[论文解读] Biomarker Investigation using Multiple Brain Measures from MRI through XAI in Alzheimer's Disease Classification

Davide Coluzzi, Valentina Bordin+5 more|arXiv (Cornell University)|May 3, 2023

Functional Brain Connectivity Studies被引用 4

一句话总结

本研究提出两种深度学习模型——用于结构磁共振成像的ResNet18和用于脑连接矩阵的BC-GCN-SE——并结合可解释人工智能（XAI）技术，通过Grad-CAM识别阿尔茨海默病的生物标志物。结果表明，两种模型均将注意力集中在临床上相关的脑区，如内侧颞叶和默认模式网络，其中位真阳性率分别为0.817和0.703，显著提升了模型的可解释性与临床信任度。

ABSTRACT

Alzheimer's Disease (AD) is the world leading cause of dementia, a progressively impairing condition leading to high hospitalization rates and mortality. To optimize the diagnostic process, numerous efforts have been directed towards the development of deep learning approaches (DL) for the automatic AD classification. However, their typical black box outline has led to low trust and scarce usage within clinical frameworks. In this work, we propose two state-of-the art DL models, trained respectively on structural MRI (ResNet18) and brain connectivity matrixes (BC-GCN-SE) derived from diffusion data. The models were initially evaluated in terms of classification accuracy. Then, results were analyzed using an Explainable Artificial Intelligence (XAI) approach (Grad-CAM) to measure the level of interpretability of both models. The XAI assessment was conducted across 132 brain parcels, extracted from a combination of the Harvard-Oxford and AAL brain atlases, and compared to well-known pathological regions to measure adherence to domain knowledge. Results highlighted acceptable classification performance as compared to the existing literature (ResNet18: TPRmedian = 0.817, TNRmedian = 0.816; BC-GCN-SE: TPRmedian = 0.703, TNRmedian = 0.738). As evaluated through a statistical test (p < 0.05) and ranking of the most relevant parcels (first 15%), Grad-CAM revealed the involvement of target brain areas for both the ResNet18 and BC-GCN-SE models: the medial temporal lobe and the default mode network. The obtained interpretabilities were not without limitations. Nevertheless, results suggested that combining different imaging modalities may result in increased classification performance and model reliability. This could potentially boost the confidence laid in DL models and favor their wide applicability as aid diagnostic tools.

研究动机与目标

• 通过多模态MRI数据的深度学习提升阿尔茨海默病的诊断准确性。
• 通过应用可解释人工智能（XAI）技术解释模型决策，增强临床对AI模型的信任。
• 通过Grad-CAM可解释性分析，将模型预测与已知的病理脑区进行验证。
• 评估结合结构与功能脑测量是否能提升分类性能与可靠性。
• 通过Grad-CAM显著性图的统计排序，识别阿尔茨海默病分类中最相关的脑区。

提出的方法

• 在T1加权结构磁共振成像扫描上训练ResNet18模型，用于分类阿尔茨海默病。
• 开发BC-GCN-SE模型，处理来自扩散磁共振成像数据的脑连接矩阵。
• 应用Grad-CAM生成132个哈佛-牛津图谱与AAL图谱中脑区的类别激活图，以增强可解释性。
• 将Grad-CAM生成的显著性分数映射到解剖区域，并与阿尔茨海默病中已知的病理区域进行比较。
• 使用统计检验（p < 0.05）及对最相关脑区前15%的排序，评估模型与领域知识的一致性。
• 通过交叉验证折中的中位真阳性率（TPR）与真阴性率（TNR）评估模型性能。

实验结果

研究问题

• RQ1 当从MRI数据分类阿尔茨海默病时，深度学习模型关注哪些脑区？
• RQ2 这些模型的注意力图在多大程度上与阿尔茨海默病中已知的病理区域一致？
• RQ3 结合结构磁共振成像与脑连接数据如何影响分类性能与可解释性？
• RQ4 XAI技术（如Grad-CAM）能否有效验证并增强深度学习模型在神经退行性疾病分类中的可信度？
• RQ5 Grad-CAM识别出的最显著脑区是否与默认模式网络或内侧颞叶等临床相关网络相对应？

主要发现

• ResNet18模型实现了中位真阳性率（TPR）0.817与中位真阴性率（TNR）0.816，表明其具有出色的分类性能。
• BC-GCN-SE模型实现了中位TPR 0.703与中位TNR 0.738，表现出中等但具有临床意义的准确率。
• Grad-CAM分析显示，两种模型均将注意力集中在内侧颞叶与默认模式网络——这些区域在阿尔茨海默病病理中具有关键作用。
• 统计检验证实，最相关脑区前15%的显著性区域与已知病理区域存在显著重叠（p < 0.05）。
• 可解释性分析表明，模型预测并非随机，而是与阿尔茨海默病的已知神经解剖学知识高度一致。
• 结果表明，结合多模态MRI与XAI可提升模型的可靠性，并支持深度学习在神经退行性疾病诊断中的临床应用。