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.
Abstract (translated)
磁共振成像（MRI）是评估膝关节损伤的重要诊断工具。然而，手动解释MRI切片既耗时又容易出现观察者之间的差异。本研究系统地评估了各种深度学习架构与可解释人工智能(xAI)技术的结合，用于自动检测膝部MRI扫描中的感兴趣区域(ROI)。我们探讨了监督和自监督方法，包括ResNet50、InceptionV3、Vision Transformers (ViT)，以及多个通过多层感知器（MLP）分类器增强的U-Net变体。为了提高可解释性和临床相关性，我们整合了诸如Grad-CAM和显著图(Saliency Maps)等xAI方法。模型性能通过分类的AUC值及重建质量的PSNR/SSIM指标进行评估，并辅以定性的ROI可视化结果。 我们的研究结果显示，在MRNet数据集的约束下，ResNet50在分类和ROI识别方面始终表现出色，优于基于变换器（transformer）的模型。尽管混合U-Net + MLP方法显示出利用空间特征于重建和可解释性方面的潜力，但它们的分类性能仍较低。Grad-CAM在整个架构中持续提供了最具临床意义的解释。总体而言，基于卷积神经网络(CNN)的迁移学习对于该数据集是最有效的策略，而未来使用更大规模预训练的研究可能会更好地解锁变换器模型的潜在优势。
URL
https://arxiv.org/abs/2508.14151
