Zebra signs disease detection deal with Google

Elad Benjamin Photo: Eyal Izhar

The Israeli company's algorithms can detect diseases before radiologists notice them.

Zebra Medical Vision, which has developed medical image analysis capabilities to aid in diagnoses by doctors and before a doctor is summoned, has announced the signing of a cooperation agreement with Google. Under the agreement, Google will offer Zebra Medical's service as part of its service for storing medical information on the cloud. Every medical scan saved on Google's cloud will automatically undergo the scans offered by Zebra Medical. 

Zebra Medical, founded in 2014 by chairman Eyal Gura, CTO Eyal Toledano and CEO Elad Benjamin, is building itself as a one-stop shop for automatic analysis of medical scans, while most of its competitors specialize in scans for detecting specific diseases or scans of a specific part of the body.

This positioning enabled Zebra Medical to launch its $1 per scan initiative last week, in which it allows any hospital to scan its medical images using all the relevant algorithms developed by Zebra Medical at $1 per scan, regardless of the number of analyses. For example, a hospital can load onto the company's software (or onto Google's cloud) a chest image, where it will automatically undergo tests to rule out all the medical problems that can be detected by Zebra Medical's algorithms in the imaged area.

Through the cooperative venture with Google, these capabilities will be available to very hospital in the world, without the hospital installing Zebra Medical's system or signing a long-term contract with the company (something that is common in the software industry, but much less common in the medical industry).

In addition to the $1 model for scanning on the cloud, Zebra Medical's product is also installed as an independent system at 50 hospitals.

Gura told "Globes" last week, "We're launching a model of equal pricing for everyone that will enable these scans to reach not only the leading hospitals, but also any hospital in the world, including in places and countries where few radiologists are available. In pricing of this type, it may take us more time to reach high sales, but we believe that later on, the total sales volume will make up for it several times over. Thanks to this approach, we have already signed agreements in India and Africa." This method also maximizes the company's information, enabling it to make its algorithms even more effective.

The company is not disclosing its current sales volume, but Gura says that Zebra Medical has made a million scans to date (including pilot scans in the framework of clinical trials).

In addition to detecting diseases indicated by the patient's clinical symptoms, the $1 per scan plan model enables hospitals to detect in advance diseases before any symptoms appear. For example, in a trial conducted by the company, it discovered that it can detect miniature compression fractures indicating the beginning of osteoporosis before fractures causing pain are felt, and sometimes even before radiologists can see the fractures when they look at the image without guidance from the software. In any image in which bones appear (even if the image was taken for other purposes), osteoporosis can be detected or ruled out, and preventative treatment can begin before the disease gets worse.

This preventative method can encourage the use of drugs from an earlier stage, and the drug companies are therefore encouraging installation of the system. On the other hand, it is likely to save health systems large amounts currently spent on supportive treatment after the disease worsens. This method, which is currently being used for osteoporosis, can also be adapted to other images later in order to prevent heart attacks, for example.

Zebra Medical was founded when Eyal Gura was hospitalized following a diving accident with no clear diagnosis. He was then chairman of image processing company PicScout (eventually sold to Getty Images for $20 million after raising only $1 million). While waiting for a radiologist, Gura thought about image processing apps for medicine. "The terms big data and deep learning were not widely known then," he says now.

The company started before the big breakthrough in the field. This enabled Zebra Medical to obtain the information needed to build its algorithm. "We had 10 million images; you can't do with less. We started with CT scans, went on to mammography, and now we're analyzing X-rays, too. We'll soon start with MRI and ultrasound," Gura predicts. Diseases that the company's algorithms can detect include osteoporosis, fatty liver, arterial aneurism, emphysema, and arterial occlusion.

Gura says, "In general, image processing systems have dramatically improved their accuracy. While Google's tagging images systems had a 16% error rate as recently as 2012, today they are better than human beings."

Radiologists are your customers in the health system. Are they trying to block you?

Gura: "The demand for radiologists today is so high that they are only glad that we're making their work more efficient and preventing them from making potentially fatal errors. We want to make each radiologist 10 times more efficient, and even then, all the radiologists in the world won't be enough. Furthermore, we're also making things easier for them outside of diagnosis, for example by adding a component to the system that writes reports for them." The final word in every scan still belongs to the radiologist.

Zebra Medical sometimes sells directly to hospitals, and it sometimes constitutes an element in the system of storing images at hospitals (PACS systems). "Israel is a power in the PAC field. All the talent in the sector is here, concentrated mainly in Haifa in companies such as Royal Philips, IBM, GE, and in Algotec Systems in Ra'anana, which was acquired by Eastman Kodak," Gura says. 

"Our relations with the PACS companies are excellent, because everyone wants to show that they have something in artificial intelligence (AI)," Gura explains.

Besides Zebra Medical, Gura is an investor in two other medical companies. One is Medisafe, which recently raised $8 million for its activity in detecting errors in giving drugs (using big data and deep learning methods, of course). Medisafe also entered the field relatively early, and is now considered a leading company in it. "The company already has several million dollars in sales and 4.5 million regular users, from whom it also produces information of great interest to the drug companies," Gura states.

KolGene, a younger company, is trying to solve a logistics problem in genetic testing - both in preliminary tests and tests during pregnancy and in the hot field of genetic specification of malignant tumors for the purpose of adapting drugs to the type of tumor.

Gura: Up to 60,000 types of genetic tests have been added in recent years. The doctors haven't yet gotten used to the tests they already had. We enable clinical personnel to find the right laboratory for every test, and we're a trading floor for these laboratories - the doctor asks for a price quote, gets a quote from 52 laboratories, and can then monitor the status of the order. Tests are currently getting lost or forgotten, and a lot of time is spent trying to understand where the test is and whether there's already an answer from it."

Published by Globes [online], Israel Business News - www.globes-online.com - on November 8, 2017

© Copyright of Globes Publisher Itonut (1983) Ltd. 2017

Elad Benjamin Photo: Eyal Izhar
Elad Benjamin Photo: Eyal Izhar
Twitter Facebook Linkedin RSS Newsletters גלובס Israel Business Conference 2018