Russian scientists from the Research Center in the field of Artificial Intelligence of Innopolis University, together with the A.S. Loginov Moscow Clinical Scientific Center, have developed a neural network for diagnosing cancer of the gastrointestinal tract. This innovative system uses AI to analyze biopsies and digital images of tissues, which allows it to automatically detect signs of malignant tumors and distinguish them from non-tumor changes.
The neural network is trained on the basis of 1.6 million medical images from 1.8 thousand samples of neoplasms. The system uses a complex ensemble of neural networks to achieve high accuracy in the segmentation and classification of tumors in ultra-high resolution images. This significantly reduces the number of medical errors and speeds up the process of analyzing biopsies of the stomach and colon.
Nikolai Karnaukhov, Head of the Pathomorphological Laboratory of the MCSC named after A.S. Loginov, noted that the system can form an integrated conclusion about the likelihood of pathology in the histological material.
Ramil Kuleev, Director of the Research Center in the field of Artificial Intelligence of Innopolis University, emphasized that the project, supported by the Analytical Center under the Government of the Russian Federation and implemented within the framework of the federal project "Artificial Intelligence", led to the creation of one of the largest and highest quality datasets in this area.
Read also:
Yandex Created a Neural Network to Detect Rare Pathology in Early Pregnancy
Faster and More Accurate: Doctors in Moscow Diagnose with Artificial Intelligence Assessment
