Student assignments Biomedical Signals and Systems (BSS) - EEMCS Faculty - University of Twente
- Master’s thesis in the Parkinson’s Vibrating Socks Project
Educational program: Technical Medicine, Health Sciences, Biomedical Engineering, or any other relevant program.
Daily supervisor: ir. Lorenzo Giuseppe Centamore (l.g.centamore@utwente.nl)
Principal investigator: Dr.ir. Ciska Heida (t.heida@utwente.nl)
Project background:
Parkinson's disease (PD) is the most common neurodegenerative disorder after Alzheimer’s disease, characterized by a range of motor and non-motor symptoms.
One notable symptom is Freezing of Gait (FOG), defined as “brief, episodic absence or marked reduction of forward progression of the feet despite the intention to walk”. When it happens, only the lower body freezes, while the upper side continues its motion, increasing the probability of falling.
One possible symptomatic treatment for FOG is cueing, which are mechanisms that provide external stimuli to make the patient focus on each step.
In the Vibrating Socks project, we aim to predict the FOG episodes in daily-life conditions. In addition, we want to better understand the effectiveness of tactile cueing in Parkinson’s patients to prevent the FOG, by testing the use of the Vibrating Socks. When activated, these socks vibrate with a fixed cadence, applying the vibrotactile cue, until they are turned off.
The Vibrating Socks project is founded by INTERREG, and we are part of a consortium that includes university and company partners from the Netherlands and Germany:
· University of Münster
· University Medical Center of Groningen
· FeelSpace (Germany)
· Sherpa (Netherlands)
Objective of the Master’s thesis project:
For the aim of the project, we want to acquire a new dataset.
During a one-day study visit, each participant will wear movement sensors (Xsens suite), physiological sensors (a wearable ECG, a wearable Skin conduction sensor, and a smartwatch measuring the heart rate and Skin conduction from the wrist), and an eye tracker.
During the study visit, participants will perform various daily activities, under your guidance.
This new dataset will enable the exploration of several research topics, depending on the sensors or combination of sensors you would like to focus on:
· Movement data:
· How does posture change before and during FOG?
· How does gait change before FOG occurs?
· Physiological data:
o Signal quality analysis
o Measuring the Skin Conduction: Is the signal acquired from the wrist comparable to that acquired from the fingers?
o How does stress influence FOG?
· Cueing:
· Do on-demand cues trigger FOG?
· Are the socks effective in preventing FOG?
We will discuss the specific topic you would like to work on.
Your tasks:
You will be fully involved in the research process, from the recruitment of the participants to the analysis of data. Your tasks will include:
1. Identifying potential participants for the study;
2. Recruiting participants by checking inclusion and exclusion criteria;
3. Scheduling the study visits;
4. Supervising the participants during the study visit;
5. Labeling and synchronizing the acquired data;
6. Processing and analyzing the data.
What you can expect from this project:
This project will provide you with valuable insights into the process of conducting medical research involving patients. You will gain more knowledge and experience in the entire process, from the experimental protocol design to the data analysis.
Specifically, you will:
· Gain more knowledge on the processes of protocol design, recruitment, acquisition, and analysis of data;
· Work closely with patients, gaining hands-on clinical experience;
· Gain more knowledge about Parkinson’s Disease, with a focus on Freezing of Gait, learning how to recognize it, and how it is possible to prevent it.
You will also be an active part of the main project, participating in the consortium and research group meetings.
What we expect from you:
We want you to:
· Independently address challenges in patient recruitment, activity labeling, freezing episode identification, and data synchronization.
· Possess strong analytical skills.
· Communicate effectively by clearly reporting findings and progress in regular updates.
· Be open to collaboration with other team members and adaptable to project changes.
· Be fluent in Dutch, as we will recruit participants from the Netherlands.
Fluency in German is a strong plus for potential recruitment in Germany.
Information and application:
Please send your application to: ir. MSc. Lorenzo Giuseppe Centamore (l.g.centamore@utwente.nl) and include:
· A curriculum vitae including your name and contact (max 2 A4 pages).
· A personal motivational letter (max 1 A4 page)
· List of courses and grades of your BSc and MSc degrees.
- Multi-sensor data fusion for health predictions
Educational program: Electrical Engineering, Computer Science, Biomedical engineering, Mathematics or other relevant direction.
Type of assignment: Internship/BSc./MSc thesis [to be decided]
Contact person: Dr. Arlene John (a.john@utwente.nl)
Research assignment
Fusion of data obtained from multiple sensors can improve detection performance, compared to that of using data from a single sensor source. It can also improve the quality and robustness of inferences when noise corrupts data from any of the input sensors or in case of missing data. Combining information from multiple sources can aid in the reduction of false alarms and can enhance clinical decision support. In this line of research, students can work on any one of the following healthcare challenges that can be solved through multisensor fusion:
1. Explainable AI for sensor fusion using 1D-CNNs in atrial fibrillation detection using electrocardiogram and photoplethsymogram signals.
2. Multimodal Fusion for respiratory rate estimation in using electrocardiogram and photoplethsymogram signals.
3. Multimodal data fusion for cardiovascular disease detection from electrocardiogram and phonocardiogram data.
4. Early Prediction of Sepsis from Clinical Data that include vital signs such as heart rate, respiration rate, temperature etc.
5. Early detection of complications using heart rate, respiration rate, and activity levels obtained through a wearable device after major abdominal surgery.
What can you expect from us?
In this project topic, you can expect to gain experience and knowledge on:
1. Exposure to real-world data for health applications
2. Gain insights into handling time-series data
3. Opportunities to enhance skills in deep learning and machine learning
4. Gain insights into explainable AI
Who do we look for?
We are looking forward to students with some experience in machine-learning and deep-learning with a strong affinity for programming, who would like to work with a data-analysis-focused project. These provide a unique opportunity for students to hone their analytical skills, as well as to put them to practice into a relevant real-world problem.
References
Zhang, Haobo & Peng, Zhang & Lin, Fan & Chao, Lianying & Wang, Zhiwei & Ma, Fei & Li, Qiang. (2023). Co-learning–assisted progressive dense fusion network for cardiovascular disease detection using ECG and PCG signals. Expert Systems with Applications. 238. 122144.
A. John, K. K. Nundy, B. Cardiff and D. John, "Multimodal Multiresolution Data Fusion Using Convolutional Neural Networks for IoT Wearable Sensing," in IEEE Transactions on Biomedical Circuits and Systems, vol. 15, no. 6, pp. 1161-1173, Dec. 2021
A. John, S. J. Redmond, B. Cardiff and D. John, "A Multimodal Data Fusion Technique for Heartbeat Detection in Wearable IoT Sensors," in IEEE Internet of Things Journal, vol. 9, no. 3, pp. 2071-2082, 1 Feb.1, 2022
- Energy-efficient deep learning architecture for anomaly detection from ECG signals
Educational program: Electrical Engineering, Computer Science, Biomedical engineering, Mathematics or other relevant direction.
Type of assignment: Internship/BSc./MSc thesis [to be decided]
Contact person: Dr. Arlene John (a.john@utwente.nl) and Dr. Ghayoor Gillani (s.ghayoor.gillani@utwente.nl)
Research description
Cardiovascular diseases, including coronary heart disease (CHD), stroke, and other circulatory diseases, contribute to around 30% of global mortality each year. CVD is a primary cause of premature death and the leading factor in morbidity among non-communicable diseases. The financial impact of CVD is significant, with an estimated cost of approximately € 169 billion per year in the European Union. This cost includes 62% direct expenses in the healthcare system, as well as productivity loss and informal care. To address the costs and healthcare risks associated with CVD, continuous monitoring of physiological signals, such as electrocardiogram (ECG), using Internet of Things (IoT) enabled wearable devices is being explored as a potential solution. However, the practical implementation of continuous monitoring with medical-grade ECG has been hindered by the high power consumption associated with constant wireless transmission for processing in the cloud. Hence, it might be more beneficial to carry out inferences on the wearable device itself if the inference algorithms are energy and area-efficient.
Approximate computing is a technique that allows controlled errors to be introduced in order to improve computing efficiency. By increasing computing efficiency, power and energy consumption can be reduced. However, the introduction of errors may affect the precision of computing and compromise the quality of the output. The goal of approximate computing is to achieve the best efficiency design while maintaining an acceptable level of output quality. In this project, the focus is on implementing approximate multiply-accumulate (MAC) units to develop a deep-learning based algorithm for detecting heartbeat anomalies. By leveraging the benefits of approximate computing, it is expected that the algorithm can achieve efficient performance while providing sufficient quality in detecting abnormal heartbeats.
What can you expect from us?
In this project, you can expect to gain experience and knowledge on:
1. Exposure to real-world data for health applications
2. Opportunities to enhance knowledge on approximate computine
3. Opportunities to enhance skills in supervised machine learning
Tentative tasks:
1. Surveying and evaluating (or developing) a suitable deep-learning-based anomaly detection algorithm using a suitable Arrhythmia ECG database (like the MIT-BIH arrhythmia ECG database).
2. Implementing the model on FPGA (and/or ASIC) to perform hardware cost analysis (chip-area, power consumption, latency)
3. Researching approximate multiplier accumulator techniques promising for deep-learning implementation.
4. Optimizing the deep-learning MAC-based architecture for minimum power/energy consumption, which satisfies the output quality criterion of ECG anomaly detection.
Note: Depending on your interests, the research can be focused on algorithm, software, or hardware part. So, the students from Mathematics, CS, EE and other branches of EEMCS are suitable
References
Awni Y. Hannun et al. “Cardiologist-level arrhythmia detection and classification in ambulatory electrocardiograms using a deep neural network”. In: Nature Medicine 25.1 (2019), pp. 65–69
Jun Luo et al. “A Smartphone-Based M-HealthMonitoring System for Arrhythmia Diagnosis”. In: Biosensors 14.4 (2024).
Chen Zhang et al. “A Low-Power ECG Processor ASIC Based on an Artificial Neural Network for Arrhythmia Detection”. In: Applied Sciences 13.17 (2023).
Gillani Abbas et al. “MACISH: Designing Approximate MAC Accelerators with Internal-SelfHealing”. In: IEEE Access (2019).
- Predicting Psychological States using Machine Learning and Digital Biomarkers from smartwatch data.
Educational program: Biomedical Engineering, Electrical Engineering, Computer Science, Mathematics or other relevant direction.
Type of assignment: Internship/BSc./MSc thesis [to be decided]
Contact person: Dr. Arlene John (a.john@utwente.nl) and Dr. Jorge Piano Simoes (j.pianosimoes@utwente.nl)
Research assignment
In this project, we aim to understand how the psychological state affects the human physiology and vice versa. To achieve this, we aim to utilize previously collected data to investigate whether digital biomarkers passively sensed with a smartwatch can predict psychological states. This dataset comprises of 34 healthy individuals who wore a research-grade smartwatch for 7 or 8 consecutive days, and who reported their psychological state (e.g., positive or negative emotions, sense of belonging, etc.) at the end of each day. The smartwatch was used to collect various physiological signals such as heart rate, pulse rate variability, electrodermal activity and skin temperature, as well as behaviour data such as actigraphy and sleep. Through feature extraction from the physiological signals, and using various classification algorithms, we aim to identify physiological and/or behaviour markers (obtained from a smartwatch) that can predict/estimate psychological states.
Specific tasks include: (1) preprocessing the raw wearable signal into meaningful features, (2) conducting exploratory data analysis, (3) performing feature engineering (including tasks like imputation of missing values and extracting time-series features from the pre-processed data), (4) applying different ML algorithms (e.g., xgboost, neural networks, random forest, etc) to predict psychological states based on physiological markers sensed with the smartwatches, (5) investigating the most relevant features based on interpretable ML techniques, and (6) applying state-of-the-art techniques (e.g., conformal prediction) to assess the uncertainty of the ML predictions.
