Computational Sensing for ISM & OCT Guided Laser Microsurgery

Lasers surgery offers precision and fewer complications, but challenges include controlling ablation depth and identifying tumor edges. Intraoperative imaging, especially optical laser scanning microscopes (LSM), such as optical coherence tomography (OCT) and image scanning microscopes (ISM), may overcome these issues. They provide benefits such as, cost-effectiveness, and no radiation exposure. This work investigates the use of OCT and ISM to enhance surgeries providing better depth perception, clearer tumor margins, and surgical precision. Transitioning optical imaging from bench-top to clinical settings requires overcoming challenges like limited focus depth, motion blur, and slow data acquisition. Computational sensing may be a solution to tackle this problem by co-designing hardware and software through techniques like optimization and machine learning. In particular, compressive sensing is of special interest here, as it enables sampling data below the Nyquist rate and is able to reconstruct original signals using computational methods. Compressive sensing is typically used in single-pixel cameras. This work explores applying compressive sensing to LSM, specifically ISM and OCT. First, we investigate how to combine laser surgery with optical imaging modalities. Then, we apply compressive sensing ISM and leverage the micro-images from a single-photon avalanche diode (SPAD) to produce better images. We address the limitations of 1D and 2D OCT scans under the separate optical path (SOP) category by using a 3D-OCT scan to determine the ablation depth. OCT scanning time is reduced by applying compressive sensing without significant loss in the quality of the depth map. Finally, we use the depth map to control the exposure time (laser on/off time) to accurately ablate a given depth across different tissue types in a feedback controller. So far, ablation over a given point has been performed using a high-speed scanner, but the ablation area remains limited. Auto-CALM extends the capability of CALM to ablate over a larger area defined by the surgeon automatically. It employs target tracking, laser tracking, and an ablation algorithm. Tested on a porcine model simulating breathing, Auto-CALM showed high precision and promised significant surgical advancements. Integrating compressive sensing with LSM modalities such as OCT and ISM is a significant step in transferring this technology from bench-top to clinical settings. We have shown that it reduces the photo-bleach of ISM, reduces the scanning time of OCT, and enables the control of the laser ablation depth. Additional efforts, such as the development of endoscopic probes that incorporate these technologies and the extension of the depth of focus, are essential for the clinical application of these advancements.