Intraoperative Stereovision for Surgical Guidance

PROJECT SUMMARY/ABSTRACT: InSight Surgical Technologies, LLC, is a start-up founded by senior members of a long-standing NIH-funded research effort to develop intraoperative image updating that maintains accurate image correspondence with the surgical field throughout a procedure. The effort has led to research innovations that have generated an intellectual property (IP) portfolio of 13 issued and 8 pending patents. Phase I demonstrated feasibility of new algorithms for assembling multiple intraoperative stereovision (iSV) acquisitions into a composite field-of-view and accomplishing iSV-to-MRI registration/re-registration, and conversion of existing iSV algorithms for 3D surface reconstruction and calibration in graphics processing unit (GPU) accelerated forms. Additionally, Phase I results obtained during resection of a large animal (swine) brain tumor indicate this model is a superb system for testing and validating InSight’s complete image updating system, I2UpdateTM, in vivo in Phase II. Specifically, we will leverage Phase I results to automate I2UpdateTM fully, and demonstrate its technical performance and clinical functionality under quality systems sufficient to meet regulatory requirements. Specific aims will (i) develop a fully-automated, GPU-enabled processing system for intraoperative image-updating (I2UpdateTM), (ii) complete verification testing and pre-clinical in vivo validation of I2UpdateTM, under quality system controls sufficient for supporting regulatory filings at the end of Phase II, and (iii) conduct a multi-site clinical-use study with I2UpdateTM under non-significant risk in patients undergoing brain tumor resection. Successful completion of Phase II will produce the I2UpdateTM product platform as a finalized commercial GPU-enabled software system with concomitant design control, software validation, procedural use and quality systems, accompanied by in vivo animal validation and multi-site, multi-surgeon use demonstrating clinically- relevant and acceptable image-updating performance in terms of automation, computational efficiency and registration accuracy.