What can you expect from us?
In this project, you can expect to gain experience and knowledge on:
1. Exposure to real-world data for health applications
2. Exposure to data obtained from wearable devices
3. Opportunities to enhance skills in supervised and unsupervised machine learning
4. Gain insights into explainable AI
Who do we look for?
We are looking forward to students with some experience in ML and who would like to work with a data-analysis-focused project. This project uses a real-world dataset, including common challenges like missing data, preprocessing raw data, and Managing situations where the number of features (p) is much greater than the number of samples (n). These provide a unique opportunity for students to hone their analytical skills, as well as to put them to practice into a relevant real-world problem.
References
Smets, Elena, Emmanuel Rios Velazquez, Giuseppina Schiavone, Imen Chakroun, Ellie D’Hondt, Walter De Raedt, Jan Cornelis et al. "Large-scale wearable data reveal digital phenotypes for daily-life stress detection." NPJ digital medicine 1, no. 1 (2018): 67.
Föll, S., Maritsch, M., Spinola, F., Mishra, V., Barata, F., Kowatsch, T., ... & Wortmann, F. (2021). FLIRT: A feature generation toolkit for wearable data. Computer Methods and Programs in Biomedicine, 212, 106461.
Wu, Rui, Scott D. Hamshaw, Lei Yang, Dustin W. Kincaid, Randall Etheridge, and Amir Ghasemkhani. "Data imputation for multivariate time series sensor data with large gaps of missing data." IEEE Sensors Journal 22, no. 11 (2022): 10671-10683.
Stachl, Clemens, Quay Au, Ramona Schoedel, Samuel D. Gosling, Gabriella M. Harari, Daniel Buschek, Sarah Theres Völkel et al. "Predicting personality from patterns of behavior collected with smartphones." Proceedings of the National Academy of Sciences 117, no. 30 (2020): 17680-17687.
Angelopoulos, A. N., & Bates, S. (2023). Conformal prediction: A gentle introduction. Foundations and Trends® in Machine Learning, 16(4), 494-591.
- Detection of micro-expression through classification of multichannel facial EMG
Aim, Motivation and the background:
This master’s assignment aims to develop a method and experimental protocol detecting subtle (micro) expressions on the face based on measuring facial muscle activities.
Facial expressions are an important aspect of non-verbal communication, showing reactions and attention. For instance, in patients with Disorders of Consciousness (DOC), facial expressions are commonly less pronounced. Whereas, the diagnosis of these patients is partly based on their response to external stimuli, measured by their facial expressions. As these can be difficult to objectively measure, a high misdiagnosis rate exists. The development of a method to detect and identify expressions could support diagnosis and possibly improve communication between patients and caregivers or loved ones. The challenge is to identify facial expressions for DOC patients who may have abrupt body/head movements and subtle manifestations of facial gestures which cannot be detected using optical systems.
To tackle this challenge, we will use high-density surface electromyography (HD-sEMG) to record from multiple muscles contributing to major facial expressions of emotions. Previous works showed that we can detect such subtle manifestations of muscles through electromyography (EMG). Using machine learning approaches, we will develop a generic EMG model of emotional expressions.
The main objectives of this research are:
• to design an experimental protocol to obtain facial EMG measurements.
• to identify an appropriate machine learning approach to classify microexpressions based on 32 channels of high-density EMG data.
• to identify minimal sensor settings to be able to classify micro-expressions with significant accuracy.
A schematic representation of the desired workflow is shown below:
Skills to be obtained during this assignment:
- multi-channel EMG data acquisition - biomedical signal processing ML
- Using machine learning to classify EMG sources.
Contact: The student will work at the laboratory of the Department of Biomedical Systems and Signals, faculty of EEMCS, University of Twente, under the supervision of
Utku S. Yavuz, (s.u.yavuz@utwente.nl)
Dr. Arlene John (a.john@utwente.nl)
- Body Sway analysis using an accelerometer
Assignment: Master
Contactperson: Frank Wouda (f.j.wouda@utwente.nl)
Project background
Over 35 years, the Centre for Human Drug Research (CHDR) in Leiden has developed a test battery that is able to quantify drug effects on the Central Nervous Systems. Central to this test battery is the Body Sway test that is looking at the anteroposterior displacement of a person standing with their eyes closed. The current implementation of the test consists of a string attached to a potentiometer and the waist of the participant.
The objective of this thesis is to study whether trunk acceleration and/or rotation movements are potential surrogates for the Body Sway test. Data is actively being collected for preliminary analysis and additional data will be collected during the course of the thesis.
If you are very experienced with optimization frameworks (such as Kalman Filters), this could potentially be shaped as an internship.
The analysis will be in Python and will be conducted in close collaboration with the clinical team at CHDR. The work can be performed remotely. For more information, please contact Vasileios Exadaktylos (vexadaktylos@chdr.nl) or Frank Wouda (f.j.wouda@utwente.nl).
- Instantaneous knee monitoring for ACL patients using IMUs
Assignment: Master/Bachelor/Internship
Start Date: Whenever you want (Preferably Immediate)
Contact Person: Sanchana Krishnakumar (s.krishnakumar@utwente.nl), Dr.Bert-Jan Van Beijnum (b.j.f.vanbeijnum@utwente.nl)
Topics: IMU, knee biomechanics, Machine learning, Physiotherapy, and Rehabilitation
Background: In the Netherlands, it is estimated that about 17% of the patients who visit a physiotherapist have knee-related problems. Among these patients, anterior cruciate ligament (ACL) injury and knee osteoarthritis are the most common complications that require long-term rehabilitation. To design and plan rehabilitation protocols for ACL patients, physiotherapists currently depend on subjective visual assessments to determine the functional status of patients and monitor progress of the knee over time. To optimally treat patients recovering from an ACL injury, the physiotherapist needs to gain insight into knee angles and net knee moments.
The objective of the INSTANT project is to develop an integrated wearable sensor-based system using Inertial and magnetic measurement units (IMMUs) that enables the physiotherapist to efficiently assess these kinetic and kinematic parameters during physiotherapy sessions and aid them in clinical decision-making. Collaboration partners in the project include Roessingh Research and Development (RRD, Enschede), Saxion University (Enschede), Gable (Hengelo), and TopvormTwente Fysiotherapeuten (Enschede).
What you can expect:
· Hands-on experience with IMUs and involvement in patient measurements in a clinical setting in close collaboration with project partners and physiotherapists.
· Develop algorithms for relevant quantitative kinematic and kinetic quantities for knee monitoring from measured data from the experiments that are crucial for tailoring the treatment of ACL patients.
· Engage in a project with direct applications in physiotherapy, contributing to advancements in patient care and rehabilitation.
· Enhance skills in biomechanics, machine learning, and programming languages such as MATLAB/Python, providing valuable expertise for future endeavours.
· A potential scientific publication based on the project outcomes.
Advised background knowledge:
Biomechanics of Human movement, Machine learning or related topics, MATLAB/Python experience
- Wearable Sensor Synchronization and Deep Learning for freezing of Gait Detection in Parkinson's Patients
Educational program: Biomedical Engineering, Electrical Engineering, Computer Science, Mathematics or other relevant direction.
Daily supervisor: ir. MSc. Juan Delgado Teran (j.d.delgadoteran@utwente.nl)
Principal investigator: Ciska Heida (t.heida@utwente.nl)
Project background:
In the INTENSE project, we aim to better understand and diagnose one of the most disabling symptoms of Parkinson’s disease – freezing-of-gait (FOG) https://youtu.be/3-wrNhyVTNE, in order to provide personalized care using cueing. External cueing consists of the application of (often rhythmic) spatial or temporal stimuli, for example, the beat of a metronome or a series of stripes on the floor, to improve the performance of movements such as gait, or to help initiate movements.
In this part of project, participants with Parkinson's disease were equipped with wearable sensors for a 7-day home data collection. We aim to address the specific challenges posed by unsynchronized wearable sensor data collected from the homes of individuals with Parkinson's disease. The diverse nature of the sensors introduces complexities in aligning and interpreting the data accurately. This project represents a unique opportunity to contribute to the advancement of technology for Parkinson's disease management by developing methods for data synchronization and leveraging unsupervised or semi-supervised learning techniques for Freezing of Gait detection.
What you can expect from us:
In this project, you can expect to obtain the knowledge and experience about:
· Developing and implementing a synchronization mechanism. This includes developing strategies for timestamping data from the pressure sensors (Moticon), and smartwatch (Empatica) to guarantee alignment across devices.
· Exposure to real-world applications of data synchronization and analysis for healthcare purposes.
· Opportunities to enhance skills in unsupervised and semi-supervised learning techniques
· on synchronized data to detect and/or predict freezing-of-gait episodes, providing a more comprehensive understanding of FOG in a real-world context.
· Gain insights into Parkinson's Disease and its specific features related to freezing-of-gait through hands-on experience with synchronized, real-world data.
What we expect from you:
· Ability to independently address challenges related to data synchronization and unsupervised learning techniques.
· Proficiency in programming languages, particularly Python, for the implementation of synchronization algorithms and data analysis.
· Analytical Skills:
· Strong analytical skills to conduct exploratory data analysis and interpret results effectively.
· Clear and concise communication of findings and progress in regular updates and presentations.
· Willingness to collaborate with other team members and adapt to evolving project requirements.
Information and application:
Please send your application to: ir. MSc. Juan Delgado Teran (j.d.delgadoteran@utwente.nl) and include:
· A curriculum vitae including your name and contact (max 2 A4 pages).
· A personal motivational letter (max 1 A4 page)
· List of courses and grades of your BSc and MSc degrees.
- Wearable health device development and application: Multimodal signal sensing and analysis for chronic disease monitoring in daily life
Master graduation project (BSS-UTwente +Nokia Bell Labs Cambridge-UK):
Educational program: Embedded System, Electrical Engineering, or other relevant direction
Supervision team:
Dr. Ying Wang (BSS-EEMC, UTwente) and Dr. Hongwei Li (Nokia Bell Labs Cambridge-UK)
Project background:
Noncommunicable diseases, including cardiovascular diseases (CVDs), diabetes, and mental disorders, are the number one cause of death and disability in the world. The prevalent cases of noncommunicable disease have continued its decades-long rise, and the global burden of the diseases, e.g. premature mortality and the loss of quality of life, will dramatically increase given a boost in aging population with unhealthy lifestyle. The daily monitoring of noncommunicable diseases can help clinicians provide timely interventions; hence, this can mitigate socioeconomic burden and increase patients’ quality of life. However, the performance of current wearable sensing techniques and monitoring techniques is still lacking behind despite the growing understanding of disease pathophysiology and treatment in the advanced stage of diseases.
Collaborating between BSS-UTwente and NokiaBell Labs, we aim at developing advanced wearable sensing technology and health-condition monitoring technology to tackle the challenges mentioned above. A wearable health device “Patchkeeper” developed by NokiaBell Labs contains an electronic stethoscope, PPG, ECG and IMU sensors to measure multimodal signals from human or animal subjects. The acquired multimodal signals are analysed for the disease monitoring in the daily life.
What you can expect from us:
In this project, you can expect to obtain the knowledge and experience about:
- Apply embedded computer architecture, electronics, embedded software optimization techniques in the development of Patchkeeper.
- Develop algorithms for noncommunicable disease monitoring using signal analysis and machine learning techniques:
- Preprocess different modal signals. Extract features from the multimodal signals. Apply machine learning techniques to identify the health condition of human subjects.
- Design and implement a polit experiment with healthy subjects to collect relevant data.
