Up to a single cell
AI recognizes individual cancer cells in the images and makes accurate maps
Svetlana Maslova, Hi-tech+
Detailed digital cancer maps allow the doctor to better predict the results of treatment of patients. The artificial intelligence algorithm is much more effective than manual evaluation by a specialist who simply does not have time for such a detailed study.
Scientists from the University of Texas Southwestern Medical Center have developed an artificial intelligence method called ConvPath, which identifies tumor cancer cells in digital images. AI classifies different cell types and evaluates their spatial distribution.
The latter factor is very important for determining the characteristics of tumor growth and development. Also, the spatial distribution indicates the relationship of the cancer focus with the microenvironment and the immune response of the body.
Manual identification of these cells in the image is an extremely time—consuming process for a doctor, so usually only the most representative areas are evaluated. This approach assumes that some important details may be overlooked.
The AI does all the work automatically and provides the doctor with a "map" of cancer for diagnosis.
In the image, this is the middle layer. The upper slice shows a snapshot of the tumor, and the lower one shows a schematic distribution of various cell types.
The use of AI improves the speed and accuracy of diagnosis, allows you to choose the most effective treatment plan. For example, to study the potential of immunotherapy for a specific patient. So far, the authors do not report when the method is planned to be introduced into clinical practice. The use of ConvPath will reduce the burden on specialists and increase the accuracy of diagnosis and treatment.
Article by Wang et al. ConvPath: A software tool for lung adenocarcinoma digital pathological image analysis aided by a convolutional neural network published in the journal EBioMedicine – VM.
Portal "Eternal youth" http://vechnayamolodost.ru