Transperineal Free-Hand Prostate Fusion Biopsy with AI-Driven Auto-Contouring: First Results of a Prospective Study

TL;DR

A new AI-powered device for prostate fusion biopsies was tested and found to be accurate, efficient, and easy to use, even for less experienced users.

Contribution

A novel AI-driven auto-contouring device for transperineal prostate fusion biopsies is introduced and evaluated in a prospective study.

Findings

The device achieved accurate targeting of MRI lesions with a median volume difference of 1 cc between ultrasound and MRI.

Cancer detection rates were 64% overall and 56% for clinically significant prostate cancer.

Outfield positive cores were found in one third of cases, emphasizing the need for systematic biopsies.

Abstract

The present study explored the role of a novel device for fusion imaging, equipped with AI-driven auto-contouring, to perform prostate fusion biopsies. We found that the device was accurate for targeting MRI lesions, with good detection rates in line with previous findings in the literature. The procedure with UroFusion appeared time-efficient, user-friendly, and well-tolerated, allowing good outcomes with novice users. Fusion imaging was accurate, with a median difference of 1 cc between the volumes estimated at ultrasound and MRI.

Background: prostate fusion biopsies are key in the diagnosis of prostate cancer (PCa); however, the fusion imaging system is not always user-friendly or reliable. The aim of this study was to assess the feasibility, accuracy, and effectiveness of transperineal fusion biopsies performed with a novel fusion imaging device equipped with AI-driven auto-contouring. Methods: data from 148 patients who underwent MRI-targeted and systematic prostate fusion biopsy with UroFusion (Esaote) were prospectively collected. All biopsies were performed in-office, under local anaesthesia. Results: cancer detection rate was 64% overall and 56% for clinically significant PCa (csPCa, ISUP ≥ 2). PCa was detected in 35%, 65% and 84% of lesions scored as PI-RADS 3, 4 and 5, respectively. Outfield positive systematic cores were found in the contralateral lobe in one third of cases. Median device-time to obtain fusion imaging was 5 min and median biopsy duration was 15 min. Median difference in volume estimation between ultrasound and MRI auto-contouring was only 1 cc. Detection rate did not differ between experienced and novice, supervised users. Conclusions: in this initial prospective experience, fusion biopsies performed with UroFusion AI-driven auto-contouring system appeared time-efficient, accurate, well tolerated, and user-friendly, with comparable outcomes between experienced and novice users. Systematic biopsies remain highly recommended given the non-negligible rates of positive outfield cores.

Figures