- Opportunities for visiting Nokia Bell Labs in Cambridge, UK for the Patchkeeper development.
- A potential scientific publication can be expected based on the project outcomes.
What we expect from you:
- We are looking for open-minded students who like to challenge themselves.
- Students should have strong skills in embedded system design, advanced signal analysis, and machine learning techniques.
- Students should have strong programming skills in Python and C language.
- Interests in physiological signal analysis and relevant research.
Information and application:
Please send your application to dr. Ying Wang (ying.wang@utwente.nl), and include:
- A curriculum vitae including your name and contact (max 2 A4 pages).
- A personal motivation letter (max 1 A4 page).
- Grade list of your BSc and MSc courses.
- Finite element method simulation of non-invasive brain stimulation
Master graduation project or internship (exceptions possible for Bachelor students with necessary skills)
Educational programs: Biomedical Engineering, Electrical Engineering, or other relevant studies.
Supervision team: Dr. Bettina Schwab (BSS-EEMC, UTwente), Msc. Silvana Huertas Penen (BSS-EEMC, UTwente)
Topics: Non-invasive brain stimulation, Computer simulations using the Finite element method (FEM)
Introduction: We are interested in optimizing stimulation montages for a non-invasive brain stimulation technique, transcranial alternating current stimulation (tACS).
- For this optimization, Electric field distributions across different conditions have to be compared.
- You will run multiple electric field simulations of tACS using the software SimNIBS in Matlab (See example below) and analyze the results.
- If you finish earlier than attempted, you can aim for additional, advanced analyses.
Required pre-knowledge: We are looking for open-minded students who like to challenge themselves and have a background in electrical engineering, biomedical engineering, or related fields.
- Students should have a strong interest in performing FEM computer simulations
- Students should have strong programming skills in Matlab.
- Experience in the physiological signal analysis is optimal but not mandatory.
Information and application: Interested students are encouraged to send an e-mail to Silvana Huertas Penen, (s.huertaspenen@utwente.nl)with their interests and course list.
- Predicting transplant detachment by analyzing movement data in Descemet Membrane Endothelial Keratoplasty (DMEK) corneal transplantation patients
Student assignments: BSc and MSc project assignments
Educational programme: BMT/BME, TG/TM, or similar
Contact person: Sigert Mevissen
Email contact: s.j.mevissen-1@utwente.nl
Project supervisors: Bert-Jan van Beijnum & Sigert Mevissen
Topic: Motion analysis for predicting graft detachment in DMEK cornea transplantation patients.
This research project of BSS and the department of ophthalmology at Deventer Hospital uses an IMU to investigate the relationship between graft detachment and the extent to which patients can adopt a calm flat posture after Descemet Membrane Endothelial Keratoplasty (DMEK) corneal transplantation.
1500 corneal transplants are performed annually in the Netherlands, of which more than 100 take place at Deventer Hospital. In a DMEK operation, the endothelial and descemet layers of the patient's cornea are replaced with the same layers of functioning donor tissue. In a DMEK, the graft is not sutured but pushed into place by an air or gas bubble. Because air rises, patients are instructed to lie flat as much as possible for the first 24 hours after surgery, so that the largest area of the bubble can push the graft against the patient's own cornea. A major postoperative complication of this surgery is graft detachment, after which a new air bubble is added or a new graft is put in place. Patients undergoing DMEK are mostly older than 65 years, and strict flat bed rest is quite a challenge for them. It is therefore important to investigate to what extent graft detachment can be predicted by looking at the patient's movements and orientation. The patient will have an IMU attached to their head for the first 24 hours after surgery to capture movements.
Assignment:
Initially, the aim is to find a general relationship between the degree of movement and graft detachment. This will then be to investigate which specific postures and movements affect the release and to what extent, so that calm flat posture instruction can be adjusted to alleviate the burden on the patient. For this, the location of the graft detachment will be taken into account.
Elements to expect in this assignment:
- Signal processing to analyze accelerometer and gyroscope data and convert it into useful information.
- Motion analysis to construct parameters representing degree of motion.
- Develop activity recognition (possibly machine learning) methods to determine how often specific movements and postures are adopted.
- Perform measurements on DMEK patients in the ophthalmology department of Deventer hospital
- Personalized cAnceR TreatmeNt and caRe (PARTNR) activity data analyses
Assignment: Bachelor, Master, Internship
Educational program: BME, TM, HS, PSY, CREATE
Daily supervisor: Kim Wijlens, MSc or ir. Lian Beenhakker
Project description
Due to improved treatment and early diagnosis of breast cancer, there are more survivors. However, this also means that more patients are struggling from the late effects of treatment. One of these late effects is cancer-related fatigue (CRF), which is defined by the American National Comprehensive Cancer Network as “a distressing, persistent, subjective sense of physical, emotional and/or cognitive tiredness or exhaustion related to cancer or cancer treatment that is not proportional to recent activity and interferes with usual functioning.”
In the Personalized cAnceR TreatmeNt and caRe (PARTNR) project, we aim to help breast cancer patients suffering from CRF. We will holistically assess patients, considering all aspects that can cause/influence the fatigue and predict who might be at risk of developing CRF. Using multimodal data, we will advise an intervention to reduce the fatigue which is based on the holistically assessed patient information about CRF and personal preferences patients might have for types of intervention. In the end, we will combine all information from patients into an intelligent self-learning platform that patients can use to keep track of and improve their fatigue.
Open assignment
Dysregulation of activity and sleep are perpetuating factors of cancer-related fatigue. The relation between physical activity behaviour and fatigue suggests that cancer survivors might be performing too much activity in the morning. This results in increased fatigue levels, which in turn result in a relapse in activity from the afternoon going into the evening. Patients’ activity can be monitored with a wearable, mobile phone apps or questionnaires. Currently, we assess activity via subjective questionnaires. Wearables or mobile phone data could provide continuous information about the patients’ activities. Wearables like Garmin or Fitbit have a protected algorithm, via the dashboard the variable outcomes are available and via the developer site the raw data can be obtained. However, the variable outcomes are subject to unknown changes due to the algorithm updates. This is a challenge in research since updates can occur during the studies and within patients. An open source platform as (https://www.beiwe.org/technology/#beiwe) can be used to retrieve raw phone data. So the question is from what source (questionnaires, wearables or raw phone data) should activity data be used to for a personalised treatment advice based on activity level. In addition, the frequency of available data differs between assessment via questionnaires and continues information via wearables or phone data. So, how can the variables of continues sources be adapted into output that are comparable to questionnaire information without content loss, to use as information to provide the patient with relevant personalized treatment advice?The PARTNR project is a collaboration between the University of Twente, Ziekenhuis groep Twente (ZGT), Helen Dowling Institute (HDI), Roessingh Rehabilitation Centre, University Medical Center Groningen (UMCG), Dutch breast cancer association (BVN), Netherlands comprehensive cancer organisation (IKNL), Ivido and Evidencio.
If you would like to work on the PARTNR project during your internship or thesis, please contact either Kim Wijlens (k.a.e.wijlens@utwente.nl) or Lian Beenhakker (l.beenhakker@utwente.nl). Supervisors will depend on the chosen assignment.
PARTNR Team:
dr. Annemieke Witteveen (BSS), prof. dr. Sabine Siesling (HTSR), dr. Christina Bode (PGT),
- Validation of a 3-IMU setup in chronic stroke patients (BSS and RRD)
Supervisors:
Bert-Jan van Beijnum (UT) (b.j.f.vanbeijnum@utwente.nl)
Jaap Buurke (RRD, UT) (j.buurke@rrd.nl) / (j.h.buurke@utwente.nl)
Type of assignments:
BSc thesis, MSc thesis, internship
Background student:
Biomedical Engineering, Movement Science, Technical Medicine, Biomechanics, or other relevant direction.
Project description:
The validation of an ambulatory three IMU setup for measuring balance parameters in adults with an asymmetric gait.
People with an asymmetric gait, for example as a result of hemiparesis caused by a stroke, or due to an amputation, experience trouble in keeping their balance. As a result of this, they need help with daily life activities and they often fall. Clinical therapy is therefore focused on improving mobility and functional capacity. However, there is a lack of objective information about the rehabilitation progress when the patient is back home. [1] In order to compare the gait and balance parameters during clinical training and performance in the home setting, a wearable and unobtrusive system is needed.
Balance is related to the position of the CoM (centre of mass) with respect to the position of the feet (i.e. base of support) [2]. Tracking foot and CoM positions is feasible with an IMU only setup, where the IMUs are placed on the segments of interest. However, as no information about relative distances are known, and due to strapdown integration, the positions measured by the IMUs drift away from each other. Current state of the art systems such as the ForceShoes™ use an ultrasound system to reduce this drift. [3] Other alternatives include measuring every segment using the Xsens suit and applying biomechanical constraints. [4,5] However, as the ForceShoes™ are bulky to be used in daily life (weight: 1 kg each), and the Xsens suit is extensive, alternatives are required.
In a previous study a minimal IMU setup for ambulatory sensing of foot and CoM positions in overground variable gait is validated in healthy (symmetric gait) adults, with the ForceShoes™ and the VICON© motion capture system as a reference. [6] In order to reduce the drift between the IMU positions, this setup makes use of the Centroidal Moment Pivot (CMP)-theory.
Aim:
The goal of the current assignment is to validate this method for patients with an asymmetric gait. This is important, since the effect of the assumptions made in the CMP-theory could be different if one has an asymmetric gait. For this, ~7 CVA-patients need to be measured, according to a pre-defined protocol. The measurements will be performed in the gait lab at Roessingh Research and Development, Enschede, after which you will post-process and analyze the data. There is already ethical approval for the study, so you can almost directly start measuring the patients!
Start time:
Whenever you want, preferably as soon as possible.
Note: it is also possible (but not mandatory) to continue during the summer holiday.
References
[1] B. Klaassen et al., “A Fully Body Sensing System for Monitoring Stroke Patients in a Home Environment”, Communications in Computer and Information Science, vol. 511, pp. 378-393, 2016.
[2] A.L. Hof, M.G.J. Gezendam, W.E. Sinke, “The condition for dynamic stability”, Journal of Biomechanics, vol. 38, no. 1, pp. 1-8, 2005.
[3] D. Weenk, D. Roetenberg, B.-J. van Beijnum, H.J. Hermens, P.H. Veltink, “Ambulatory Estimation of Relative Foot Positions by Fusing Ultrasound and Inertial Sensor Data”, IEEE Transactions on neural systems and rehabilitation engineering, vol. 23, no. 5, pp. 817-826, 2015.
[4] D. Roetenberg, H. Luinge, P. Slycke, “Xsens MVN: Full 6DOF Human Motion Tracking Using Miniature Inertial Sensors”, Xsens Technologies, Enschede, The Netherlands, version April 3, 2013.
[5] H. Zhoa et al., “Heading Drift Reduction for Foot-Mounted Inertial Navigation System via Multi-Sensor fustion and Dual-Gait Analysis”, IEEE Sensors Journal, vol. 19, no. 19, pp. 8514-8521, 2019.
[6] M.I.M.Refai, B.-J. F. van Beijnum, J.H. Buurke, P.H. Veltink, “Portable Gait Lab: Tracking Relative Distances of Feet and CoM Using Three IMUs”, IEEE Transactions on neural systems and rehabilitation engineering, vol. 28, no. 10, 2020.
