HOME ABOUT CV ADVENTURE CONTACT
Fluorescence Neurosurgery

Hyperspectral Fluorescence Spectroscopy for Neurosurgery

Date: January 2018 - Present

Gliomas are a type of tumor affecting glial cells in the brain. These make up 80% of malignant brain tumors and have no effective treatment options. While surgery is often performed for symptom reduction, the recurrence rate remains near 100%. This is due to the tumors' infiltrative nature and the fact that they are almost indistinguishable from healthy tissue. We are working on a novel hyperspectral imaging system to greatly improve the sensitivity of fluorescence-guided neurosurgery, allowing the surgeon to "see" differences in tissues that are not visible to the human eye.

I helped develop a device for this, which was installed at the University Hospital of Muenster, Germany. This allowed the neurosurgeons there to measure ex-vivo brain tumor tissue from hundreds of patients. With this data, we have determined multiple important factors about 5-ALA fluorescence, how to use it clinically, and how to better use the image data. We also developed several methodological improvements to the imaging system including deep learning and optimization-based data processing, machine learning-based tissue classification, and more.

I started this work in early 2018 when I did a 4-month internship in the Advanced Development department at Carl Zeiss Meditec AG in Oberkochen Germany. Here I worked in a small team on a new technology in the field of quantitative fluorescence microscopy. Upon completion of the 4-month internship, I was hired through Zeiss Canada to continue work on the project for the subsequent year (40 hrs per month), and then continued collaborating with neurosurgeons Dr. Eric Suero Molina and Prof. Dr. Walter Stummer at the University Hospital of Muenster to progress the reserach in my free time. Our publications are listed below. We have also collaborated with Dr. Pablo Valdes at the University of Texas and Profs. Antonio Di leva and Benoit Liquet at Macquarie University, Sydney, Australia.

This work has involved lab research and design work in the fields of optics and biophotonics, as well as programming and data analysis using deep learning, classical machine learning, computer vision, image processing, and augmented reality, predominantly in MATLAB and Python.

Hyperspectral Imaging

Publications

At the end of the co-op, I wrote a ~50 page report outlining our findings, remaining questions, next steps, and usage instructions so future people can continue the project. I also left almost 300 pages of documentation, notes, and logs from experiments. Finally, I was involved in several publications from the project. For these publications, I helped come up with the questions, design and build the measurement device, write all the analysis software, analyze multiple terabytes of data, and write much of the paper.