It was named after experts of the digital microscopic analysis laboratory of Moscow State Medical University. Sechenov, together with scientists from Yaroslav the Wise Novgorod State University, taught artificial intelligence to identify blood and lymphatic vessels with tumor cells in histological sections of lung cancer patients. socialbites.ca was told about this at Sechenov University.
This will help minimize the risk of errors when determining the nature of tumor metastasis and the course of the disease, as well as correctly choosing adequate treatment.
Lung cancer is one of the leading causes of cancer deaths today. Doctors detect more than 2 million new cases every year. One of the unfavorable prognostic factors for lung cancer is the presence of vascular invasion, that is, a process in which tumor cells extend beyond the primary focus, grow along the walls of lymphatic and blood vessels and begin to spread throughout the body.
“The presence of vascular invasion in lung cancer increases the risk of distant metastasis and negatively affects overall survival. If such areas are found in histological sections, additional treatment methods such as radiation therapy may be required after surgery. But finding such areas is not an easy task,” Anna Timakova, a young researcher at the Laboratory for Digital Microscopic Analysis, explained to socialbites.ca.
The scientists marked scans of nearly 200 microslides and trained the model to find vessels in scan images of histological slides. The sensitivity of the method is already 75-80%.
“With the help of our development, it will be possible to detect even vessels in whose wall there are already tumor cells, but which have not grown into the lumen in the examined section. There is still no consensus in the professional community whether such suspicious areas should be taken into account when determining further treatment tactics. However, using machine learning methods Automating their search will give experts the opportunity to evaluate the true prognostic significance of such controversial areas,” noted laboratory head Alexey Fayzullin.
By the end of this year, scientists will submit data on the method’s specificity and sensitivity as part of an internal dataset. Next year, the model will be tested on external data sets and real patients.
Russian scientists before discovered It is the most important anti-cancer mechanism of the uterine tissue.