METHOD OF USING ARTIFICIAL INTELLIGENCE TOOLS FOR INTERPRETING RESULTS OBTAINED BY HIGH-RESOLUTION COMPUTED TOMOGRAPHY (HRCT) FOR DIAGNOSING HUMAN INTERSTITIAL LUNG DISEASES (ILD) Russian patent published in 2024 - IPC G16H30/00 A61B8/13 G06T7/10 

FIELD: data processing.

SUBSTANCE: disclosed group of inventions relates to a method and a system for intelligent determination of the presence of a pathognomonic sign of a fibrosing process of pulmonary tissue, namely traction bronchial and bronchiectasis and adjacent areas of fibrosis, by means of analysis of data of high-resolution computed tomography (HRCT) of lungs using artificial intelligence (AI). Disclosed is a method implemented by the system for determining a pathognomonic sign of a fibrosing process of pulmonary tissue, namely traction bronchiectasis and bronchioloectasis and adjacent areas of fibrosis, by means of analysis of high-resolution computed tomography (HRCT) of lungs, executed by means of at least one processor and comprising stages, at which: obtaining pulmonary HRCT examinations in DICOM format; classifying the presence of a pathognomonic sign of a fibrosing process in the obtained studies using a convolutional neural network (CNN); in the absence of a pathognomonic sign, the result is the absence of CT signs of a fibrosing process; in the presence of a pathognomonic sign, it is localized in the obtained studies—segmentation; segmentation is carried out for each obtained HRCT study using the CNN, trained on an array of images of HRCT studies with a section thickness of not more than 1.5 mm of the lungs of patients with a pathological fibrosing process in the pulmonary tissue, wherein fibrosis areas are selected to obtain a pulmonary tissue mask, in which each pixel corresponds to a pixel on the initial examination, where each pixel is assigned value of 1 if it relates to an area with a pathognomonic sign of a fibrous process, namely traction bronchial and bronchiectasis and adjacent areas of fibrosis, or value of 0 otherwise; after segmentation, the volume of pulmonary tissue damage is calculated by a fibrosing process based on the obtained segmentation mask.

EFFECT: group of inventions provides higher accuracy of determining the presence and prevalence of a pathological fibrosing process in pulmonary tissue, namely traction bronchiectasis and bronchioloectasis and adjacent areas of fibrosis, on studies of HRCT of lungs of patients using artificial intelligence (AI).

2 cl, 7 dwg, 5 tbl