Accuracy of prostate measurements based on magnetic resonance imaging results using artificial intelligence technologies: retrospective diagnostic research

INTRODUCTION: Magnetic resonance imaging provides accurate and reliable detection of prostate pathologies, as well as staging of prostate cancer. The capabilities of the method are used in conducting targeted biopsy, treatment, and also for assessing metastatic lesions. Currently, more attention is focused on the possibilities of radiomics and artificial intelligence to improve the diagnostic capabilities of magnetic resonance imaging, in particular to improve the accuracy and timing of detection of formations. Artificial intelligence (AI) technologies have a huge number of potential applications in classifying and improving the quality of multiparametric prostate images, segmenting the prostate gland and suspicious lesions, detecting and differentiating clinically insignificant and significant cancers at the 3D level, and classifying lesions according to the Prostate Imaging Reporting and Data System (PI-RADS).

OBJECTIVE: To evaluate the diagnostic accuracy of automated measurement of prostate size and volume using artificial intelligence.

MATERIALS AND METHODS: A retrospective diagnostic research was conducted in accordance with methodology «STARD 2015». An original methodology for testing and monitoring AI services at life cycle stages was used. The accuracy assessment was carried out by binary classification: correct measurement and incorrect measurement. The proportion of correct measurements was assessed, then the accuracy of the morphometric AI service was calculated using the original formula — the ratio of the number of studies with the doctor’s consent to the measurements of the AI service to the total number of studies successfully processed by the AI service, multiplied by 100.

RESULTS: During the second calibration test, the accuracy of determining the vertical size of the prostate gland was 94.68%, the anteroposterior (sagittal) size accuracy was 97.87%, and the frontal (transverse) size accuracy was 96.81%. Overall, the accuracy of the morphometric AI service in measuring prostate gland size from MRI results was 96.45%.

DISCUSSION: The use of artificial intelligence for the analysis of MRI of the pelvic organs particularly of the prostate gland, has been studied by a limited number of authors. In contrast, automated analysis of studies of the chest organs, mammary gland or brain is significantly more common. Objective reasons explaining this imbalance are difficult to find. In fact, we can only limit ourselves to the standard statement about the lack of data for training algorithms — in the required volume and quality. Most researchers focus on the problems of detection, risk assessment and differential diagnosis of focal lesions of the prostate gland. In particular, among 11 AI-based that have received medical device approval in the USA and/or Western Europe, 5 perform the segmentation function, the other 5 focus on the detection of foci (with 2 presenting the results as a heat map and 3 providing reports), and one solution both segments the gland and calculates «several biomarkers» (e.g., PSA density) while also offering a description project of the study results. It should be noted that, in the latter case, Dice coefficient values for the segmentation task have been published, but data on the accuracy of biomarker calculations are not available.

CONCLUSION: The Moscow Experiment, within the framework of which our research was conducted, represents an impartial and objective external validation, moreover, conducted according to a standardized methodology and with «transparent» results. The task of automating the morphometry of routine measurements of the prostate gland has been successfully implemented. The use of artificial intelligence technologies for the analysis of pelvic organs, in particular, the prostate gland, remains a relevant and little-studied area.

1. Apolihin O.I., Komarova V.A., Nikushina A.A., Sivkov A.V. Prostate diseases in the Russian Federation: statistical data for 2008–2017. Healthcare organization in urology, 2019, No. 4, pp. 4–13 (In Russ.). doi: 10.29188/2222-8543-2019-11-2-4-12.

2. Fernandes M.C., Yildirim O., Woo S., Vargas H.A., Hricak H. The role of MRI in prostate cancer: current and future directions // Magma. 2022. Vol. 35, No. 4. Р. 503–521. https://doi.org/10.1007/s10334-022-01006-6.

3. Belue M.J., Turkbey B. Tasks for artificial intelligence in prostate MRI // Eur. Radiol. Exp. 2022. Vol. 6, No. 1. Р. 33. doi:10.1186/s41747-022-00287-9.

4. Bozgo V., Roest C., van Oort I., Yakar D., Huisman H., de Rooij M. Prostate MRI and artificial intelligence during active surveillance: should we jump on the bandwagon? // Eur. Radiol. Published online June 27. 2024. doi: 10.1007/s00330-024-10869-3.

