A Systematic Study of Deep Learning Models and xAI Methods for Region-of-Interest Detection in MRI Scans

TL;DR

This study systematically evaluates deep learning and xAI methods for automated knee MRI ROI detection, finding CNNs like ResNet50 most effective, with Grad-CAM providing meaningful explanations, highlighting the potential of transfer learning.

Contribution

It provides a comprehensive comparison of various deep learning architectures and xAI techniques for knee MRI ROI detection, emphasizing the effectiveness of CNNs over transformers in this context.

Findings

ResNet50 outperforms transformer models in classification and ROI detection.

Grad-CAM offers the most clinically relevant explanations.

CNN transfer learning is most effective for the MRNet dataset.

Abstract

Magnetic Resonance Imaging (MRI) is an essential diagnostic tool for assessing knee injuries. However, manual interpretation of MRI slices remains time-consuming and prone to inter-observer variability. This study presents a systematic evaluation of various deep learning architectures combined with explainable AI (xAI) techniques for automated region of interest (ROI) detection in knee MRI scans. We investigate both supervised and self-supervised approaches, including ResNet50, InceptionV3, Vision Transformers (ViT), and multiple U-Net variants augmented with multi-layer perceptron (MLP) classifiers. To enhance interpretability and clinical relevance, we integrate xAI methods such as Grad-CAM and Saliency Maps. Model performance is assessed using AUC for classification and PSNR/SSIM for reconstruction quality, along with qualitative ROI visualizations. Our results demonstrate that ResNet50 consistently excels in classification and ROI identification, outperforming transformer-based models under the constraints of the MRNet dataset. While hybrid U-Net + MLP approaches show potential for leveraging spatial features in reconstruction and interpretability, their classification performance remains lower. Grad-CAM consistently provided the most clinically meaningful explanations across architectures. Overall, CNN-based transfer learning emerges as the most effective approach for this dataset, while future work with larger-scale pretraining may better unlock the potential of transformer models.