MESA+ University of Twente
Multiscale Modeling and Simulation

BSc & MSc projects

Development of an electronic nose wavelet for medical diagnostics

Supervisors: Bernard Geurts (University of Twente),
Jan Willem Gerritsen (The eNose Company, Zutphen)

Since ancient times, it has been known that smell contains information on a person’s physical condition. A possible source of smell is the exhaled breath, containing large numbers of different volatile substances.

Detecting specific volatiles, e.g., using an electronic nose (eNose), would make it possible to distinguish between sick and healthy individuals. Such technology enables new, non-invasive diagnostic opportunities, easing patients’ lives, and supporting physicians in their diagnostic decisions. Finally, the eNose has the potential to contribute to decreasing the financial burden of healthcare for the community.

The eNose Company (located in Zutphen) is capable of manufacturing electronic noses for dedicated high-volume applications. Composed of standard electronic components, small and robust devices can be produced at low costs. Calibration models for specific diseases can be developed and subsequently transferred to an unlimited number of electronic noses. For the first time, this enables high-volume application of electronic noses.

A dedicated eNose has been developed in recent years for exhaled-breath analysis: the Aeonose. The patient should breathe gently into this battery-powered instrument for a couple of minutes. The data obtained are analysed using embedded software as well as computing power available remotely. This can give a screening result within minutes, after which further medical steps can be initiated if the results so indicate. After replacing the mouth piece and carbon filters, the device is ready for the next measurement.

A major component for the robust and reliable operation of an eNose is the analysis and interpretation of the data that are collected. Currently, data analysis is being performed using standard data compression and classification techniques. In order to achieve higher sensitivity and specificity for the device, we aim at developing a wavelet, dedicated for the signal structures of the Aeonose. For this purpose, knowledge on the excitation and chemo/ physical properties of the sensors will be incorporated to arrive at new compression methods tailored for the specific device and disease-specific volatiles.

Specific steps are:


Literature study on (a) thermo-physical properties of the Aeonose-sensors and gas mixtures and on (b) wavelet transform


Creating specific wavelet transforms for the eNose based on a mathematical model and analysis of the main processes on which it relies


Comparing results obtained with the new wavelet and existing techniques. A software module for comparing different approaches is available.


Optimization of the preferred wavelet transform and testing it on a number of diseases


Writing a report documenting the findings, developments, and recommendations

To apply for this position, please send your email to: Bernard Geurts