Israeli startup Zebra Medical Vision has two new algorithms to better detect clogged arteries and fatty liver in patients’ CT scans. The company uses deep learning and computer vision to better detect anomalies in CTs, with its development now bearing fruit. Zebra’s founder describes why, in his company’s view, such a specific set of algorithms is necessary now.
“Radiologists are trained (and have time) to look for findings related to the original reason for the scan. It is nearly impossible for them to analyze all the various possibilities and organs in the scan,” Zebra CEO and Co-founder Elad Benjamin told Geektime. That padded the need to bring in some machine backup to catch other telltale signs of illness that doctors would otherwise not look for and would likely miss.
Both algorithms have been tested against every scan that Zebra has access to, as well as scans from pilot programs with health networks and hospitals. Zebra likely was able to pitch the completion or near-completion of these algorithms in the run-up to raising $12 million in May. Zebra has partnerships with Intermountain Healthcare and Dell, as well as the latter’s library of 150 million scans. They also count Khosla Ventures and Marc Benioff as investors.
The new algorithm specifically detects calcium buildup in the arteries, the common foundation of later heart attacks. A fatty liver may be an indicator of insulin-resistant diabetes, as well as an underlying cause of chronic liver disease, cirrhosis and liver cancer. An exploratory study published earlier this year also suggested that fatty tissue causes heart attacks.
“Quantifying calcium deposits in multiple different images and trying to generate a 3D volume in our heads that gives overall calcium volume is very difficult and prone to significant error,” Benjamin suggests as an illustration of his argument. He puts it in another, more frank way: Zebra’s algorithms will find things doctors don’t have the time to find. “Our algorithm makes that process accurate and repeatable.”
They say the next three likely rollouts will have to do with chest, abdomen and brain analysis, but could also expand to X-rays and MRIs.
Based in Kibbutz Shefayim, the company was founded in 2014 by Benjamin, Eyal Gura and Eyal Toledano.