Artificial intelligence used to identify bacteria quickly and accurately 2017-12-19 11:18:00 / INTERESTING INFORMATION

Scientists are sure: microscopes coupled with artificial intelligence can quickly and accurately diagnose, revealing deadly blood infections that kill up to 40% of patients. often blood infections cause rod-like bacteria, including Escherichia coli, the bacteria of Staphylococcus and Streptococcus species.

Collegues of the Bet-Izreyel Medical Center used an automatic microscope designed to collect high-resolution images and a convolutional neural network,  reports.The latter is a form of artificial intelligence, constructed on the basis of the visual cortex of mammals for image analysis. The network was taught to recognize the type of bacteria depending on the form and its distribution in the sample. The training of the network was due to the acquaintance of more than 100,000 images of blood samples.

In general, the system divided the images of bacteria into three categories: rod-like, spherical and spherical chains or pairs. The accuracy of the separation reached almost 95%. According to scientists, in the future the system will be able to replace entire laboratories, in terms of diagnosis.

    Copyright © AMEA İnformasiya Texnologiyaları İnstitutu, 2017