5. Harmon S.A., Tuncer S., Sanford T., Choyke P.L., Türkbey B. Artificial intelligence at the intersection of pathology and radiology in prostate cancer // Diagn. Interv. Radiol. 2019. Vol. 25, No. 3. Р. 183–188. doi: 10.5152/dir.2019.19125.

6. Turkbey B., Haider M.A. Artificial Intelligence for Automated Cancer Detection on Prostate MRI: Opportunities and Ongoing Challenges, From the AJR Special Series on AI Applications // AJR Am. J. Roentgenol. 2022. Vol. 219, Nо. 2. Р. 188–194. doi: 10.2214/AJR.21.26917.

7. Karnaukhov N.S., Maksimenko A.V., Sogomonian M.G. Artificial Intelligence Capabilities in Oncourology. Modern Problems of Science and Education, 2024, No. 4 рр. 1–12. (In Russ.). doi: https://doi.org/10.17513/spno.33556.

8. Morozov A.O., Bazarkin A.K., Vovdenko S.V., Taratkin M.S., Balashova M.S., Enikeev D.V. Application of Artificial Intelligence in Molecular and Genetic Diagnostics of Prostate Cancer. Bulletin of Urology, 2024, Vol. 12, No. 1, pp. 117–130 (In Russ.). https://doi.org/10.21886/2308-6424-2024-12-1-117-130.

9. Popov G.V., Chub A.A., Lerner Yu.V., Tsoi L.V., Dubinina A.V., Varshavsky V.A. Artificial intelligence in the diagnosis of prostate cancer. Archives of Pathology, 2021, Vol. 83, No. 2, pp. 38–45 (In Russ.). https://doi.org/10.17116/patol20218302138.

10. Talyshinsky A.E., Kamyshanskaya I.G., Mishchenko A.V., Guliev B.G., Bakhtiozin R.F. Application of artificial intelligence in the detection and stratification of prostate cancer: a literature review. Bulletin of St. Petersburg University. Medicine, 2023, Vol. 18, No. 2, рр. 150–166 (In Russ.). https://doi.org/10.21638/spbu11.2023.204.

11. Aboyan I.A., Redkin V.A., Nazaruk M.G., Polyakov A.S., Pakus S.M. et al. Artificial intelligence in the diagnosis of prostate cancer using magnetic resonance imaging. A new approach. Oncourology, 2024, Vol. 20, No. 2, pp. 35–43 (In Russ.). https://doi.org/10.17650/1726-9776-2024-20-2-35-43.

12. Kaneko M., Magoulianitis V., Ramacciotti L.S., Raman A., Paralkar D. et al. The Novel Green Learning Artificial Intelligence for Prostate Cancer Imaging: A Balanced Alternative to Deep Learning and Radiomics // The Urologic clinics of North America. 2024. Vol. 51, No. 1. Р. 1–13. https://doi.org/10.1016/j.ucl.2023.08.001.

13. Sunoqrot M.R.S., Saha A., Hosseinzadeh M., Elschot M., Huisman H. Artificial intelligence for prostate MRI: open datasets, available applications, and grand challenges // European radiology experimental. 2022. Vol. 6, No. 1. Р. 35. https://doi.org/10.1186/s41747-022-00288-8.

14. Saha A., Bosma J.S., Twilt J.J., van Ginneken B., Bjartell A. et al. (2024). Artificial intelligence and radiologists in prostate cancer detection on MRI (PI-CAI): an international, paired, non-inferiority, confirmatory study // The Lancet. Oncology. 2024. Vol. 25, No. 7. Р. 879–887. https://doi.org/10.1016/S1470-2045(24)00220-1.

15. Morozov S.P., Vladzymyrskyy A.V., Shulkin I.M., Ledikhova N.V., Arzamasov K.M., Andreychenko A.E., Logunova T.A., Omelyanskaya O.V., Gusev A.V. Feasibility of using artificial intelligence in radiology (first year of Moscow Experiment on Computer Vision). Medical Doctor and Information Technologies. 2022. No. 1. P. 12–29. (In Russ.). https://doi.org/10.25881/18110193_2022_1_12.

