Scientists have created a neural network for early detection of pancreatic cancer

Researchers at St. Petersburg Electrotechnical University LETI have developed a neural network capable of detecting pancreatic cancer at an early stage. About this on October 17, "Gazeta.Ru" was reported at the university.

According to the developers, the model is based on deep machine learning algorithms and is designed to analyze CT scans of the abdominal cavity. In the future, it will become part of a medical decision support system that will be able to assess the likelihood of tumor malignancy and the extent of its spread.

"Together with the Vishnevsky National Research Medical Center of Surgery, we have created a neural network model for automated segmentation of pancreatic tissues and detection of lesions. It will help doctors diagnose cancer at an early stage," said Nikolay Staroverov, Associate Professor of the Department of Electronic Devices and Devices at LETI.

The neural network showed an accuracy of 92.55% in detecting pathological changes. Scientists note that further training of the model will reduce the number of missing pathologies during CT, but the final decision on the diagnosis will always remain with the doctor.

Scientific model: neural network will detect early signs of Parkinson's disease by EEG

How will doctors determine the pathology even before the first symptoms appear

Earlier, on August 15, it was reported that Sechenov University scientists have created a web application based on artificial intelligence for the diagnosis of brain tumors. The system analyzes MRI scans, determining the location and nature of the neoplasms. The accuracy of the model, trained on more than 5,000 images, reached 97.1%.

All important news is on the Izvestia channel in the MAX messenger.