Photo: Ilya Pitalev / RIA Novosti
Experts from Lomonosov Moscow state University, together with colleagues from the UK, Spain and Germany have developed a computer vision system that can diagnose cancer. This was reported TASS the press service of the Russian University.
When training their neural network experts used photographs of 17 thousand slices of tumors and healthy tissues obtained professional histopathology. As a result, they have ensured that the system has learned to recognize the characteristics of 28 types of cancer cells, and highlight the tumors on the pictures of organs and tissue sections. While in some cases the accuracy of its work was superior to the results that can be obtained using traditional methods of histopathology.
“This method allows us to find not only a large-scale genetic change occurring at the level of chromosomes, and single substitutions of nucleotides in the genes-the drivers of the tumors, including portions PTEN, BRAF and TP53. Interestingly, in the case of the gene BRAF, he showed the best prediction results for thyroid cancers than traditional histopathological approach, which suggests the possibility to further improve the conventional classification,” explained MSU student Artyom Shmatko, who worked on the creation of the neural network.
In addition, machine vision has shown good results in predicting the survival of patients, also surpassing the rate of conventional methods. In the future, scientists hope to implement their proposed approach in clinical practice to help physicians more quickly diagnose cancer and to find more effective treatments.
The increase in the number of domestic developments and patents for inventions is provided by the national project “Science”, which should be implemented in Russia to 2024. Only its implementation is planned to more than 625 billion rubles. He also suggests the creation of 15 scientific and educational centers of the world level, commissioning research facilities class “magicians” and the formation of an integrated system of training scientific and scientific-pedagogical personnel.
Video, photo All from Russia.