[Artificial intelligence-based fusion prostate biopsy].

Abstract

Introduction: Prostate cancer imaging is constantly evolving. The previously used “blind sampling” technique of prostate biopsy was superseded by developing the ultrasound- guided methods and later assisted by more precise imaging. The current state-of-the-art version relies on multiparametric MRI analysis and fusion planned biopsy. Objective: Our aim was to present, for the first time in Hungary, our experiences gained through artificial intelligence-assisted image analysis and three-dimensional virtual modeling in transperineal prostate biopsy procedures. Method: Between November 2020 and April 2024, at our institute, we performed 227 prostate biopsies following Watson Elementary® artificial intelligence-assisted image analysis and PI-RADS (v. 2.1) classification, as well as three-dimensional digital modeling. The procedures were conducted under anesthesia using the BiopSee® perineal biopsy station. Our results were analyzed retrospectively. Results: During the specified period, 227 biopsies were performed on patients suspected of prostate cancer. The mean age was 66.7 ± 7.8 years, and the average PSA level was 14.3 ng/mL. The average time in the operating room was 26.2 minutes. No infectious complications occurred. Spontaneous hematuria or hematospermia was recorded in 86 cases. A transurethral catheter was inserted in 7 cases; however, 5 (71%) of them were removed within 24 hours. The tumor detection rate was 18% for PI-RADS 3, 72% for PI-RADS 4, and 95.5% for PI-RADS 5. The distribution of clinically significant prostate cancer was 7%, 44%, and 84%, respectively, for these PI-RADS categories. For confirmed prostate cancer cases, 48 radical prostatectomies were performed, and 20 patients are under active surveillance. Discussion: The two emerging problems in relation to prostate biopsies – the detection rate of clinically significant cancers and the risk of infection – encourage medicine to continue to develop. Conclusion: Artificial intelligence-assisted analysis and digital model-based transperineal fusion biopsy offer high patient safety and transparent diagnostic assistance. Transperineal sampling, performed according to our institute’s protocol, has minimized the risk of infection to virtually zero. The standardization of imaging and the technical unification of sampling protocols can support diagnostic and therapy-based decision-making. Orv Hetil. 2025; 166(13): 503–510.