- Freezing-of-Gait detection in Parkinson’s Disease using wearable sensors
Student assignment: Master/Bachelor BMT, EE, related
Supervisor: Juan Delgado Teran (j.d.delgadoteran@utwente.nl)
Principal investigator: Ciska Heida (t.heida@utwente.nl)
Topics: signal analysis, Parkinson’s disease, movement sensing, physiological sensing, daily monitoring, e-health, machine/deep learning
Introduction: In the PROMPT and INTENSE projects, we aim to better understand and diagnose one of the most disabling symptoms of Parkinson’s disease – freezing-of-gait (FOG) https://youtu.be/3-wrNhyVTNE, in order to provide personalized care using cueing. External cueing consists of the application of (often rhythmic) spatial or temporal stimuli, for example, the beat of a metronome or a series of stripes on the floor, to improve the performance of movements such as gait, or to help initiate movements
To observe patients in a semi-daily living environment we use multiple sensors to record the patient’s behaviour during half a day in the eHealth House (TechMed Centre). During data collection, various wearable sensors are attached to the person with Parkinson’s Disease including motion sensors (Xsens suit), pressure sensors (Moticon) and a smartwatch (Empatica). The subject is asked to perform everyday tasks (e.g. cooking, cleaning, etc.) in the eHealth House, as well as take a 10-minute walk outside with and without medication. In the eHealth house. Videos are taken of the subject moving around in the eHealth House, in order to label freezing episodes. The labelled video data is then utilized to train classification algorithms to detect freezing-of-gait based on sensor data.
In this project, you can expect to obtain the knowledge and experience about:
- Applying machine learning techniques on different types of data to detect and/or predict FOG episodes.
- Use of signal analysis techniques to pre-process and clean the data from artefacts.
- Learning about Parkinson's Disease and features about FOG.
- Possibly becoming a co-author of a publication depending on the results.
What we expect from you:
- Proactive and an open-minded attitude to new challenges
- Strong interest in advanced signal analysis and/or Machine Learning/Deep learning
- Experience with movement sensors would be ideal but is not required.
For more information contact Juan Delgado Terán (j.d.delgadoteran@utwente.nl)
- Modulation of brain activity using transcranial alternating current stimulation
Master graduation project (exceptions possible for internships and Bachelor students)
Educational programs: Biomedical Engineering, Electrical Engineering, (M3 for Technical Medicine), or other relevant studies.
Supervision team: Dr. Ciska Heida (BSS-EEMC, UTwente), Dr. Bettina Schwab (BSS-EEMC, UTwente), Msc. Silvana Huertas Penen (BSS-EEMC, UTwente)
Topics: Transcranial alternating current stimulation (tACS), neuromodulation, EEG data analysis.
Introduction: We are working on researching how transcranial alternating current stimulation (tACS) modulates brain activity/connectivity. The project will include: acquisition of EEG data in healthy participants during tACS and analysis of EEG data. Projects can be adapted to personal interests and skills.
Example assignments:
1. Acquisition of EEG data with tACS + analysis of brain activity using EEG data
2. Acquisition of EEG data with tACS+ analysis of functional connectivity (EEG data)
Required pre-knowledge:
1. We are looking for open-minded students who like to challenge themselves and have a background in electrical engineering, biomedical engineering, or related fields.
2. Students should have strong interest in advanced signal analysis techniques.
3. Students should have strong programming skills in Matlab.
4. Experience in physiological signal analysis andclinical research is optimal but not mandatory.
Information and application:
Motivated students are encouraged to send an e-mail to Silvana Huertas Penen (s.huertaspenen@utwente.nl) with their interests and course list.
- Signal Separation in the Monitoring of Respiratory Muscle Activity of ICU patients
Master graduation project (BSS+CRPH)
Educational program: Embedded Systems, Electrical Engineering, Biomedical Engineering, or other relevant direction
Supervision team:
Dr. Ying Wang (BSS-EEMC, UTwente) and Dr. Eline Mos-Oppersma (CRPH-TNW, UTwente)
Project background:
The diaphragm represents the major muscle of the so called ‘human respiratory pump’, which allows us to breath several times every minute of every day. As all skeletal muscles, the diaphragm and accessory respiratory muscles are composed of multiple groups of muscle fibers and innervated by motor neurons. Activation of the motor neurons results in the propagation of action potentials and the activation of the muscle fibers. This diaphragmatic wave of electrical activation can be recorded using dedicated electrodes to obtain electromyograms (EMG). When the respiratory pump fails and a patient needs mechanical support of breathing at the Intensive Care, it is essential to monitor diaphragmatic activity, both to prevent further failure and to optimize treatment.
A novel easy-to-use and noninvasive approach is to measure the electromyogram (EMG) via electrodes attached to the skin. Yet, analysis of these data is complex, in part based on the inherent crosstalk of other (thoracic or abdominal) muscles. More important, the main disturber of the surface EMG of the diaphragm is the heart. The geometry of the heart relative to the diaphragm determines the amplitude and timing of the electrical activity of the heart to be represented in the measured signal. Separation of these sources of electrical activity would improve interpretation and clinical application of EMG measurements, and thereby improve individual patient care.
What you can expect from us:
We aim to investigate Signal Separation methods, e.g., Independent Component Analysis and deep neural network, for the monitoring of respiratory muscle activity of ICU patients. Through Signal Separation methods, different physiological source signals are expected to be well separated, and hence, the quality of respiratory muscle activity will be satisfied for accurate ICU patients’ monitoring.
In this project, you can expect to obtain the knowledge and experience about:
- Design and implement a polit experiment with healthy subjects to collect relevant signals as reference.
- Compare and evaluate different signal separation methods on the reference signals.
- Apply and customize the outperformed signal separation method on ICU patients’ dataset.
- A potential scientific publication can be expected based on the project outcomes.
What we expect from you:
- We are looking for open-minded students who like to challenge themselves and have a background in electrical engineering, biomedical engineering, or other relevant direction.
- Students should have strong interest in advanced signal analysis techniques.
- Students should have strong programming skills in Matlab\Python.
- Experience in physiological signal analysis and\or clinical research is optimal but not mandatory.
Information and application:
Please send your application to Dr. Ying Wang (ying.wang@utwente.nl), and include:
- A curriculum vitae including your name and contact (max 2 A4 pages).
- A personal motivation letter (max 1 A4 page).
- Lists of courses of your BSc and MSc programs.
- Integrating physiological simulator and machine learning for the monitoring of cardiorespiratory responses during daily physical activity
Master graduation project (BSS+CRPH)
Educational program: Robotics, Systems&Control, Electrical Engineering, Biomedical Engineering, or other relevant direction
Supervision team:
Dr. Ying Wang (BSS-EEMC, UTwente) and Dr. Libera Fresiello (CRPH-TNW, UTwente)
Project background:
The human cardiovascular system is a highly dynamic and complex system that has intrigued scientists for several centuries. The dynamicity of the cardiovascular system is particularly important during daily physical activity, when heart and vessels have to adapt to deliver more oxygen to muscular tissues.
Nowadays, with the advancement of computer powers, new and sophisticated techniques of investigations are emerging as machine learning. This technique can accelerated the building of physiological models with data-driven strategies and can potentially serve the purpose of studying the human cardiovascular system during daily physical activity. However, these data-driven models lack in explaining the underlying physiological mechanism. This could lead to unreliable models used in real healthcare applications.
On the other hand, in-silico physiological simulators (e.g., the cardiorespiratory simulator described below) offer a detailed description of cause-effect mechanisms occurring in the human body, but are relatively complicated with unknown parameters which need to be manually tuned; and hence, the physiological simulator is not ideal to be directly used in the daily life applications.
Given this complementarity, physiological simulators and data-driven models derived by machine learning techniques can benefit from each other. Building such synergy can provide a reliable model for the monitoring of cardiorespiratory responses during daily physical activity.
Brief description of the physiological model:
The in silico cardiorespiratory simulator is a lumped parameter model of heart, vessels, ventilation and respiration. It reproduces pressure and flows profiles over a heart cycle in the main circulatory districts as well as O2 and CO2 content in the arterial and venous vessels. The simulator includes also negative feedback loops (e.g. autonomic and metabolic controls) that allow to reproduce human homeostasis and the adaptation of the human body to stimuli and change of status. As for example, if exercise is input, the simulator automatically reproduces the response of the human body in terms of increase of heart rate, ventilation, cardiac output etc. The simulator was also adapted to reproduce exercise in heart failure patients, with related physical impairments and limitations. As such, it is a detailed and sophisticated deterministic model of human cardiorespiratory (patho)physiology.
Main reference:
Fresiello, L., Meyns, B., Di Molfetta, A., & Ferrari, G. (2016). A Model of the Cardiorespiratory Response to Aerobic Exercise in Healthy and Heart Failure Conditions. Frontiers in physiology, 7, 189. https://doi.org/10.3389/fphys.2016.00189
Alber, M., Buganza Tepole, A., Cannon, W.R. et al. Integrating machine learning and multiscale modeling—perspectives, challenges, and opportunities in the biological, biomedical, and behavioral sciences. npj Digit. Med. 2, 115 (2019). https://doi.org/10.1038/s41746-019-0193-y
What you can expect from us:
In this project, you can expect to obtain the knowledge and experience about:
- Apply machine learning techniques and dynamic system techniques in building up the synergy between cardiorespiratory simulators and data-driven models.
- Use signal analysis techniques to get physiological parameters.
- Design and implement a polit experiment with healthy subjects to collect relevant data.
- Learning basic cardiovascular physiological concepts and modelling techniques
- A potential scientific publication can be expected based on the project outcomes.
What we expect from you:
- We are looking for open-minded students who like to challenge themselves.
- Students should have strong interest in advanced signal analysis, machine learning, and dynamic system techniques.
- Students should have strong programming skills in Matlab\Python.
- Experience in physiological signal analysis and\or clinical research is optimal but not mandatory.
Information and application:
Please send your application to Dr. Ying Wang (ying.wang@utwente.nl), and include:
- A curriculum vitae including your name and contact (max 2 A4 pages).
- A personal motivation letter (max 1 A4 page).
- Lists of courses of your BSc and MSc programs.
- Early detection of noncommunicable diseases using multimodal physiological signal and system analysis
Educational program: Biomedical Engineering, Electrical Engineering or other relevant direction
Daily supervisor: Dr. Ying Wang <ying.wang@utwente.nl>
Project background:
Noncommunicable diseases, including cardiovascular diseases (CVDs) and diabetes, are the number one cause of death and disability in the world. For example, the number of people with CVDs nearly doubled to 523 million, and the one with diabetes almost quadrupled
Student assignment: Bachelor/Master project
to 463 million over the past three decades. The prevalent cases of noncommunicable disease have continued its decades-long rise, and the global burden of the diseases, e.g. premature mortality and the loss of quality of life, will dramatically increase given a boost in aging population with unhealthy lifestyle.
The early detection of noncommunicable diseases can help clinicians provide timely interventions; hence, this can mitigate socioeconomic burden and increase patients’ quality of life. However, the effectiveness of current early detection techniques is still lacking far behind despite the growing understanding of disease pathophysiology and treatment for in the advanced stage of diseases.
What you can expect from us:
We aim to develop algorithms to track the health condition of individuals at the risk of noncommunicable diseases using different modal physiological signals, such as accelerometer, electrocardiography (ECG), and photoplethysmogram (PPG) signals.
In this project, you can expect to obtain several or all knowledge and research experience listed below:
· Design a polit experiment for healthy subjects as the control group of the targeted population.
· Collect physiological signals from human subjects using wearable sensors.
· Develop algorithms for the early detection of noncommunicable diseases:
o Preprocess different modal signals.
o Extract features from the signals.
o Apply machine learning and/or physiological system techniques to identify the health condition of human subjects.
· Test the developed algorithms on data collected from the targeted population: you will have chances to collaborate with hospitals and improve your algorithms accordingly.
· A potential scientific publication can be expected based on the project outcomes.
