Previous methodological approaches like gradient-weighted class activation mapping (Grad-CAM), occlusion sensitivity analyses, and local interpretable model-agnostic explanations (LIME) had the limitation that deriving the relevance or saliency maps provided only group-average estimates, required long runtime, or provided only low spatial resolution. In general, relevance or saliency maps indicate the amount of information or contribution of a single input feature on the probability of a particular output class. Novel methods for deriving relevance maps from CNN models may help to overcome the black-box problem. The poor intuitive comprehensibility of CNNs is one of the major obstacles which hinder the clinical application. Despite the high accuracy levels achieved by CNN models, a major drawback is their algorithmic complexity, which renders them black-box systems. Various studies have evaluated the performance of CNNs for the detection of Alzheimer’s disease in MR images with promising results regarding both separation of diagnostic groups and the prediction of conversion from MCI to manifest dementia. Ĭonvolutional neural networks (CNNs) provide a powerful method for image recognition. Automated detection of subtle brain changes in early stages of Alzheimer’s disease could improve diagnostic confidence and early access to intervention. Therefore, hippocampus volume is currently the best-established MRI marker for diagnosing Alzheimer’s disease at the dementia stage as well as at its prodromal stage amnestic mild cognitive impairment (MCI). Particularly at earlier stages of AD, atrophy patterns are relatively regionally specific, with volume loss in the medial temporal lobe and particularly the hippocampus. The presented data-driven and hypothesis-free CNN modeling approach might provide a useful tool to automatically derive discriminative features for complex diagnostic tasks where clear clinical criteria are still missing, for instance for the differential diagnosis between various types of dementia.Īlzheimer’s disease (AD) is characterized by widespread neuronal degeneration, which manifests macroscopically as cortical atrophy that can be detected in vivo using structural magnetic resonance imaging (MRI) scans.
The high hippocampus relevance scores as well as the high performance achieved in independent samples support the validity of the CNN models in the detection of AD-related MRI abnormalities. This strengthens the comprehensibility of the CNN models, which were trained in a purely data-driven manner based on the scans and diagnosis labels. The relevance maps highlighted atrophy in regions that we had hypothesized a priori. Relevance scores within the hippocampus were highly correlated with hippocampal volumes (Pearson’s r ≈ −0.86, p < 0.001). Relevance maps indicated that hippocampal atrophy was considered the most informative factor for AD detection, with additional contributions from atrophy in other cortical and subcortical regions.
ResultsĪcross the three independent datasets, group separation showed high accuracy for AD dementia versus controls ( AUC ≥ 0.91) and moderate accuracy for amnestic MCI versus controls ( AUC ≈ 0.74). To improve model comprehensibility, we implemented an interactive visualization of 3D CNN relevance maps, thereby allowing intuitive model inspection. We evaluated the association of relevance scores and hippocampus volume to validate the clinical utility of this approach. We trained a CNN for the detection of AD in N = 663 T1-weighted MRI scans of patients with dementia and amnestic mild cognitive impairment (MCI) and verified the accuracy of the models via cross-validation and in three independent samples including in total N = 1655 cases. We investigated whether models with higher accuracy also rely more on discriminative brain regions predefined by prior knowledge. Recently developed visualization methods for deriving CNN relevance maps may help to fill this gap as they allow the visualization of key input image features that drive the decision of the model. One important reason for this is a lack of model comprehensibility. for the ADNI, AIBL, DELCODE study groupsĪlzheimer's Research & Therapy volume 13, Article number: 191 ( 2021)Īlthough convolutional neural networks (CNNs) achieve high diagnostic accuracy for detecting Alzheimer’s disease (AD) dementia based on magnetic resonance imaging (MRI) scans, they are not yet applied in clinical routine.Improving 3D convolutional neural network comprehensibility via interactive visualization of relevance maps: evaluation in Alzheimer’s disease