Image recognition is one of the main directions in artificial intelligence theory. Providing a universal solution for the recognition of any given image by a computer is virtually impossible. However, the characteristic features of each image indicate that it follows certain regular patterns. In this study, medical diagnosis is considered as a three-stage process, and closed sub-objects are examined based on complex image representations, which corresponds to the earliest stage of diagnosis. In modern times, medical equipment has become an integral part of medical diagnostics. The research was conducted on 238 ultrasound images obtained using the Toshiba-SAL-38B ultrasound imaging system. A preprocessing package is proposed for the initial processing of the images, which enables tasks such as boundary detection, segmentation, noise removal, and filtering. For image recognition, mathematical morphology methods and a classifier are used. In the final stage, closed or semi-closed contours are detected, and a predictor concept is introduced for their evaluation. Three new indicators characterizing the predictor are proposed: the area enclosed by the contour, the centroid of the area, the color palette of the area. A software package has been developed for image recognition. If a predictor is detected, its indicators are measured and logged. The monitoring frequency is determined by the physician (pp.18-25).