What we expect from you:
· We are looking for talented and open-minded students who have a background in biomedical engineering, electrical engineering, or other relevant direction.
· Students should have strong interest in signal and system analysis and machine learning techniques.
· Students should have strong programming skills in Matlab/Python.
· Experience in physiological signal analysis and/or clinical research is optimal but not mandatory.
Information and application:
Please send your application to Ying Wang (ying.wang@utwente.nl), and include:
· A curriculum vitae including your name and contact (max 2 A4 pages).
· A personal motivation letter (max 1 A4 page).
· Lists of courses of your BSc and MSc degrees.
- Clinical adverse event detection and prediction with vital sign signals
Student assignment: Master graduation project (40-45 credits, 28-32 weeks)
Educational program: Electrical Engineering, Biomedical Engineering, or other relevant direction
Daily supervisor: Dr. Ying Wang <ying.wang@utwente.nl>
Project background:
Continuous wearable sensing technologies have been widely applied in the vital sign monitoring of in-hospital patients for timely and personalized intervention. These advanced techniques play an essential role in the monitoring of patients’ clinical adverse events to decrease patients’ mortality rate and release the burden of national health care systems. However, the performance of current remote monitoring systems is still not satisfied given several interference factors, such as, daily body movement artefacts, missing data, and patients’ heterogeneity.
What you can expect from us:
We aim to further improve the performance of our current monitoring algorithms using different modal vital sign signals and patients’ medical records, such as electrocardiography (ECG), photoplethysmogram (PPG) signals and body temperature. In this project, you can expect to obtain the knowledge and research experience about:
· Develop an algorithm to detect or even predict clinical adverse events:
o Preprocess different modal signals.
o Extract features from the time-series signals, such as, using long-termed trend analysis, time-frequency and morphological signal analysis.
o Apply machine learning techniques for the detection and/or prediction.
· Test the developed algorithm on data collected from in-hospital patients: you can visit the hospital to observe the patients’ daily activities to inspire and improve your algorithms.
· A potential scientific publication can be expected based on the project outcomes.
What we expect from you:
· We are looking for talented and open-minded students who have a background in electrical engineering, biomedical engineering, or other relevant direction.
· Students should have strong interest in signal processing, analysis, and machine learning.
· Students should have strong programming skills in Matlab/Python.
· Experience in physiological signal analysis and/or clinical research is optimal but not mandatory.
Information and application:
Please send your application to Ying Wang (ying.wang@utwente.nl), and include:
· A curriculum vitae including your name and contact (max 2 A4 pages).
· A personal motivation letter (max 1 A4 page).
· Lists of courses of your BSc and MSc degrees.
- NSP-Team: Observation of altered nociceptive processing based on brain activity using paired probing
Supervisors: Dr. ir. J.R. Buitenweg (j.r.buitenweg@utwente.nl)
Our research group recently developed a method to observe altered sensory processing in chronic pain patients. This method stimulates small nerve fibers in the skin, responsible for the sensation of pain, and measures detection probability and brain activity in response to these stimuli. In order to specifically stimulate nociceptive nerve fibers in the skin, the current should remain below twice the detection threshold (Mouraux, 2010). The current stimulation paradigm stimulates close to this detection threshold to remain nociceptive-specific and to accurately estimate the detection threshold. However, the signal-to-noise ratio of brain potentials evoked in this range of stimulus amplitudes is poor, and does not allow for observation of all evoked potential components that are typically observed during nociceptive stimulation, nor observation of event-related synchronization and desynchronization in the brain. Recently, a first pilot showed that we can enhance the signal-to-noise ratio by pairing each near-threshold stimulus with one stimulus of twice the number of pulses, also referred to as ‘paired-probing’. The next step is to use this paradigm in a larger number of participants and check whether this paradigm allows for observation of all typical evoked potential components as well as event-related (de)synchronization. As such, this master assignment includes the following objectives:
1) Use the paired-probing paradigm to assess nociceptive detection thresholds, and evoked EEG responses (evoked potential, event-related (de)synchronization) on the hands and feet in 30 healthy participants.
2) Assess whether stimulation using this paradigm remains limited to nociceptive nerve fibers, or co-activates tactile nerve fibers in the skin.
3) Evaluate whether we can improve the signal-to-noise ratio of evoked potential components and event-related (de)synchronization using this paradigm.
References
Mouraux, A., Iannetti, G. D., & Plaghki, L. (2010). Low intensity intra-epidermal electrical stimulation can activate Aδ-nociceptors selectively. Pain, 150(1), 199-207. doi:10.1016/j.pain.2010.04.026
- NSP-Team: Feature-based machine learning to explore signs of neuropathic pain in electroencephalography data
Supervisors: Dr. ir. J.R. Buitenweg (j.r.buitenweg@utwente.nl)
Starting date: as soon as possible
Required background: Biomedical engineering or electrical engineering, with a strong emphasis on signal processing, pattern recognition and machine learning. Experience with programming in Python and Matlab. Knowledge about electroencephalography and experience with experiments on human subjects are a pre.
Additional information: This master assignment can potentially be extended into a PhD project provided that funding is available.
Description:
About 50% of patients with diabetes eventually develop diabetic polyneuropathy during their lifetime, resulting in the loss of function of peripheral sensory nerve fibers (Hicks & Selvin, 2019). This loss of function is often followed by the development chronic neuropathic pain. Currently, there are no non-invasive methods available to monitor the development of chronic neuropathic pain and guide medical treatment.
Our research group recently developed a method to observe altered sensory processing in chronic pain patients. This method stimulates small nerve fibers in the skin, responsible for the sensation of pain, and measures detection probability and brain activity in response to these stimuli. Based on this evoked brain activity, we might be able to observe emerging neuropathic pain. However, the signal-to-noise ratio of single preselected features is too low accurately identify neuropathic pain in patients. Brain activity based observation of neuropathic pain in these patients might be improved by extracting a large set of features in the time, frequency, and/or time-frequency domain and combining these features using machine learning methods such as SVM, random forest, or Riemannian-geometry-based decoding (Gemein et al., 2020). During this master assignment, you will develop methods to automatically extract features from EEG data, to train and to evaluate classification performance of several feature-based machine learning methods. Objectives:
1) Develop a pipeline for data preprocessing and augmentation.
2) Develop machine learning pipelines for a collection of advanced feature-based machine learning methods for EEG classification.
3) Evaluate the performance of data augmentation and machine learning methods.
References
Gemein, L. A. W., Schirrmeister, R. T., Chrabąszcz, P., Wilson, D., Boedecker, J., Schulze-Bonhage, A., . . . Ball, T. (2020). Machine-learning-based diagnostics of EEG pathology. NeuroImage, 220, 117021. doi:https://doi.org/10.1016/j.neuroimage.2020.117021
Hicks, C. W., & Selvin, E. (2019). Epidemiology of Peripheral Neuropathy and Lower Extremity Disease in Diabetes. Current Diabetes Reports, 19(10), 86-86. doi:10.1007/s11892-019-1212-8
- Optimized network stimulation in the human brain
Teachers:
Bettina Schwab
Student assignments:
MSc project assignment (exceptions possible for BSc projects and internships)
Educational program:
Biomedical Engineering, Electrical Engineering, (M3 for Technical Medicine)
Topics:
Transcranial alternating current stimulation (tACS), deep brain stimulation (DBS), brain-computer interfaces, neuromodulation, neural network dynamics, EEG data analysis
Required pre-knowledge:
Programming in matlab, signal analysis
Introduction:
We are working on a range of different topics related to brain stimulation and modulation of networks in the brain. Using computational (neural network and neural mass) modeling and field simulations, we aim to predict and further enhance the effects of this stimulation, and to use it specifically to steer functional connectivity. Thus, projects range from a technical level via quantitative neurophysiology to clinical applications. Dependent on the project, collaborations with the University Medical Center Hamburg-Eppendorf (Germany) and the University of Oxford (UK) are possible.
Motivated students are encouraged to send me an e-mail (b.c.schwab@utwente.nl) with their interests, course list and CV. Projects can then be adapted to personal interests and skills.
Example assignments:
- Development of a software phase-locked loop to couple DBS and tACS
- Measurement & analysis of tACS effects on EEG dynamics in healthy control participants
- Computational modeling to predict and enhance the effects of tACS
- Electric field simulations for tACS and/or DBS
- Closed-loop stimulation dependent on EEG or kinematic signals
- Clinical optimization of brain stimulation for Parkinson’s disease or stroke
Email contact: b.c.schwab@utwente.nl
- Oncologie (BSS-ZGT)
Educational programme: BMT, TG, HS.
a) Optimizing health outcome in (neo)adjuvant treatment for breast cancer patients.
Background information:
In the Netherlands 1, out of 7 women will develop breast cancer, in 2017 17.423 new cases were diagnosed. A substantial part of these women are treated with (neo) adjuvant systemic therapy, hormonal- and/or chemotherapy. In case of (neo)adjuvant chemotherapy, this is accompanied by a significant loss in physical condition and by a substantial weight gain. In part these women experience chronic fatigue afterwards, loss of concentration and difficulty in regaining normal activity at work and at home. Our understanding is that if preventive measures are taken the loss in physical condition and weight gain can be (partially) prevented.
Previous exercise and nutritional habits combined with lifestyle are important factors in maintaining, as good as possible, weight and physical fitness. The majority of women with breast cancer have some overweight and they do not engage in enough physical activity.
Tailoring for which patient and at what moment intervention is needed is difficult at this moment by lack of tools and practical guidelines. Recently a personalized and technology-supported coaching system to support the patient’ self-management is developed for diabetes patients. A comparable system can be used for oncology patients treated with chemotherapy.
Supervisors: Dr. Laverman – g.d.laverman@utwente.nl, Dr. Oving.
b) Impact of (Neo)Adjuvant Chemotherapy on Long-Term Performance and Employment of Early-Stage Breast Cancer Survivors in our region.
Background information:
Many women with early-stage breast cancer are working at the time of diagnosis and survive without recurrence. The short-term impact of chemotherapy on employment (<1 year) has been demonstrated, but the long-term impact merits further research. Even less is known about long-term impact of cancer treatments on social functioning, physical activity or sports, managing family live and maintaining relationships. Recently a database was made of all our breast cancer patients. This database can be used for further research.
Supervisors: Dr. G. Laverman, Dr. Irma Oving, Dr. Ester.Siemerink
- Daily health monitoring using contactless sensing techniques
Student assignment: Master graduation project (40-45 credits, 28-32 weeks)
Educational program: Embedded Systems, Electrical Engineering, Biomedical Engineering, or other relevant direction
Supervision team: Dr. Ying Wang (BSS, UTwente) and Dr. Yang Miao (RS, UTwente)
Project background:
Continuous health monitoring plays an essential role in timely and personalized disease prevention. Daily remote monitoring with advanced sensing techniques can help increase individuals’ quality of life and release the burden of national health care systems. Currently, two types of sensing techniques—wearable and contactless sensing—have been used for health monitoring. Compared with wearable sensing techniques, contactless sensing can provide users comfort and natural feeling meanwhile reducing the device-management burden of healthcare professionals. However, the technical usage of contactless sensing systems, e.g., radio-based techniques, is still not sufficiently validated in daily health monitoring.
What you can expect from us:
We aim to investigate the technical usage of contactless sensing system in daily health monitoring. Through a radio-based contactless system, vital signs (e.g., heart rate and breath rate) and movement signals will be collected from experiment subjects. Daily health monitoring system will be developed in this project, and the project results will be transferred to clinical applications in future.
In this project, you can expect to obtain the knowledge about:
- Design a pilot experiment for healthy subjects to simulate the activities of in-\out-hospital patients.
- Collect multimodal physiological signals from healthy subjects to learn about the potential effects of real-life factors on the quality of signals.
