COMPARATIVE ANALYSIS OF THE RESULTS OF DIGITAL MAMMOGRAPHY DATA

EVALUATION BASED ON ARTIFICIAL INTELLIGENCE "PLATFORM THIRD OPINION"

AND RADIOLOGISTS

Moiseev D.A1,2, Yusufov A.A2, Maksimov D.A3, Fomina E.E.4

 

1 - White Rose Medical Center.

2 - Tver State Medical University of the Ministry of Health of Russia.

3 - Tver Regional Clinical Oncologic Dispensary

4 - Tver State Technical University. Tver, Russia.

P

urpose. To conduct a comparative analysis of the evaluation results of mammographic images by radiologists and artificial intelligence system. To evaluate the sensitivity of the AI system to detect breast cancer in the group with verified diagnosis and to evaluate the false positive rate in the study of the control group.

Materials and Methods. The results of a retrospective cohort study of 2 groups consisting of 93 patients (group №1 and group №2) screened at ANO "White Rose Medical Center" in the city of Tver are presented. Group №1 (main): patients with a verified diagnosis of malignant breast neoplasm according to histologic examination. Group No. 2 (control): patients who underwent routine examination in the center from 13.11.2023 to 18.11.2023 (random sampling of 93 patients for the selected period of time).

Results. Quantitative sensitivity analysis of the artificial intelligence platform was performed. The value of the sensitivity statistic ranged from 84.9% to 96.8% depending on the given category and score. An intraclass correlation coefficient was calculated to analyze the consistency between the results of mammographic image evaluation by the AI platform and the radiologists' physicians.

Discussion. A retrospective comparative study of data analysis of digital mammograms by radiologists and the Third Opinion AI platform showed good sensitivity and evaluation results of AI actions in analyzing digital mammograms. The algorithm of digital mammogram analysis design is clearly shown on clinical examples.

The information report of AI in decoding digital mammograms will be useful for radiologists, especially beginners, where lack of experience will be leveled.

Conclusion. The feasibility of using an AI system both for analytical support of radiologists in mammographic examinations and in mammography in general has been demonstrated.

 

Keywords: breast cancer, screening, BI-RADS, Breast Imaging Reporting and Data System, artificial intelligence, mammography.

 


Corresponding author: Moiseev D.A., e-mail: Этот e-mail адрес защищен от спам-ботов, для его просмотра у Вас должен быть включен Javascript

For citation: Moiseev D.A, Yusufov A.A, Maksimov D.A, Fomina E.E. Comparative analysis of the results of digital mammography data evaluation based on artificial intelligence "platform third opinion" and radiologists. REJR 2024; 14(4):41-56. DOI: 10.21569/2222-7415-2024-14-4-41-56.

Received:        10.04.24 Accepted:       22.10.24