METHOD FOR AUTOMATED ANALYSIS OF DIGITAL FLUOROGRAPHY IMAGES Russian patent published in 2019 - IPC A61B6/00 G03B42/02 G06T1/40 

FIELD: medicine.

SUBSTANCE: invention relates to digital images processing in medicine and is intended for automated performance of chest radiographs of a patient for the presence of changes or pathologies in the lung area. Method of analyzing digital photofluorographic images for pathology detection comprises steps of using an initial base of X-ray photographs with classification of each region for training three convoluted neuron networks. Loading is performed by means of a client module, after which successive processing of loaded images is performed by means of neural network to identify areas of interest depending on response of each neural network. Obtained processing is transmitted to a module for analysis and integration of results of operation of convolution networks, which processes and superimposes the results of operation of the neural network ensemble on the initial downloaded digital image for detecting pathologies, wherein one of the convolutional neural networks is trained with the possibility of processing and detecting dangerous pathologies of the first kind, and the second and the third one - of pathologies of the first type and non-dangerous pathologies of the second kind. Each network is trained by using different threshold values and segmentation maps.

EFFECT: use of the invention enables to minimize errors of false triggering of the SNS ensemble.

1 cl, 3 dwg