Open AccessPERSPECTIVE METHODS FOR CONTOUR ANALYSIS OF RADIOGRAPHIC IMAGES OF MALIGNANTBREAST TUMORS
Author(s) -
М. К. Михайлов,
Е. А. Романычева,
В. В. Севастьянов,
Я. А. Фурман
Publication year - 2018
Publication title -
lučevaâ diagnostika i terapiâ
Language(s) - English
Resource type - Journals
eISSN - 2079-5351
pISSN - 2079-5343
DOI - 10.22328/2079-5343-2018-9-2-40-45
Subject(s) - mammography , radiology , artificial intelligence , computer science , medicine , pattern recognition (psychology) , computer vision , breast cancer , cancer
X-ray mammography is considered one of the primary diagnostic methods for malignant breast tumors. Since the conclusion is mainly based on the visual analysis of analog or digital X-ray images of the breast, the objectivity of the method is highly dependent on the professional experience of the radiologist. Therefore, the automation of the process of analyzing X-ray mammograms is a relevant task. The present study aims to develop a method for the automatic classification of the types of tumors on x-ray mammograms. To this end, the boundaries of breast tissue densities were described analytically using the mathematical apparatus of contour analysis. It has been found that malignant tumors are characterized by rough contours, which enables the determination of the tumor type by calculating the straightness coefficient of its contour. The straightness coefficient values for malignant and benign tumors have been found. Based on a representative sample from the patients with a previously known diagnosis, consistent classification results have been obtained which is an indication of the feasibility of the proposed method.