SyMRI histogram analysis for diagnosing clinically significant prostate cancer

Area of Science:

• Radiology and Imaging Science
• Oncology Diagnostics
• Medical Physics

Background:

• Quantitative parameters from synthetic magnetic resonance imaging (SyMRI) show promise for diagnosing clinically significant prostate cancer (csPCa).
• Histogram analysis, by evaluating spatial heterogeneity, can enhance diagnostic accuracy for csPCa.

Purpose of the Study:

• To evaluate the diagnostic performance of histogram analysis models using SyMRI relaxation maps for csPCa detection.
• To compare histogram analysis models with mean-value-based models for csPCa differentiation.

Main Methods:

• Prospective enrollment of 124 men with suspected csPCa.
• Analysis of 224 ROIs including csPCa, insignificant PCa, peripheral zone lesions, and benign prostatic hyperplasia.
• Construction of histogram analysis models using SyMRI, diffusion-weighted imaging (DWI), apparent diffusion coefficient (ADC), and combined modalities; comparison with mean-value models.

Main Results:

• Histogram analysis models significantly outperformed mean-value models in both training and validation groups.
• SyMRI-based histogram models showed diagnostic effectiveness comparable to DWI and ADC models.
• The combined SyMRI, DWI, and ADC histogram model achieved the highest AUC values in the peripheral zone (0.898) and transition zone (0.944), significantly outperforming PI-RADS in the TZ.

Conclusions:

• Histogram analysis of SyMRI relaxation maps is a valuable tool for differentiating csPCa from clinically insignificant disease (CIS).
• Combining SyMRI with DWI and ADC in histogram analysis models substantially improves diagnostic accuracy for csPCa.