Computer-aided detection for prostate cancer detection based on multi-parametric magnetic resonance imaging

Area of Science:

• Medical Imaging
• Oncology
• Computer-Aided Diagnosis

Background:

• Prostate cancer (CaP) is a leading global cancer diagnosis in men.
• Magnetic resonance imaging (MRI) advancements have improved CaP detection, but diagnosis faces challenges like observer variability and lesion complexity.
• Computer-aided detection and diagnosis (CAD) systems offer potential to assist radiologists.

Purpose of the Study:

• To develop and evaluate a novel CAD system for prostate cancer detection.
• To integrate multiple MRI modalities (T2-W-MRI, DCE-MRI, DW-MRI, MRSI) for enhanced diagnostic accuracy.
• To generate probabilistic maps indicating cancer location within the prostate.

Main Methods:

• Development of a CAD system integrating data from T2-W-MRI, DCE-MRI, DW-MRI, and MRSI.
• Utilizing various feature fusion approaches to combine information from different MRI modalities.
• Extensive testing and validation of the proposed CAD system's performance.

Main Results:

• The CAD system successfully integrated multi-modal MRI data.
• Feature fusion strategies were evaluated for their impact on diagnostic performance.
• The system generated probabilistic maps to aid in localizing prostate cancer.

Conclusions:

• The proposed multi-modal MRI-based CAD system shows promise for improving prostate cancer diagnosis.
• Integration of diverse MRI sequences through feature fusion enhances the potential of CAD systems.
• The open-source release of the code and dataset facilitates further research and development.