16. Morozov S.P., Gavrilov A.V., Arkhipov I.V., Dolotova D.D., Lysenko M.A. et al. The impact of artificial intelligence technologies on the duration of descriptions of the results of computed tomography of patients with COVID-19 in inpatient healthcare. Preventive Medicine, 2022, Vol. 25, No. 1, pp. 14–20 (In Russ.). https://doi.org/10.17116/profmed20222501114.

17. Bobrovskaya T.M., Vasiliev Yu.A., Nikitin N.Yu., Arzamasov K.M. Approaches to the formation of data sets in radiation diagnostics. Doctor and information technology, 2023, No. 4, p. 14 (In Russ.). https://doi.org/10.25881/18110193_2023_4_14.

18. Vasiliev Yu.A., Vladzimirsky A.V., Omelyanskaya O.V. et al. Methodology of testing and monitoring software based on artificial intelligence technologies for medical diagnostics. Digital Diagnostics, 2023, Vol. 4, No. 3, pp. 252–267 (In Russ.). doi: 10.17816/DD321971.

19. Maki J.H., Patel N.U., Ulrich E.J., Dhaouadi J., Jones R.W. Part I: prostate cancer detection, artificial intelligence for prostate cancer and how we measure diagnostic performance: a comprehensive review // Current problems in diagnostic radiology. 2024. Vol. 53, No. 5. Р. 606–613. https://doi.org/10.1067/j.cpradiol.2024.04.002.

20. Belue M.J., Turkbey B. Tasks for artificial intelligence in prostate MRI // European radiology experimental. 2022. Vol. 6, No. 1. Р. 33. https://doi.org/10.1186/s41747-022-00287-9.

21. Jimenez-Pastor A., Lopez-Gonzalez R., Fos-Guarinos B., Garcia-Castro F., Wittenberg M. et al. Automated prostate multi-regional segmentation in magnetic resonance using fully convolutional neural networks // European radiology. 2023. Vol. 33, No. 7. Р. 5087–5096. https://doi.org/10.1007/s00330-023-09410-9.

22. Sanford T.H., Zhang L., Harmon S.A., Sackett J., Yang D. et al. Data Augmentation and Transfer Learning to Improve Generalizability of an Automated Prostate Segmentation Model // AJR. American journal of roentgenology. 2020. Vol. 215, No. 6. Р. 1403–1410. https://doi.org/10.2214/AJR.19.22347.

23. Ushinsky A., Bardis M., Glavis-Bloom J., Uchio E., Chantaduly C. et al. A 3D-2D Hybrid U-Net Convolutional Neural Network Approach to Prostate Organ Segmentation of Multiparametric MRI. // AJR. American Journal of Roentgenology. 2021. Vol. 216, No. 1. Р. 111–116. https://doi.org/10.2214/AJR.19.22168.

24. Wang B., Lei Y., Tian S., Wang T., Liu Y. et al. Deeply supervised 3D fully convolutional networks with group dilated convolution for automatic MRI prostate segmentation // Medical physics. 2019. Vol. 46, No. 4. Р. 1707–1718. https://doi.org/10.1002/mp.13416.

25. Cao R., Mohammadian Bajgiran A., Afshari Mirak S., Shakeri S., Zhong X. et al. Joint Prostate Cancer Detection and Gleason Score Prediction in mp-MRI via FocalNet // IEEE transactions on medical imaging. 2019. Vol. 38, No. 11. Р. 2496–2506. https://doi.org/10.1109/TMI.2019.2901928.

26. Le M.H., Chen J., Wang L., Wang Z., Liu W. et al. Automated diagnosis of prostate cancer in multi-parametric MRI based on multimodal convolutional neural networks // Physics in medicine and biology. 2017. Vol. 62, No. 16. Р. 6497–6514. https://doi.org/10.1088/1361-6560/aa7731.

27. Belue M.J., Harmon S.A., Lay N.S., Daryanani A., Phelps T.E. et al. The Low Rate of Adherence to Checklist for Artificial Intelligence in Medical Imaging Criteria Among Published Prostate MRI Artificial Intelligence Algorithms // Journal of the American College of Radiology: JACR. 2023. Vol. 20, Nо. 2. Р. 134–145. https://doi.org/10.1016/j.jacr.2022.05.022.