Comparative study of data analysis results of digital mammography AI-based system "Celsus" and radiologists

The article discusses the possibility of using the AI-based system for the analysis of mammological images Celsus for the detection of malignant neoplasms of the mammary gland during screening mammographic studies. The article presents the results of a retrospective cohort study performed on a group of 49 patients screened at the National Medical and Surgical Center named after N.I. Pirogov and with a verified diagnosis of malignant lesions according to histological examination data. The quantitative characteristics of the analysis of digital mammograms by the Celsus system were obtained. In particular, malignant neoplasms were detected in 92% of cases, and a complete agreement with a group of radiologists specializing in mammography was made in 90% of cases. The results of a comparative clinical analysis of the most difficult cases for diagnosis are presented. The expediency of using the Celsus system for analytical support of radiologists during screening mammographic studies is shown and use-cases of their use are determined.