Cognitive Versus Software-based Fusion Targeted Biopsy for the Diagnosis of Clinically Significant Prostate Cancer: A Multicenter, Randomized, Noninferiority Trial (IMAGINATION)

Background Magnetic resonance imaging–targeted prostate biopsy relies heavily on costly software-fusion platforms, limiting global accessibility. Objective We aimed to determine whether cognitive fusion (mental registration) is noninferior to software-based fusion targeted biopsy for detecting clinically significant prostate cancer (csPCa). Design, setting, and participants This multicenter, randomized, blinded, noninferiority trial enrolled 648 predominantly biopsy-naive men (prostate-specific antigen 4–20 ng/ml; Prostate Imaging Reporting and Data System ≥3) at nine centers in China. Intervention Participants were randomly assigned 1:1 to cognitive fusion or software-based fusion targeted biopsy, followed by systematic biopsy. Outcome measurements and statistical analysis The primary outcome was csPCa (ISUP grade group ≥2) detection on targeted biopsy, evaluated using generalized estimating equations with a prespecified −10 percentage point noninferiority margin. Results and limitations In the intention-to-treat population (327 cognitive and 321 software), targeted biopsy detected csPCa in 32% and 34% of the patients, respectively. The adjusted risk difference was –1.02 percentage points (90% confidence interval [CI], –5.6 to 3.6), establishing noninferiority. Combined targeted and systematic biopsy detected csPCa in 35–39% of the patients, respectively (difference, –1.50% [90% CI, –6.4 to 3.4]). A significant body mass index (BMI) interaction was observed ( p < 0.001): cognitive fusion performed best in men with BMI <24 kg/m 2 , whereas software-based fusion improved detection in men with BMI ≥24 kg/m 2 . Adverse event rates were comparable (15% vs 18%). Limitations include conducting the trial predominantly in Asian populations, necessitating cautious geographic extrapolation of specific BMI cutoffs. Conclusions Cognitive fusion is diagnostically noninferior to software-based fusion within a prespecified –10% margin overall. However, a significant BMI interaction raises serious concerns regarding the diagnostic feasibility of cognitive fusion in obese populations, where software-based fusion is strictly superior. Future prospective trials in both Asian and Western populations must evaluate the BMI effect as a primary end point. Trial registration: ClinicalTrials.gov NCT04271527.