- Develop an algorithm to track the physiological states of individuals:
- Preprocess different modal signals. Extract clinical-relevant features from the signals. Apply machine learning techniques to classify different physiological states.
- A potential scientific publication can be expected based on the project outcomes.
What we expect from you:
- We are looking for talented and open-minded students who have a background in embedded systems, electrical engineering, biomedical engineering, or other relevant direction.
- Students should have a strong interest in signal processing, analysis, and machine learning.
- Students should have strong programming skills in Matlab\Python.
- Experience in physiological signal analysis and\or clinical research is optimal.
Information and application:
Please send your application to Dr. Ying Wang (ying.wang@utwente.nl), and include:
- A curriculum vitae including your name and contact (max 2 A4 pages).
- A personal motivation letter (max 1 A4 page).
- Lists of courses and grades of your BSc and MSc degrees.
- Design of an interactive tool for patient - healthcare professional consultations (RE-SAMPLE project)
Supervisors:
Roswita Vaseur, Wendy Oude Nijeweme - d’Hollosy, Monique Tabak
Student assignments:
BSc and MSc project assignment
Educational programs:
Interaction Technology, Industrial Design Engineering, Communication Science
RE-SAMPLE Project description:
Chronic Obstructive Pulmonary Disease (COPD) is a common, progressive lung condition with a high impact on quality of life and life expectancy. Many patients with COPD have multiple chronic conditions, like diabetes, cardiovascular diseases or mental health issues. The care of patients with COPD and co-morbid chronic conditions is very complex. There are overlapping risk factors and symptoms which can delay the selection of appropriate treatment. Furthermore, multiple healthcare professionals are involved in the treatment of people with COPD. The challenge of the increasing number of patients with COPD and multi-morbid chronic conditions requires an integrated, personalized, approach to support and manage care for these patients. RE-SAMPLE will work to transform the healthcare journey of patients with other chronic conditions by using real world data (RWD) to monitor symptoms beyond scheduled medical check-ups, to provide doctors, caregivers, and patients a unique insight into common, day-today triggers that can lead to health complications and to support personalized treatment and develop a virtual companionship program.
Student assignment description
The objective of this project is to develop an interactive tool that will be used during consultation with a healthcare professional. This tool will visualize the progression and burden of disease to set goals and decide on a treatment plan together. To develop our interactive tool, students will work on:
Focus groups
- Developing and conducting a protocol for focus groups with patients and healthcare professionals to explore presentation of information (communication strategies) for shared decision making. For example; identifying what input we need for the interactive tool or what kind of information doctors and patients with COPD would like to see in this tool.
Co-designing and prototyping
- Based on the information elicited during focus groups, the interactive tool will be developed based on co-designing and prototyping which include visualizing shared decision making processes.
Contact:
Roswita Vaseur (r.m.e.vaseur@utwente.nl)
Wendy Oude Nijeweme - d’Hollosy (w.dhollosy@utwente.nl)
Monique Tabak (m.tabak@utwente.nl)
- Coaching strategies for patients with COPD and co-morbid chronic conditions (RE-SAMPLE project)
Supervisors:
Roswita Vaseur, Wendy Oude Nijeweme - d’Hollosy, Monique Tabak
Student assignments:
BSc and MSc project assignment
Educational programs:
Health Sciences, Biomedical Engineering
RE-SAMPLE Project description:
Chronic Obstructive Pulmonary Disease (COPD) is a common, progressive lung condition with a high impact on quality of life and life expectancy. Many patients with COPD have multiple chronic conditions, like diabetes, cardiovascular diseases or mental health issues. The care of patients with COPD and co-morbid chronic conditions is very complex. There are overlapping risk factors and symptoms which can delay the selection of appropriate treatment. Furthermore, multiple healthcare professionals are involved in the treatment of people with COPD. The challenge of the increasing number of patients with COPD and multi-morbid chronic conditions requires an integrated, personalized, approach to support and manage care for these patients. RE-SAMPLE will work to transform the healthcare journey of patients with other chronic conditions by using real world data (RWD) to monitor symptoms beyond scheduled medical check-ups, to provide doctors, caregivers, and patients a unique insight into common, day-today triggers that can lead to health complications and to support personalized treatment and develop a virtual companionship program.
Student assignment description
The objective of this project is to develop a high-level overview of a health coaching strategy for patients with COPD and co-morbid chronic conditions. Personalized coaching actions and lifestyle coaching will be specified to counteract or delay changes in disease progression. To develop our coaching strategy, students will:
1. Identify what coaching domains are relevant for patients
Examples of coaching domains are physical activity, mental health promotion, sleep behavior, coping, nutrition and smoking cessation. The student will possibly help with analyzing and processing data of interviews with patients to decide which domains should be offered to patients.
2. Provide content to the identified coaching domains.
After relevant coaching domains are identified, the content for these domains must be designed. Relevant aspects which may be considered for the content are:
a. Purpose/goals
b. Requirements
c. Treatment
d. Behavior Change Techniques
e. Measurable parameters
f. Delivery of information
g. Etc.
3. Personalize the coaching domains
After the relevant coaching domains are identified and the content is designed, the coaching domains will be personalized based on patient characteristics and preferences (applying elicitation methods). Thus, specific patient profiles will be designed and adjusted based on patient characteristics and preferences.
Contact:
Roswita Vaseur (r.m.e.vaseur@utwente.nl)
Wendy Oude Nijeweme - d’Hollosy (w.dhollosy@utwente.nl)
Monique Tabak (m.tabak@utwente.nl)
- Sleep monitoring: classification of sleep apnea events from physiological signals (Onera)
Teachers: Bert-Jan van Beijnum
Student assignments: MSc project assignment
Educational programme: EE, BME
Topics: sleep apnea, detection algorithms, signal analysis
Project Summary:
Sleep apnea is a common sleep-related breathing disorder which leads to a temporary obstruction of the upper airways during sleep. Despite its temporary nature, sleep apnea has a dramatic impact on people’s lives. Early symptoms include daytime sleepiness and reduced concentration and learning abilities. If it remains untreated, sleep apnea can lead to health-threatening diseases, such as cardiomyopathies, stroke, and pulmonary hypertension. Early detection and classification of the sleep apnea events are the first necessary steps to prompt treatment and prevent the development of more severe conditions.
At Onera, the future of sleep medicine is being developed: a sleep diagnostic system composed of wearable devices able to reliably acquire a comprehensive picture of a subject’s sleep. Physiological signals are recorded and processed by machine learning algorithms to automatically detect sleep disorders. The figure below shows a number of typical examples of apnea when measuring the nasal airflow and abdominal movement.
Figure 1: Types of sleep apnea. Changes in breathing pattern as recorded from nasal airflow and abdominal movements. Figure is adapted from Scientific Reports 2020 Kang Sun et al.
The objective of this master thesis is the development and validation of an algorithm for the automatic detection and classification of apnea events using data acquired from the Onera Sleep Test System, a multi-sensor diagnostic system.
The research will address the following questions:
- What is the current state of the art in sleep apnea detection algorithms?
- Which signal conditioning methods need to be applied to accurately and reliable detect apnea biomarkers (e.g., various signal processing filters, noise rejection/reduction methods, time – frequency analysis and processing)?
- What are the evaluation criteria needed to evaluate the performance of classification algorithms?
- Which classification algorithm to detect and classify apnea event perform best, and to what extend do apnea biomarkers improve classification results?
It is envisioned that the report of the research will be in the form of a scientific paper.
The following skills are required:
- Following one of the following educational programmes: Biomedical Engineering or Electrical Engineering with demonstrated interest in biomedical applications.
- Good knowledge in Python (scikit-learn, pandas, matplotlib, seaborn) or Matlab.
- Interest and experience in signal processing, and machine learning (specifically classification)
- Good English communication skills.
- Positive attitude, can-do mentality, and flexibility.
Contact
The UT contact persons for this Master assignment are:
- Bert-Jan van Beijnum (b.j.f.vanbeijnum@utwente.nl)
The Onera contact persons for this assignment are:
- Flavio Raschella’ (Flavio.Raschella@onerahealth.com)
- Tineke de Vries (Tineke.deVries@onerahealth.com)
- Detection and eventual prediction of freezing-of-Gait in Parkinson's Disease
Supervisor: Juan Delgado Teran: j.d.delgadoteran@utwente.nl
Student assignment: Master BMT
Principal investigator: Prof. Dr Richard van Wezel: r.j.a.vanwezel@utwente.nl
Detection and eventual prediction of freezing-of-Gait in
Parkinson’s Disease
My current research focuses on the detection and eventual prediction of Freezing-of-Gait in people living with Parkinson’s Disease. Making use of wearable sensors, such as insoles, ankle sensors and smartphones, we investigate freezing episodes in an at-home situation.Freezing-of-Gait detection in Parkinson’s Disease from wearable sensors
This project, “PROMPT” (Personalised care and Research On Motoric-dysfunctioning for Patient-specific Treatments), aims to better understand and diagnose a symptom of Parkinson’s disease – freezing-of-gait – in Parkinson’s disease https://youtu.be/3-wrNhyVTNE, in order to provide for personalized care.
Making use of the eHealth House at University of Twente www.utwente.nl/en/techmed/facilities/htwb-labs/ehealth-house/, your tasks are:1- to assist in the recruitment of participants, and 2- to assist in the collection and post-processing of data from participants. In this study, we aim to include a total of 25 individuals with Parkinson’s disease. During data collection, various motion sensors are attached to the joints of a person, who is then asked to perform everyday tasks (e.g. cooking, cleaning, etc.) in the eHealth House, as well as take a 10-minute walk outside. In the eHealth lab, videos are taken of the individual moving around, in order to label any freezing episodes. The labelled video data is then utilized to train classification algorithms to detect freezing-of-gait based on motion sensor data.
Experience with recruitment and management of Parkinson’s patients and/or usage of movement sensors would be ideal but is not required.
Start period: 09-2021
- The national ArmCoach4Stroke project: developing an innovative sensing system for measuring arm exercises in stroke patients (BSS & Erasmus MC)
Supervisors: dr. Ruben Regterschot, dr. Bert-Jan van Beijnum.
Type of assignments: bachelor projects, master projects, internships.
Background student: biomechanical engineering, biomedical technology, movement science, health sciences, technical medicine, biomechanics, computer science, data science, or other relevant direction.
Project description & project aims:
The student projects contribute to the national ArmCoach4Stroke project: a large research project where multiple leading universities (UTwente, TU Delft, Erasmus MC, Amsterdam UMC), companies, and rehabilitation centres work closely together to develop the ArmCoach4Stroke (see Figure). The ArmCoach4Stroke is an innovative and new medical device that enables arm rehabilitation after stroke in the home environment.
The student projects focus on the development and validation of algorithms for the recognition of arm exercises and the calculation of arm exercise metrics based on data from body-fixed motion sensors (IMU sensors). In the project, you will develop algorithms for the recognition of specific arm movements (e.g., reaching) and the calculation of arm exercise metrics (e.g., the time duration of a reaching movement) based on motion sensor data. You will evaluate the sensor-based method in the lab of UTwente in comparison to camera data (Vicon system).
In the project, you will gain extensive experience with signal analysis, algorithm development (in Matlab, Python, R), sensor fusion, machine learning, body-fixed motion sensors (Xsens), camera systems for movement analysis (Vicon), and the biomechanics of the arm. This project provides you with an exciting opportunity to train research skills and to work on a national research project with leading technical and clinical universities. The project is technically oriented but with a clear clinical application as goal.
