UNIM, a Russian portfolio company of the Internet Initiatives Development Fund and a Skolkovo resident, has developed and trained a neural network that helps doctors identify cancer metastasis.
Metastasis is a remote secondary site of the pathological process. If a doctor cannot find the metastasis in the specimen, then the stage of cancer will be determined incorrectly, and a patient will receive insufficient treatment. This can deprive a person of a chance to recover and live.
The neural network reduces the amount of time a doctor spends on determining this characteristic of cancer a hundredfold. The neural network doesn’t “miss” the metastasis contained in a specimen. Using the neural network, a doctor can obtain information on any metastasis-like mass contained in a specimen and its location within 10 seconds. Usually, a doctor spends around 1 hour doing it on his own. An error in determining the presence or absence of metastasis can lead to an incorrect diagnosis and choosing the wrong method and scope of treatment.
The neural network is based on the analysis of more than 60 terabytes of diagnostic data and has been trained by dozens of pathologists. The neural network already runs in test mode in the company's own laboratory. Data of 600 diagnosed patients have been evaluated with its help.
According to the World Health Organisation 9,6 mln people worldwide are estimated to die from cancer in 2018. The estimated total annual economic cost of cancer in 2010 was 1,16 trillion US$.
“Neural networks bring histology to a whole new level. In the next couple of years, we will see a new level of evidence and accuracy of cancer diagnostics. Doctors will have a reliable assistant who will significantly reduce labor costs and increase work efficiency,” explained Alexey Remez (CEO at UNIM).
Reference
UNIM was established in 2014. At the moment, the Digital Pathology© cloud platform, developed by the company, is used by hundreds of doctors in Russia, Asia, Western Europe, and the US. The company develops fundamentally new cancer diagnostic tools. The Internet Initiatives Development Fund, Medme, and Alexey Basov are company's investors. The company is a Skolkovo resident. In 2018, UNIM has launched its own digital pathology laboratory and announced the development of a neural network that automatically determines the characteristics of the tumor growth rate (Ki67 index).
UNIM press service contact details: Olga Karpova, +7 903 019-68-65, karpova@unim.su