Advancing Prostate Cancer Assessment: A Biparametric MRI (T2WI and DWI/ADC)-Based Radiomic Approach to Predict...

Area of Science:

• Radiology and Medical Imaging
• Oncology
• Artificial Intelligence in Medicine

Background:

• Prostate cancer (PCa) management benefits from accurate noninvasive assessment of tumor characteristics.
• Tumor-stroma ratio (TSR) is a histopathological prognostic factor in PCa.
• Biparametric MRI (bpMRI) offers detailed imaging for PCa evaluation.

Purpose of the Study:

• To develop and validate a bpMRI-based radiomics model for noninvasive TSR prediction in PCa.
• To explore the relationship between TSR and lesion distribution in the peripheral zone (PZ) and transition zone (TZ).

Main Methods:

• A multicenter retrospective study of 223 PCa patients.
• Radiomic feature extraction from preoperative bpMRI (T2WI, DWI, ADC maps).
• Machine learning classifiers (Ridge model selected) trained and validated for TSR prediction.

Main Results:

• The Ridge radiomics model achieved AUCs of 0.846 (training), 0.789 (validation), and 0.745 (testing).
• No significant association was found between TSR and zonal localization (p = 0.867).

Conclusions:

• A bpMRI-based radiomics model shows potential for noninvasive TSR estimation in PCa.
• This approach may offer insights into PCa aggressiveness beyond histopathology.
• Further research can explore its clinical utility in PCa management.