Predictive modelling for prostate cancer aggressiveness using non-invasive MRI techniques

Area of Science:

• Radiology
• Oncology
• Medical Imaging

Background:

• Magnetic Resonance Imaging (MRI) is vital for prostate cancer (Pca) diagnosis.
• Improving diagnostic accuracy for Pca grading is crucial for treatment decisions.

Purpose of the Study:

• To develop and validate predictive models for distinguishing Pca Grade Groups (GGs).
• Specifically, to differentiate GG2 from GGs3-5 and GG2 from GG3 using MRI data.

Main Methods:

• Retrospective analysis of Double Inversion Recovery MRI (DIR-MRI) and Multiparametric MRI (mpMRI) from 53 patients.
• Calculation of Lesion-to-Normal Ratios (LNRs) using Regions of Interest (ROIs).
• Development of predictive models integrating MRI variables and patient characteristics using the R statistical package CARRoT.

Main Results:

• The developed predictive models demonstrated high diagnostic performance.
• Achieved an Area Under the Receiver Operating Characteristic Curve (AUROC) of 0.86 and 0.91 after 1000 cross-validations.

Conclusions:

• Explainable and rigorously cross-validated models were developed to differentiate Pca aggressiveness.
• These models utilize T2 LNR and axial T2 tumor dimensions, offering greater clinical applicability than existing methods.
• The models provide a robust tool for clinicians to improve Pca diagnosis and clinical decision-making.