Comparing and Combining Artificial Intelligence and Spectral/Statistical Approaches for Elevating Prostate Cancer...

Area of Science:

• Radiology
• Medical Imaging
• Oncology

Background:

• Accurate prostate cancer management requires non-invasive, quantitative tumor evaluation.
• Current methods include visual inspection and AI (artificial intelligence) / deep learning (DL) on MRI.
• Spectral/statistical approaches have shown success in evaluating biparametric MRIs for prostate cancer.

Purpose of the Study:

• To assess and improve spectral/statistical methods for prostate cancer evaluation.
• To benchmark these methods against AI-based deep learning approaches.
• To investigate the combined potential of spectral/statistical methods and AI.

Main Methods:

• A zonal-aware self-supervised mesh network (Z-SSMNet) was applied to MRI data from 42 patients.
• Tumor volume and eccentricity were computed using Z-SSMNet output.
• Linear and logistic regression analyzed correlations with ISUP grade and prostate cancer significance (PCsPCa).
• Multi-variate regression combined Z-SSMNet and spectral/statistical outputs.

Main Results:

• Z-SSMNet (AI/DL) showed poorer performance compared to previous spectral/statistical methods.
• Multi-variable regression combining AI (average blob size) and spectral/statistical (SCR) results achieved a higher correlation (R=0.70) than univariate approaches.
• The combined approach significantly enhanced tumor grade prediction accuracy.

Conclusions:

• Spectral/statistical approaches demonstrated robust performance in prostate cancer assessment.
• Combining AI (Z-SSMNet) with spectral/statistical methods significantly improved tumor grade prediction.
• This integrated approach may offer a valuable alternative for current prostate tumor assessment strategies.