BACKGROUND: The global amount of investment in companies developing artificial intelligence (AI)-based software technologies for medical diagnostics reached $80 million in 2016, rose to $152 million in 2017, and is expected to continue growing. While software manufacturing companies should comply with existing clinical, bioethical, legal, and methodological frameworks and standards, there is a lack of uniform national and international standards and protocols for testing and monitoring AI-based software. AIM: This objective of this study is to develop a universal methodology for testing and monitoring AI-based software for medical diagnostics, with the aim of improving its quality and implementing its integration into practical healthcare. MATERIALS AND METHODS: The research process involved an analytical phase in which a literature review was conducted on the PubMed and eLibrary databases. The practical stage included the approbation of the developed methodology within the framework of an experiment focused on the use of innovative technologies in the field of computer vision to analyze medical images and further application in the health care system of the city of Moscow. RESULTS: A methodology for testing and monitoring AI-based software for medical diagnostics has been developed, aimed at improving its quality and introducing it into practical healthcare. The methodology consists of seven stages: self-testing, functional testing, calibration testing, technological monitoring, clinical monitoring, feedback, and refinement. CONCLUSION: Distinctive features of the methodology include its cyclical stages of monitoring and software development, leading to continuous improvement of its quality, the presence of detailed requirements for the results of the software work, and the participation of doctors in software evaluation. The methodology will allow software developers to achieve significant outcomes and demonstrate achievements across various areas. It also empowers users to make informed and confident choices among software options that have passed an independent and comprehensive quality check.

Background: The implementation of radiological artificial intelligence (AI) solutions remains challenging due to limitations in existing testing methodologies. This study assesses the efficacy of a comprehensive methodology for performance testing and monitoring of commercial-grade mammographic AI models. Methods: We utilized a combination of retrospective and prospective multicenter approaches to evaluate a neural network based on the Faster R-CNN architecture with a ResNet-50 backbone, trained on a dataset of 3641 mammograms. The methodology encompassed functional and calibration testing, coupled with routine technical and clinical monitoring. Feedback from testers and radiologists was relayed to the developers, who made updates to the AI model. The test dataset comprised 112 medical organizations, representing 10 manufacturers of mammography equipment and encompassing 593,365 studies. The evaluation metrics included the area under the curve (AUC), accuracy, sensitivity, specificity, technical defects, and clinical assessment scores. Results: The results demonstrated significant enhancement in the AI model’s performance through collaborative efforts among developers, testers, and radiologists. Notable improvements included functionality, diagnostic accuracy, and technical stability. Specifically, the AUC rose by 24.7% (from 0.73 to 0.91), the accuracy improved by 15.6% (from 0.77 to 0.89), sensitivity grew by 37.1% (from 0.62 to 0.85), and specificity increased by 10.7% (from 0.84 to 0.93). The average proportion of technical defects declined from 9.0% to 1.0%, while the clinical assessment score improved from 63.4 to 72.0. Following 2 years and 9 months of testing, the AI solution was integrated into the compulsory health insurance system. Conclusions: The multi-stage, lifecycle-based testing methodology demonstrated substantial potential in software enhancement and integration into clinical practice. Key elements of this methodology include robust functional and diagnostic requirements, continuous testing and updates, systematic feedback collection from testers and radiologists, and prospective monitoring.

Aim. To develop an artificial intelligence-based system for the diagnosis of focal liver lesions aimed at supporting clinical decision-making in surgical hepatology.Materials and methods. An artificial intelligence-based technological service was developed for the automatic segmentation and classification of contrast-enhanced computed tomography (CT) images of four types of liver lesions: focal nodular hyperplasia, carcinoma, hemangioma, and simple cyst. The service was trained and tested on datasets comprising 725 CT images using the nnU-Net architecture. Diagnostic performance was evaluated by calculating the AUC-ROC, sensitivity, specificity, and accuracy.Results. The service achieved high performance metrics. The AUC-ROC ranged from 0.847 to 0.928, with a maximum sensitivity of 0.940 for carcinoma and a specificity of 0.900 for focal nodular hyperplasia. Accuracy ranged from 0.883 to 0.922, which demonstrates the algorithm's ability to clearly differentiate between malignant and benign lesions.Conclusion. The machine learning-based service demonstrated high diagnostic performance and shows promise for integration into clinical practice, offering improved detection and classification of liver lesions.

BACKGROUND: The increased number of computed tomography scans during the COVID-19 pandemic has emphasized the task of decreasing radiation exposure of patients, since it is known to be associated with an elevated risk of cancer development. The ALARA (as low as reasonably achievable) principle, proposed by the International Commission on Radiation Protection, should be adhered to in the operation of radiation diagnostics departments, even during the pandemic. AIM: To systematize data on the appropriateness and effectiveness of low-dose computed tomography in the diagnosis of lung lesions in COVID-19. MATERIALS AND METHODS: Relevant national and foreign literature in scientific libraries PubMed and eLIBRARY, using English and Russian queries low-dose computed tomography and COVID-19, published between 2020 and 2022 were analyzed. Publications were evaluated after assessing the relevance to the review topic by title and abstract analysis. The references were further analyzed to identify articles omitted during the search that may meet the inclusion criteria. RESULTS: Published studies summarized the current data on the imaging of COVID-19 lung lesions and the use of computed tomography scans and identified possible options for reducing the effective dose. CONCLUSION: We present techniques to reduce radiation exposure during chest computed tomography and preserve high-quality diagnostic images potentially sufficient for reliable detection of COVID-19 signs. Reducing radiation dose is a valid approach to obtain relevant diagnostic information, preserving opportunities for the introduction of advanced computational analysis technologies in clinical practice.

Objective. To estimate the incidence of repeated mammography in patients with BI-RADS 3. Material and methods. We analyzed the results of mammography in medical organizations of the Moscow Department of Healthcare providing outpatient medical care for the period from October 1, 2020 to December 31, 2020. The inclusion criterion was BI-RADS 3. Data on repeated mammography performed within 2 years were obtained from electronic medical records. There was no control group. Results. We found 346 mammograms and their text protocols for 346 patients. Repeated mammography was performed in 24.9% (86 out of 346) of patients. The need for repeated mammography was indicated in 34.7% (120 of 346) of protocols. Mean period between baseline and repeated mammography was 327 days. BI-RADS 3 was downgraded to BI-RADS 2 in 81.4% (70 of 86) of cases, increased to BI-RADS 4 in 3.5% of cases (3 of 86) and remained the same in 12.8% of cases (11 of 86). Conclusion. Low rate of repeated mammography indicates the need for strategies to improve mammography adherence, including information technology-based strategies, and additional work with population.