Development and clinical applicability of MRI-based 3D prostate models in the planning of nerve-sparing robot-assisted radical prostatectomy.

ABSTRACT

The interpretation of conventional MRI may be limited by the two-dimensional presentation of the images. To develop patient-specific MRI prostate-based virtual and three-dimensional (3D)-printed models. To assess the association between 3D imaging and the pathological outcome of RARP specimen. To assess the clinical applicability of 3D models to guide nerve-sparing robot-assisted radical prostatectomy (RARP). We created virtual 3D and 3D-printed 3D models of 20 prostate cancer patients retrospectively. A comparison was made between conventional MRI and 3D-reconstructed images. The concordance between tumour lesion location in 3D models and pathology reporting of RARP specimens was assessed. Seven urologists assessed the side-specific extent of nerve-sparing based on (1) conventional MR images, (2) virtual 3D models, and (3) 3D-printed models. Clinically relevant changes in nerve-sparing and the absolute agreement between observers was analyzed using the Chi-square test and intra-class correlation coefficient (ICC). The index lesion was correctly visualized in 19/20 (95%) 3D models and the expected location of extraprostatic extension was correctly visualized in all 3D models. Clinically relevant changes in the planned extent of nerve-sparing between MRI and virtual 3D models and MRI and 3D-printed models were found in 25% and 26%. The ICC of the planned extent of nerve-sparing between urologists was 0.40 (95% CI 0.28-0.55) for conventional MRI, 0.52 (95% CI 0.39-0.66) for virtual 3D models and 0.58 (95% CI 0.45-0.71) for 3D-printed models. 3D models of the MRI prostate to guide RARP could aid urologists in the planning of nerve-sparing surgery as shown by a higher inter-observer agreement.