Computer Vision News - May 2023

37 Hypervision Surgical AI is at the heart of everything Hypervision Surgical does. The low-resolution data it captures has many inherent artifacts, but its AI tools reconstruct pristine images for surgeons. It also employs AI to analyze the captured data and extract relevant information displayed to the surgeon to enhance their intraoperative decision- making abilities . “ Instead of dividing the entire spectrum of light into three bands – the standard red, green, and blue – we split it into multiple spectral bands, ” Tom explains. “ That’s not something we can visualize as humans, as our eyes are limited to three color bands.We capture 16 and use AI to extract imaging biomarkers of interest in real-time and feed that back to the surgeon in a visually plausible manner. ” It takes roughly 20,000 hours to train as a surgeon. However much you believe in your technology, convincing surgeons to change their regular practice can be a difficult task, but it holds across the board. Whether in colorectal surgery, urology, or neurosurgery, there’s a need to help surgeons see better and understand better. ” Hypervision Surgical aims to augment the vision of surgeons and provide them with tools that are easy to use and offer real- time guidance. Its technology is based on hyperspectral imaging , a technique long used in biomedical imaging, but its surgical use has been restricted due to acquisition and reconstruction complications. “ Recently, changes on the machine vision side have led to new sensors being able to capture low-resolution hyperspectral data in real-time , ” Tom reveals. “ We took advantage of those developments, and together with our computational expertise, we could bridge the gap to deliver high- resolution real-time hyperspectral imaging . A technology that fits the surgical need! ”