Detecting cancer with hyperspectral imaging - To improve the outcome of breast-conserving surgery
Lynn-Jade Jong is a PhD student in the Department of Nanobiophysics. (Co)Promotors are prof.dr. T.J.M. Ruers and prof.dr. M.M.A.E. Claessens from the Faculty of Science & Technology, prof.dr. H.J.C.M. Sterenborg and dr. B. Dasht Bozorg from the Netherlands Cancer Institute and Antoni van Leeuwenhoek.
Breast-conserving surgery is a crucial procedure for treating breast cancer, offering significant advantages in overcoming the disease and improving both survival rates and patients’ quality of life. With this surgery, surgeons aim to selectively remove the tumor so that the maximum amount of surrounding healthy tissue can be spared. However, surgeons often face challenges in accurately distinguishing tumor boundaries from healthy tissue, often relying solely on visual and tactile feedback. This can lead to either excessive removal of healthy tissue or inadequate tumor excision, known as a positive resection margin, both resulting in suboptimal outcomes. Excessive removal impacts cosmetic appearance and recovery time, while inadequate excision may necessitate a reoperation or additional treatments, affecting the patient’s quality of life and increasing healthcare costs.
Currently, assessment of the excised breast tissue, called a lumpectomy specimen, is performed at the pathology department, where histopathological analysis of resection margins can take several days. This delay highlights the need for a margin assessment technique that provides real-time feedback to surgeons during the surgery.
This dissertation addressed this challenge by investigating hyperspectral imaging (HSI) as a potential solution. HSI is a quick, harmless, non-invasive, and non-contact optical imaging technique that uses diffuse reflected light from a broad wavelength range to obtain spectral information of the tissue, serving as its “optical fingerprint” for enabling tissue type discrimination.
The dissertation has evaluated the diagnostic performance of HSI for margin assessment in breast-conserving surgery, aiming to bridge the gap between experimental research and clinical implementation. Several studies were conducted using a machine learning approach to advance HSI’s application for this purpose, which incorporated methodological improvements to optimize dataset utilization and tissue classification. The final part of the dissertation focused on clinical implementation, presenting a novel framework for utilizing snapshot HSI cameras to facilitate their integration into the operating room and address challenges such as motion artifacts and specular reflections during surgery.
