Computer Vision News - November 2022

63 Qritive “ Histopathological images tend to be huge, and to be able to process all this information with an AI module, specific engineering is needed, ” Bruno explains. “ If you open up one of these images, it looks like Google Maps. You can zoom in to the maximum resolution and look at the cell morphology and very subtle details that normal or tumor cells might show. If you zoom out, you’ll see how huge the image is. To process this sort of image, you must address it at many different magnifications and levels. You start with very low magnification to get a broad sense of this, then go deeper and deeper, and aggregate all this information into insights the system generates. ” The use of AI in healthcare is still in its early stages. Its adoption has been slow compared to other industries, despite increasing interest from stakeholders. Why does Bruno think this is? “ AI in histopathology is a given. It’s coming. There’s no question about it, ” he responds. “ One of the reasons it’s been slow is because of a lack of infrastructure. At Qritive, we’re building all the important pieces of this infrastructure required to bring AI to histopathology . We’re pretty sure we’ll be successful, but it’s about the market, execution, and many other factors. We’re doing our best to deliver it. ” Qritive has already built an extensive network of interest, but with global ambitions, it is always looking for diverse partners and collaborators worldwide. “ Other companies like us are mostly in the US or Europe, whereas Qritive is in Asia, ” Bruno points out. “ In Asia, access to a pathologist is a massive challenge. You don’t have the skill sets. We want to use AI to unlock access to diagnostics. In a world where you see so many cases of cancer, the fact we can open access to diagnostics could impact healthcare dramatically, and we’re on the right side of the world to do it. ” ) unts AI Module Non-ROI Area ROI area Ki 67 Negative Ki 67 Positive a) & Non-Tumor Area Benign area Malignant area AI v1.0 9:07:44 s: High Risk 3