The overarching goal of our project was the completion of a device prototype suitable for intraoperative tumor surgeries. We have almost completed this and will be soon testing it as planned.
The first steps towards the implementation of the project were based on multiple visits to the hospital and careful observation of the surgical procedure involved in the removal of brain tumors. Further, through numerous meetings with the lead surgeon Prof Rachel Grossman (Head of Brain Tumor Center and Vice Chairman Department of Neurosurgery Rambam Medical Center,Haifa Israel). This project phase proved to be fundamental to reshaping our technological approach. Namely, we shifted from a scanning approach to wide-field imaging. The key benefits are speed and field of view coverage at the expense of spatial resolution. While our initial approach was to attain cellular resolution, the surgeons strongly discouraged this approach and deemed it secondary compared with imaging speed and extent of field of view. From a practical point of view, we were given a slot of 15min in the course of the surgery for completing all imaging and analysis. Our initial approach laser scanning would have taken two to three hours to complete which was prohibiting in the real conditions.
This strategic decision forced us to change hardware components while maintaining our multimodal imaging concept. For example, measurements of metabolic state were traditionally done using fluorescence life-time imaging (FLIM) which was traditionally developed for laser scanning approaches and more recent technology allow us to do so using a fast-gated sensor array. Our prototype is now based on the SPAD512S photon-counting camera from Axion (Somerville, USA) and a picosecond laser source (765nm, 70ps, 20MHz. PicoQuant, Berlin, Germany). As second fundamental development is imaging further into the SWIR band. This was achieved with a proprietary light source and a SWIR compatible camera (WiDy SensS 640, New Imaging Technologies, Verrières le Buisson, France). Combined, we are now able to perform multimodal imaging as initially planned albeit at a much faster speed making our future product a much better suit for its intended application.
Our second objective focused on the generation of tissue samples obtained during surgery and their use to create an image database. This turned to be more complicated than anticipated and our Helsinki protocol has not been approved in hospital of our clinical partner (we have submitted and obtained positive comments the ethics committee albeit in the previous hospital where Prof Grossman performed surgery). Per our contingency plan, we set to complete this task based on mouse brain tumor models used in other projects in my laboratory and with collaborators in our department (see for example our recent publication Banerjee et al 2023, Eur J Immunol). We foresee return to the original plan upon receipt of the needed Helsinki ethics approval. Another key component of this multidisciplinary component of the project was the application of advance machine learning tools to the segmentation of different components of the tissue. We have honed down our team skill in this direction and worked together with Prof. Lior Wolf from the Computer Science Department) and the use of “transformers” to analyse blood flow dynamics in volumetric brain samples (see our joint publication Choukroun et al 2023, International Conference on Medical Image Computing and Computer-Assisted Intervention).
The third objective focused on commercialization of our invention. Here we presented our work in several opportunities; two private investors and a start-up incubator (MedEx, Israel). From these series of meetings, it became clear that, a higher TRL was needed (main focus on a real scenario proof of principle).
No subcontracting was involved in the execution of this action,
