A classifier model for prostate cancer diagnosis using CNNs and transfer learning with multi-parametric MRI

Area of Science:

• Oncology
• Medical Imaging
• Artificial Intelligence

Background:

• Prostate cancer (PCa) is a leading cause of male cancer deaths globally, necessitating improved diagnostic tools.
• Magnetic Resonance Imaging (MRI) offers high resolution for PCa detection, but efficient analysis remains challenging.
• Computer-aided diagnostic (CAD) and deep learning (DL) methods are increasingly vital for enhancing diagnostic accuracy and reducing variability.

Purpose of the Study:

• To develop and evaluate a novel deep learning approach for accurate prostate cancer classification using MRI images.
• To leverage transfer learning (TL) to overcome limitations of small datasets in PCa image analysis.
• To improve the efficiency and accuracy of PCa identification for radiologists.

Main Methods:

• Implementation of a deep learning model utilizing the EfficientNet architecture, pre-trained on ImageNet.
• Integration of a multi-branch approach for feature extraction from diverse MRI sequences.
• Application of transfer learning (TL) to a limited dataset for robust PCa classification.

Main Results:

• The proposed model achieved a high accuracy rate of 88.89% in classifying prostate cancer.
• Comparative analysis demonstrated superior performance over traditional feature engineering and existing DL methods.
• The methodology effectively extracts distinctive features from prostate MRI, leading to accurate cancer identification.

Conclusions:

• The combined deep learning and transfer learning approach significantly enhances prostate cancer classification accuracy from MRI.
• This method offers a promising tool for radiologists, aiding in the timely and precise detection of PCa.
• The study highlights the potential of advanced AI techniques in addressing resource constraints and improving oncological diagnostics.