Mechanism for evaluating the results of detection and classification of objects in medical images

Abstract

The article proposes a method for evaluating the results of detecting and classifying objects in medical images obtained from computed tomography of human internal organs by the YOLO architecture neural network, containing an algorithm and mathematical models of fuzzy estimation. The developed algorithm and models make it possible to classify objects depending on their location and image projection, automate and reduce the time for diagnosing a disease, move from evaluating two-dimensional images to assembling and evaluating three-dimensional objects, increase the accuracy of evaluating object parameters, reduce the risks of incorrect surgical decisions when planning and conducting operations. The proposed algorithm and models were implemented in a prototype of a medical decision support system in surgery and urology using computer vision technologies as part of software modules for detecting objects and calculating object parameters. The presented method for evaluating the results of detecting and classifying objects in medical images has shown high efficiency.