Contact:
Dr. Ruben Regterschot (University of Twente & Erasmus MC): g.r.h.regterschot@utwente.nl
Dr. Bert-Jan van Beijnum (University of Twente): b.j.f.vanbeijnum@utwente.nl
- Personalized cAnceR TreatmeNt and caRe (PARTNR)
Assignment: Bachelor, Master, Internship
Educational program: BME, TM, HS, PSY, CREATE
Daily supervisor: Kim Wijlens, MSc or ir. Lian Beenhakker
Project description
Due to improved treatment and early diagnosis of breast cancer, there are more survivors. However, this also means that more patients are struggling from the late effects of treatment. One of these late effects is cancer-related fatigue (CRF), which is defined by the American National Comprehensive Cancer Network as “a distressing, persistent, subjective sense of physical, emotional and/or cognitive tiredness or exhaustion related to cancer or cancer treatment that is not proportional to recent activity and interferes with usual functioning.”
In the Personalized cAnceR TreatmeNt and caRe (PARTNR) project, we aim to help breast cancer patients suffering from CRF. We will holistically assess patients, considering all aspects that can cause/influence the fatigue and predict who might be at risk of developing CRF. Using multimodal data, we will advise an intervention to reduce the fatigue which is based on the holistically assessed patient information about CRF and personal preferences patients might have for types of intervention. In the end, we will combine all information from patients into an intelligent self-learning platform that patients can use to keep track of and improve their fatigue.
The PARTNR project is a collaboration between the University of Twente, Ziekenhuis groep Twente (ZGT), Helen Dowling Institute (HDI), Roessingh Rehabilitation Centre, University Medical Center Groningen (UMCG), Dutch breast cancer association (BVN), Netherlands comprehensive cancer organisation (IKNL), Ivido and Evidencio.
If you would like to work on the PARTNR project during your internship or thesis, please contact either Kim Wijlens (k.a.e.wijlens@utwente.nl) or Lian Beenhakker (l.beenhakker@utwente.nl). Supervisors will depend on the chosen assignment.
PARTNR Team:
dr. Annemieke Witteveen (BSS), prof. dr. Sabine Siesling (HTSR), dr. Christina Bode (PGT),
- An Enriched Stimulation Protocol to the NDT-EP Method for Improved Observation of Nociceptive Evoked Potentials
Teachers: Jan R. Buitenweg, j.r.buitenweg@utwente.nl
Educational Programme: BSc - BMT, TG
BackgroundChronic pain often results from disturbed processes in the central nervous system. Once chronic pain is established, treatment is relatively ineffective, with – at best – one patient in three or four achieving 50% pain intensity reduction. Early detection and therapeutic action would mean better treatment outcomes and fewer clinical efforts per patient, but appropriate diagnostic tools are lacking.
Increased sensitivity to noxious stimuli is widely recognized as a key factor in chronic pain development. Noxious stimuli are processed by neural mechanisms at several stages in the ascending pathway from the periphery to the brain, into a conscious pain experience. As a response to injury or disease, maladaptive changes in this pathway may result in increased pain sensitivity. Clinical observation of the specific malfunctioning of peripheral and central components of this pathway is limited at present but would permit a better understanding and early selection of interventions for treatment or prevention of chronic pain.
Recently, we developed a new method for observing the properties of nociceptive processing utilizing subjective detection of electrocutaneous stimuli in combination with objective neurophysiological brain responses. In this method, nociceptive afferents are activated by temporally defined current stimuli with a varying number of pulses and varying interpulse intervals. As these different temporal stimulus properties result in different excitation of nociceptive processing mechanisms of the ascending system, subsequent processing of stimulus-response pairs (SRPs) into estimated nociceptive detection thresholds (NDTs) and Evoked brain Potentials (EPs) of multiple stimulus types may provide information about the properties of these mechanisms.
The EPs comprises mainly of two elements: (1) activity related to the processing of the nociceptive stimulus and (2) task-related activity. At present, the majority of the activity seems to be related to the task. For an improved characterization of the nociceptive processing, we would like to better evaluate the activity related to the processing of the nociceptive stimulus. By providing an enriched stimulation protocol to the participant, we might be able to reduce the extent of task-related activity. This could allow us to better evaluate the activity related to nociceptive processing. Thus far, we have however not experimentally evaluated this protocol. In this study, we want to evaluate whether this enriched stimulation protocol is feasible and useful.
Assignment – Prepare and execute experiments at the University of Twente to evaluate the effect of the enriched stimulation protocol on the EP's.
Components – (1) Perform a literature study. (2) Implement and conduct a human subject experiment. (4) Analyze the data and (5) report and (6) present the results. - Motoneuron adaptations to exoskeleton training
Student assignment: Master assignment (28-32 weeks, 40-45 ECs)
Contact: Dr. Yavuz US.
email: s.u.yavuz@utwente.nl
Background
Lower limb rehabilitation exoskeletons support and assist patients in locomotion after neurological injuries. Identifying the mechanisms underlying neural adaptations to exoskeleton training is key to design optimal strategies for assistance. Previous studies1,2 indicate reduced muscle activation (EMG) after short-term exoskeleton training. However, the specific mechanisms for this adaptation remain unclear.
High-density surface EMG (HD-EMG) is a non-invasive technique to measure neural activity from multi-channel grids on the muscle. Blind source separation (BSS) techniques, such as Convolution Kernel Compensation (CKC) 3, enable exploiting the high-resolution capability of HD-EMG to open a window into constituent motor neuron spike trains. By these means, we aim at investigating the neural mechanisms that govern locomotor adaptations to short-term exoskeleton training.
Goal: The main goal of this assignment is to develop and validate methodologies for identifying motoneuron adaptations to short-term exoskeleton training.
Keywords: HD-EMG, motor neuron, decomposition, blind source separation, exoskeleton training
Main Activities:
· Literature study on motor neuron properties (recruitment threshold, variability, discharge rate and synchronicity, stretch reflex, H-reflex and exoskeleton training.
· Development of a methodology to identify physiological changes between before and after-training trials (in time or frequency domain).
· Statistical analysis to compare results.
· Scientific report about the findings.
References:
1. Kao PC, Lewis CL, Ferris DP. Short-term locomotor adaptation to a robotic ankle exoskeleton does not alter soleus Hoffmann reflex amplitude. J Neuroeng Rehabil. 2010;7(1):33. doi:10.1186/1743-0003-7-33
2. Gordon KE, Ferris DP. Learning to walk with a robotic ankle exoskeleton. J Biomech. 2007;40(12):2636-2644. doi:10.1016/j.jbiomech.2006.12.006
3. Holobar A, Zazula D. Gradient Convolution Kernel Compensation Applied to Surface Electromyograms. Indep Compon Anal Signal Sep. 2007:617-624. doi:10.1007/978-3-540-74494-8_77
- Optimized network stimulation in the human brain
Teachers:
Bettina Schwab
Student assignments:
MSc project assignment (exceptions possible for BSc projects and internships)
Educational program:
Biomedical Engineering, Electrical Engineering, (M3 for Technical Medicine)
Topics:
Transcranial alternating current stimulation (tACS), deep brain stimulation (DBS), brain-computer interfaces, neuromodulation, neural network dynamics, EEG data analysis
Required pre-knowledge:
Programming in matlab, signal analysis
Introduction:
We are working on a range of different topics related to brain stimulation and modulation of networks in the brain. Using computational (neural network and neural mass) modeling and field simulations, we aim to predict and further enhance the effects of this stimulation, and to use it specifically to steer functional connectivity. Thus, projects range from a technical level via quantitative neurophysiology to clinical applications. Dependent on the project, collaborations with the University Medical Center Hamburg-Eppendorf (Germany) and the University of Oxford (UK) are possible.
Motivated students are encouraged to send me an e-mail (b.c.schwab@utwente.nl) with their interests, course list and CV. Projects can then be adapted to personal interests and skills.
Example assignments:
- Development of a software phase-locked loop to couple DBS and tACS
- Measurement & analysis of tACS effects on EEG dynamics in healthy control participants
- Computational modeling to predict and enhance the effects of tACS
- Electric field simulations for tACS and/or DBS
- Closed-loop stimulation dependent on EEG or kinematic signals
- Clinical optimization of brain stimulation for Parkinson’s disease or stroke
Email contact: b.c.schwab@utwente.nl
- Validatie van biosensoren in gezonde vrijwilligers
Research Question: Wat is de accuraatheid en betrouwbaarheid van de Everion biosensor en de Fitbit Charge 3 in gezonde vrijwilligers?
Department: Chirurgie Universitair Medisch Centrum Groningen
Language: Dutch/English
Educational programme: BME
Description project content and methods:
Bij de chirurgie loopt een groot project waarbij telemonitoring toepassingen worden gevalideerd en geïmplementeerd voor patiënten die grote operaties ondergaan. Het gebruik van telemonitoring, als onderdeel van eHealth, komt tegemoet aan de tendens naar personalized medicine en is geassocieerd met betere klinische uitkomsten en kosteneffectiviteit van zorg in verschillende takken van de gezondheidszorg, met name in chronische ziekten. Het inzichtelijk maken van het welzijn van een patiënt zorgt ervoor dat eerder of beter omgegaan kan worden met veranderingen in gezondheid. De verwachting is dat telemonitoring ook in de perioperatieve periode, van toegevoegde waarde kan zijn. Telemonitoring zou kunnen bijdragen aan de optimalisatie en individualisatie van het perioperatieve (herstel)proces en daarbij te vroeg ontslag, onnodig lange ziekenhuisopnames of ongeplande heropnames kunnen reduceren.
Op dit moment zijn draadloze biosensoren beschikbaar om fysieke parameters te monitoren. De Everion biosensor van Biovotion (Biovotion AG, Zürich, Switzerland) is een CE-gecertificeerde sensor die op de bovenarm wordt gedragen (zie afbeelding) en vitale parameters en activiteit meet met een frequentie van 1Hz, waaronder: hartslag, hartslag variabiliteit, oxygenatie, huid temperatuur, ademhalingsfrequentie, aantal stappen en type activiteit. Hartritme is al een gevalideerde parameter. Bekend is dat de hoeveelheid postoperatieve dagelijkste stappen en fysieke activiteit van een patiënt tijdens ziekenhuisopname geassocieerd is respectievelijk het risico op heropname binnen 30 dagen na de operatie en de lengte van de ziekenhuisopname. Om deze informatie te implementeren in de perioperatieve zorg van hoog-risicopatiënten, moeten in kaart worden gebracht wat de betrouwbaarheid van verschillende sensoren is wat betreft het meten van vitale parameters, het aantal stappen en activiteit (en de gevoeligheid van vitale parameters voor activiteit). Daarom gaan we een validatiestudie doen bij gezonde vrijwilligers.
De studenten zullen telemonitoring devices valideren door het uitwerken van strategieën om looppatronen te simuleren, mee ontwikkelen van protocol en meetopstelling en uitvoeren van tests bij gezonde vrijwilligers. In dit project hebben studenten de mogelijkheid om zelf metingen uit te voeren en kennis te maken met de faciliteiten van het eHealth house (Techmed Centre, Universiteit Twente) en de nieuwste ontwikkelingen op het gebied van telemonitoring.
