Diagnostic value of intravoxel incoherent motion-based virtual MR elastography for differentiating prostate cancer from benign prostate lesion.

OBJECTIVE

To evaluate the diagnostic efficacy of virtual magnetic resonance elastography (vMRE) derived from intravoxel incoherent motion (IVIM) imaging in distinguishing prostate cancer (PCa) from benign prostate lesion (BPL).

METHODS

In this prospective study, patients with suspected prostate lesions underwent 3.0-T MRI, including an IVIM-vMRE sequence, prior to biopsy or surgery. Virtual stiffness maps (µdiff) were generated using a custom MATLAB algorithm. Two radiologists, blinded to pathology, performed manual region-of-interest segmentation and PI-RADS v2.1 scoring. The protocol focused on comparing µdiff values between malignant and benign lesions. Diagnostic performance was assessed via receiver operating characteristic (ROC) analysis, and the correlation between µdiff and Gleason scores was evaluated using Spearman's rank correlation. Internal validation was performed via 1,000-sample bootstrapping.

RESULTS

The final cohort comprised 120 patients (70 PCa; 50 BPL). PCa lesions exhibited significantly higher µdiff values compared to BPL (9.25 ± 1.06 kPa vs. 7.27 ± 0.53 kPa, P < 0.001). ROC analysis revealed an optimal µdiff cutoff of 7.90 kPa, which yielded an AUC of 0.931 (95% CI: 0.886-0.976). Furthermore, µdiff values were significantly positively correlated with pathological Gleason scores (r = 0.670, P < 0.001). The combined model of µdiff and PI-RADS v2.1 significantly outperformed PI-RADS v2.1 alone (AUC: 0.941 vs. 0.811, P < 0.001), and it achieved an optimism-corrected AUC of 0.940 in bootstrap validation.

CONCLUSION

IVIM-vMRE provides an objective, quantifiable measure of prostate tissue stiffness that correlates with tumor grade. As a non-invasive adjunct to bpMRI, vMRE significantly improves the diagnostic accuracy of PI-RADS v2.1 and facilitates the stratification of tumor aggressiveness.