[Clinical study of artificial intelligence-guided image fusion assisted transperineal prostate biopsy].

Objective: To compare the diagnostic efficacy of AI-guided mpMRI-TRUS fusion assisted transperineal systematic biopsy, targeted biopsy and combined biopsy in the diagnosis of PCa, and to evaluate the clinical application value of combined biopsy.

Methods: From April 2022, the general personal information and clinical data of patients with suspicious prostate lesions (PI-RADS≥3) detected by 3.0T mpMRI were collected, then underwent AI-guided mpMRI-TRUS fusion-assisted transperineal prostate biopsy. The data included age, PSA level, PV, PSAD, PI-RADS score, Gleason score of biopsy tissue, etc. The mpMRI image data were imported into the real-time fusion imaging system before biopsy. After image fusion, the suspected PCa lesion was taken as the target, 2 to 3 cores of targeted biopsy were first performed, then 12 cores of systematic biopsy were continued. The results of targeted biopsy + systematic biopsy were defined as the results of combined biopsy. The detection rate of PCa, CsPCa and pathological Gleason score were compared among different biopsy methods, and the diagnostic efficacy in different PI-RADS score groups was further evaluated.

Results: A total of 118 PCa cases were detected in 220 patients enrolled in this study. The PCa detection rates of systematic biopsy and targeted biopsy were 40.45% and 43.64%, the result reveals no statistical significance (P=0.562). The PCa detection rate of combined biopsy was 53.64%, higher than single biopsy method and the differences were statistically significant (P<0.05). The detection rates of CsPCa in systematic biopsy and targeted biopsy were 28.18% and 37.27% which reveals significant statistical difference (P=0.042). The CsPCa detection rate of combined biopsy was 41.82%, higher than single biopsy method, the difference was statistically significant compared with systematic biopsy (P=0.003), but was not compared with targeted biopsy (P=0.330). In PI-RADS score 3 group, the PCa detection rate of systematic biopsy and targeted biopsy was 39.29% and 21.43%, which reveals no statistical significance (P=0.146). The PCa detection rate of combined biopsy was 50%, higher than single biopsy method, the difference was statistically significant compared with targeted biopsy (P=0.026), but was not compared with systematic biopsy (P=0.420). In PI-RADS 4 ~5 group, the PCa detection rate of systematic biopsy and targeted biopsy was 40.10%, and 46.88% which reveals no statistical significance (P=0.181). The PCa detection rate of combined biopsy was 54.17%, higher than single biopsy method, the difference was statistically significant compared with systematic biopsy (P=0.006), but was not compared with targeted biopsy (P=0.153). Among PCa patients detected by both systematic and targeted biopsy, 39 had concordant pathologic Gleason scores, 13 had escalating pathologic Gleason scores for systematic biopsy, and 18 had escalating pathologic Gleason scores for targeted biopsy.

Conclusion: Compared with systematic biopsy, AI-guided mpMRI-TRUS image fusion assisted transperineal targeted prostate biopsy has a higher detection rate of CsPCa and is probably closer to the true pathological Gleason score. Compared with single biopsy, combined biopsy has higher diagnostic efficiency for PCa, which can be used as one of the options of prostate biopsy in clinical practice.