EVALUATION OF THE RESULTS OF THE ARTIFICIAL INTELLIGENCE USAGE IN THE
ULTRASOUND DIAGNOSTICS OF BREAST TUMORS
Marushchak E.A.1,2, Zubareva E.A.2, Glushkov P.S.1, Fisenko E.P.1
1 - B.V. Petrovsky Russian Research Surgery Center.
2 - N.I. Pirogov Russian National Research Medical University. Moscow, Russia.
|
T |
o evaluate the results of the use of AI in the ultrasound diagnosis of breast tumors.
Materials and methods. In 2024, 91 multiparametric ultrasounds (In-mode, CDI, elastography, microvascularization assessment) were performed in patients with 129 tumors of the breast, including using the Samsung RS85 ultrasound scanner AI software. The age of the patients ranged from 18 to 82 years. Morphological verification of neoplasms of various levels was performed in all patients (fine needle aspiration biopsy, trepan biopsy, vacuum aspiration biopsy, total histological examination with immunohistochemical analysis).
The patients were divided into 2 groups: the coincidence and discrepancy of the opinion of the diagnostician and the AI data. There are 117 neoplasms in the data coincidence group, and 12 neoplasms in the data discrepancy group. The data obtained were statistically processed.
Results. The coincidence of the doctor's ultrasound and morphological verification data was 90% and of the AI was 87%. The sensitivity of the ultrasound performed by the doctor and the AI turned out to be the same and amounted to 82%; the specificity of the ultrasound of the doctor was 92% and of the AI was 88%.
Discussion. The data of the conducted analysis indicate that at the current stage of information technology development, the AI system does not show significant advantages in the stratification of breast tumors over an ultrasound doctor with extensive work experience. Our experience using the AI program has shown its effectiveness for screening nodular tumors of the breast, which may help reduce the dependence of ultrasound results on the doctor's experience. If the node was classified as I-III by BI-RADS by the AI system, then this was highly likely to exclude breast cancer, which was morphologically verified. At the same time, the AI system is prone to over diagnosis, which is due to algorithms that imitate clinical thinking that are still not sufficiently advanced at this stage of its evolution.
Conclusion. An important point when performing ultrasound using AI software is the clear positioning of the survey window on the object to be analyzed, as well as good image quality. The ultrasound scanner AI software can be used in the practice of an ultrasound diagnostics doctor as a decision support system, but not to replace his conclusion. The most promising niche for the use of AI are regions with limited medical resources, where the results of AI will not only reduce the burden on medical staff, but will also contribute to a more targeted analysis of ultrasound data by a clinician.
Keywords: ultrasound diagnostics, mammary gland, artificial intelligence, morphological verification, BI-RADS.
Corresponding author: Marushchak E.A., e-mail: Этот e-mail адрес защищен от спам-ботов, для его просмотра у Вас должен быть включен Javascript
For citation: Marushchak E.A., Zubareva E.A., Glushkov P.S., Fisenko E.P. Evaluation of the results of the artificial intelligence usage in the ultrasound diagnostics of breast tumors. REJR 2025; 15(1):119-129. DOI: 10.21569/2222-7415-2025-15-1-119-129.
Received: 21.12.24 Accepted: 14.02.25