Supervisors:
dr.ir. M. Tabak, associate professor vakgroep Biomedische Signalen en Systemen, UTwente
dr. R.C.L. Schuurmann, post-doc multi-modality medical imaging group, UTwente
prof. dr. J.P.P.M. de Vries, vaatchirurg UMCG
M.E. Haveman, MSc, promovendus UMCG
Contact e-mail: m.e.haveman@umcg.nl / m.tabak@utwente.nl
- Functional electro-tactile stimulation to augment the postural control: development of rehabilitation technology fall
Contact person: Utku Yavuz
Student assignments: MSc project assignment
Educational programme: BME, TM, EE, Create
Project Summary:
The dramatic increase in average life expectancy during the 20th century is one of society’s greatest achievements. In many countries, the oldest age (people aged 85 or older) are now the fastest-growing proportion of the total population (the 85-and-over population is projected to increase 351% between 2010 and 2050). As people grow older they are increasingly at risk of falling with resultant injuries. One out of three persons older than 65 and 50% of those older than 80, fall at least once per year (Todd & Skelton, World Health Organization, 2004) based on various reasons. In some cases, fall may be an indication of chronic neurological disorders affecting postural stability and gait (such as Parkinson disease). Therefore, understanding the neural mechanism underlying the postural control has utmost importance to develop an accurate intervention.
It is well established that the sensory information conveyed from muscle and joint proprioceptors play an important role in the control of posture and gait in humans. In a recent study, we showed that cutaneous mechanoreceptive inflow from the neck is also integrated into the control of posture. More importantly, we showed that the electro-tactile stimulation of cutaneous afferents at the neck region, with a subtle (140% of perception threshold) stimulation current intensity, led to a drift towards forward (leaning forward) in the centre of pressure (Nunzio et.al., Experimental Brain Research 2018). Within the proposed study, we aim to investigate the integration of cutaneous receptors located at the foot sole and determining the pattern and strength of stimulation current that leads to an efficient effect on body orientation (posture). The findings are relevant for the exploitation of electro-tactile stimulation for rehabilitation interventions where induced body orientation is desired.
Assignment:
In this study, you will use biomedical equipment to record the electrical activity produced by muscles contraction (electromyography - EMG) and to acquire the oscillations of the human body through force platforms during quite upright stance conditions. You will then administer a series of electro-tactile stimulation and analyze the postural effects induced by such stimuli.
The acquired data will be analyzed with standardized and validated algorithms to study the physiological basis of these postural effects. This will lead to the development of new rehabilitation protocols and devices to enhance postural control in the elderly population and neurological patients. You will learn:
- neurophysiology of postural control
- recording muscle electrical signals and the movement of the centre of foot pressure using professional biomedical equipment
- developing acquisition systems with stimulation units
- statistically analyze the acquired data
Contact:
The student will work at the laboratory of the Department of Biomedical Systems and Signals, faculty of EEMCS, University of Twente, under the supervision of S. U. Yavuz, PhD. email: s.u.yavuz@utwente.nl, tel:+31534898158
- E-manager Chronic Diseases
Teachers: Wendy d’Hollosy, Anouk Middelweerd en Eclaire Hietbrink
Student assignments: BSc and MSc project assignments
Educational programme: HS, TG, CS, HMI/Create, BME
Topics: tailoring and personalization, technology supported lifestyle coaching, experimental designs, mHealth, machine learning, human computer interfaces
Project Summary:
Over 5.3 million people in the Netherlands have a chronic disease. The importance of a healthy lifestyle in chronic disease management has been increasingly recognized in recent years. There is a growing body of evidence that lifestyle interventions contribute to a reduction in disease burden and to an improvement in the Quality of Life (QoL) in people with chronic diseases. eHealth technologies show promising results regarding lifestyle coaching. Compared to regular face-to-face lifestyle coaching, eHealth interventions have several advantages as eHealth is less effort-, time- and cost intensive and enables more continuous support in chronic disease management in daily life.
For this reason, this project focuses on the development, implementation and evaluation of the E-manager Chronic Diseases for people with type 2 diabetes mellitus, asthma, chronic obstructive pulmonary disease (COPD) and heart failure. In the E-Manager project, a digital coaching platform, the E-Supporter, is being developed for people with chronic diseases who want to improve their lifestyle. The E-Supporter motivates and helps patients in everyday life to pursue lifestyle goals by providing tailored e-coaching that ties in with the patients’ behavioural state, character and abilities. The coaching content that is developed will be integrated into existing apps. For example, the current content of the E-Supporter for people with type 2 diabetes mellitus is built into the “Diameter” app and the “MiGuide” app.
Assignments:
The E-Supporter has not yet been fully developed and evaluated, which is why bachelor and master assignments are available focusing on different topics:
· Developing and evaluating more tailored coaching content for physical activity and/or nutrition;
· Broadening of the existing coaching content to other chronic conditions (e.g., COPD) or lifestyle domains (e.g., smoking);
· Developing a rule based decision support system to automate the provision of tailored lifestyle coaching;
· Using data science techniques to tailor the coaching strategies to an individual patient.
Email contact: a.middelweerd@utwente.nl
- Sensing and analyzing (sports) biomechanics with (inertial) sensors
Possible project supervisors:
Bouke Scheltinga B.Scheltinga@rrd.nl
Robbert van Middelaar R.P.vanmiddelaar@utwente.nl
Jasper Reenalda J.Reenalda@rrd.nl
Educational programmes: EE and BMT
Topic: Sensing and analyzing (sports) biomechanics with (inertial)sensors
Project summaries:
Several PhD students are involved in research related to (sports) biomechanics and (inertial) sensors. Below you can find a short description of the different topics. If you have an interest in a Bachelor of Master project related to one of these topics, please contact the PhD student mentioned by the project.
Bouke Scheltinga: Model and quantify physiological and biomechanical training load over time using workload data. Ideally with a minimal sensor setup, which can consist of inertial sensors and physiological sensors.
Robbert van Middelaar: Development of new algorithms and tools to monitor an athlete (runner, rower, volleyball etc.) in the sports-specific setting to improve their performance or to prevent them from injuries; detect changes in their movement or posture and give feedback via an interactive system". (www.sports-data-interaction.com)
The project is related to 'Sports, Data & Interaction Team', meaning there is close collaboration with projects from Human Media Interaction and Mathematics.
- Pediatric asthma monitoring
Teachers: Monique Tabak, Mattienne van der Kamp
Student assignments: BSc and MSc project assignments
Educational programme: BMT, TG, EE, TCS, CT, CS, ATLAS
Contact: m.tabak@utwente.nl
Topic: eHealth in the pediatric asthma care: From smart monitoring to implementing a new care standard for personalized health.
Project Summary:
This project focuses on innovative, personalized eHealth for pediatric asthma patients. Currently, pediatric asthma care heavily relies on information acquired during scheduled visits when children usually are asymptomatic. Relevant (symptomatic) information can therefore be missed due to for example recall bias, hampering the treatment of the child. Smart monitoring (e.g. sensing, ESM, video) allows regular monitoring of asthma control at home and helps to identify worsening of asthma control (e.g. prediction modelling, decision-support). This enables the pediatrician to anticipate timely for personalized treatment and self-management of the child and parents, potentially resulting in less symptoms and a better quality of life.
The challenges of this project lie in;
· finding the right combination of monitoring tools and devices for the right physiological parameters (i.e. spirometry, smart inhalers, activity, EMG, etc),
· developing a child-friendly and attractive monitoring platform,
· creating evidence-based clinical decision support algorithms,
· increasing self-management of asthmatic children and parents (with i.e. automated feedback),
· and implementing eHealth in the current clinical practice in a clear and efficient way.
During the course of this project, we as Medisch Spectrum Twente, University of Twente and Roessingh Research and Development devoted ourselves to many of the above-mentioned challenges and are working hard to progress towards extending, personalizing and validating eHealth strategies in pediatric asthma care. So that it benefits both the caregivers and the care providers.
- DIAMETER (Diabetes) (BSS-ZGT)
Educational programme: BMT, TG, HMI, ATLAS
Achtergrond en probleem
Diabetes Mellitus Type 2 (T2DM) is een chronische ziekte waarbij het risico op complicaties is verhoogd door ontregelde glucosewaarden. Deze patiënten hebben hulp nodig om meer grip te krijgen op de diabetes. Een belangrijke factor dat winst kan opleveren bij T2DM is een gezondere leefstijl, echter ontbreekt de tijd in de gezondheidszorg om patiënten hier optimaal in te kunnen begeleiden. Uit vooronderzoek is gebleken dat patiënten niet voldoen aan de norm gezonde voeding en gezond bewegen en de kennis over de voordelen van een gezonde leefstijl ontbreekt. Daarom is het ZGT in samenwerking met de Universiteit Twente bezig om een gepersonaliseerde diabetes coach, in de vorm van een mobiele app, te ontwikkelen die de diabetespatiënt helpt de glucosewaarde op peil te houden. De coach, de Diameter, maakt gebruik van continue monitoring van glucosewaarden, lichaamsbeweging, hartslag, voeding en medicatie op basis waarvan een individueel voorspellend model voor glucoseregulatie wordt ontwikkeld. Dit model dient o.a. als input voor de coaching module die gebaseerd op theorieën van gedragsverandering en motivatietheorieën de patiënt ondersteunt in het maken en volhouden van verantwoorde keuzes. De verwachting is dat patiënten hiervan voordelen zullen merken op de korte termijn (groter gevoel van welbevinden) en lange termijn (verminderde kans op complicaties).
Doel
Het doel van het onderzoek is het ontwikkelen van de mobiele applicatie die diabetes patiënten kan ondersteunen. De ontwikkeling van de app bestaat uit meerdere subonderdelen. De verschillende app versies moeten bijvoorbeeld getest worden op usability, de coachingsmodule moet (verder) ontwikkeld worden en verschillende onderdelen van de app, zoals ziektebeleving en mogelijkheid tot een community, moeten nog vormgegeven worden.
Niveau: Bachelor/master
Supervisor: Dr. G. Laverman
UT-BSS, ZGT Almelo, afdeling: nefrologie
E-mail: g.d.laverman@utwente.nl of g.laverman@zgt.nl
- Modeling and analyzing tremor related phenomena of Parkinson’s disease
Teachers: Ciska Heida
Student assignments: BSc and MSc project assignments
Educational program: BMT/BME, TG/TM, EE
Topics: Modeling and signal analysis
Introduction
Parkinson's disease (PD) is a long-term degenerative disorder of the central nervous system that affects the motor system. The cardinal symptoms are (rest) tremor, rigidity (muscle stiffness), and akinesia/bradykinesia (lack or slowness of movement, resp.). These symptoms result from the loss of cells in the substantia nigra, a region of the midbrain that produces dopamine, an essential neurotransmitter for relaying neuronal information that plan and control voluntary movements.
A number of phenomena occur in Parkinson’s disease (PD) that are related to tremor, which have not yet been combined into a single conceptual and/or computational model, but probably all involve basal ganglia and cerebellar circuitry:
- (Rest) Tremor in PD is hypothesized to involve the basal ganglia as well as the cerebellum with the basal ganglia switching tremor on and off, and the cerebellar circuit modulating tremor amplitude (known as the dimmer-switch hypothesis).
- Rest tremor is reduced/suppressed by voluntary movements.
- Rest tremor disappears during sleep.
- Rest tremor may respond to Levodopa medication, but this is not necessarily the case in all patients. Furthermore, while in a single patient rigidity and slowness of movement may respond well to Levodopa, tremor may not, which may suggest that tremor is not a direct effect of the dopaminergic deficiency.
- Rest tremor responds to deep brain stimulation (DBS) in the STN and Vim.
A) Schematic overview of the connections of the basal ganglia with the thalamus and cortex, and the cerebello-thalamo-cortical circuit. (Blue: inhibitory connections; Red: excitatory connections)
B) DBS electrode in STN: continuous stimulation of the STN reduces tremor.
Assignments
- Development of a computational model of the neuronal networks involved in central motor control and couple a number of the phenomena related to tremor at the neuronal circuit level.
- Analyzing experimental data containing movement registrations performed by PD patients during different (movement) tasks that can be used to validate computational models of the mechanisms and neuronal circuits involved in tremor under different behavioural conditions.
Email contact: t.heida@utwente.nl