Yapay Zeka Hızlı Bakış & Çıkarımlar
AI InsightReal-time radiologist assistance appears to refine diagnostic yield by increasing csPCa detection while minimizing insignificant findings, particularly in challenging anatomical locations.
📌 3 Maddede Bu Makale
Abstract
Aim: To evaluate the impact of real-time radiologist assistance on the detection rates of clinically significant prostate cancer (csPCa) and the reduction of overdiagnosis in patients with “challenging” lesions (PI-RADS 3 and 4), using a propensity score-matched analysis. Material and Methods: We retrospectively analyzed patients who underwent software-based MR–US fusion biopsy for PI-RADS 3 or 4 lesions. PI-RADS 5 lesions were excluded to better assess operator-dependent targeting. Patients were stratified into two cohorts: the Multidisciplinary Group (MDG), with active real-time radiologist assistance, and the Single-Operator Group (SOG), performed by the urologist alone. To minimize selection bias, 1:3 propensity score matching (PSM) was conducted based on age, PSA level, prostate volume, and index lesion size. Results: Following matching, 98 patients (28 MDG, 70 SOG) were included. Overall cancer detection rate (CDR) was numerically higher in the SOG (67.1% vs. 53.6%), largely driven by a numerically higher detection of insignificant (ISUP Grade Group 1) cancers (41.4% vs. 21.5%). Conversely, csPCa detection (ISUP ≥2) was higher in the MDG (32.1% vs. 25.7%). “First-strike success” (detection in the first core) was 88.9% in the MDG vs. 75.0% in the SOG. Subgroup analyses showed a strong clinical trend toward superior csPCa detection in the MDG for anterior lesions (60.0% vs. 23.3%, p=0.053). Conclusion: In PI-RADS 3–4 lesions, the single-operator approach yields high sensitivity but may carry a risk of overdiagnosis. Real-time radiologist assistance appears to refine diagnostic yield by increasing csPCa detection while minimizing insignificant findings, particularly in challenging anatomical locations.