Artificial intelligence detects outbreaks of lung cancer in less than 20 seconds

Artificial intelligence detects outbreaks of lung cancer in less than 20 seconds

Cancerous tumors in the lungs

Moscow: Taha Abdel Wahed

Russian scientists from the Polytechnic University, in cooperation with the doctors of the Scientific Center for Cancer Treatment in St. Petersburg, have developed an intelligent system that can diagnose the presence of cancerous tumors in the lungs in just 20 seconds . "The artificial intelligence program designed for this purpose can be installed on any computer and within 20 seconds, it will transfer the patient's tomography to the patient, then give a result," said the researcher. Polytechnic University of St. Petersburg in a statement posted on its official website. Minute on condition ". Scientists have called their invention Asimov, a tribute to the memory of the American science fiction writer and doctor Isaac Asimov, who put the "three robotic laws" in his novel "Evasion".
The university said that "closed" tests on the new innovation had taken place by the end of 2018. A cross-sectional imaging of 60 patients had been taken. The results were positive. The new program identified tumor foci in the lungs, even small foci (2 mm). "At first we developed an algorithm to look for 6mm foci, because the doctors themselves adhere to this tactic," said Lev Utkin, project manager and head of the Neural Network Research Lab. Artificial intelligence. "But the program has been so well taught that is able to detect new pathogens of less than 6 mm."
"It was necessary before the system started to work," said Anna Mildo, co-author of the project, head of the department of radiology. "It was necessary to learn about tumors and distinguish malignant tumors from benign tumors." In CT, you can see a lot of things. The basic task is to teach the system to identify what each represents, bringing the system closer to the physician's logic and using a clinical and radiological classification. We try to teach the system not only the detection of tumors, but also the distinction between other diseases similar to cancer. The new program has trained more than 1,000 CT images. The database currently contains images of more than 250 patients treated with the new program and is expected to quadruple by mid-2019.
In addition to the above, the program develops with each new image processed and assisted by the giant computer called Polytechnic Institute.
, Thanks to a neural network-based functionality that can take into account new data and any task resolved to speed up the process in the future. It is planned to send the scanner from a patient to this supercomputer via the intranet, after the training, the program will be able to complete the process of detecting tumors in the lungs in one or two seconds.
After completing the "closed tests", he plans larger tests and then inserts the new program on computers at the Cancer Center. Scientists plan to develop the "Asimov" program and to cooperate with other medical institutions in this field. The spotlight says that the program hopes to learn ultrasound and radiological images from other internal organs, and that the central computer processes all the data. That ensures great speed in achieving results, and sent to the doctor through internal networks.

Moscow

Moscow