In Laboratory investigation; a journal of technical methods and pathology ; h5-index 42.0
Neuropathological assessment at autopsy is the gold standard for diagnosing neurodegenerative disorders. Neurodegenerative conditions, such as Alzheimer's disease (AD), neuropathological changes are a continuous process from normal aging rather than categorical; therefore, diagnosing neurodegenerative disorders is a complicated task. We aimed to develop a pipeline for diagnosing AD and other tauopathies, including corticobasal degeneration (CBD), globular glial tauopathy (GGT), Pick's disease (PiD), and progressive supranuclear palsy (PSP). We used a weakly supervised deep learning-based approach called clustering-constrained-attention multiple instance learning (CLAM) on whole slide images (WSIs) of patients with AD (n=30), CBD (n=20), GGT (n=10), PiD (n=20), and PSP (n=20), as well as non-tauopathy controls (n=21). Three sections (A: motor cortex; B: cingulate gyrus and superior frontal gyrus; C: corpus striatum) that had been immunostained for phosphorylated-tau were scanned and converted to WSIs. We evaluated three models (classical multiple instance learning, single-attention-branch CLAM, and multi-attention-branch CLAM) using 5-fold cross-validation. Attention-based interpretation analysis was performed to identify morphological features contributing to the classification. Within highly attended regions, we also augmented gradient-weighted class activation mapping (Grad-CAM) to the model to visualize cellular-level evidence of the model's decisions. The multi-attention-branch CLAM model using Section B achieved the highest area under the curve (0.970 ± 0.037) and diagnostic accuracy (0.873 ± 0.087). A heatmap showed the highest attention in the gray matter of the superior frontal gyrus in AD and the white matter of the cingulate gyrus in CBD. Grad-CAM showed the highest attention in characteristic tau lesions for each disease (e.g., numerous tau-positive threads in the white matter inclusions for CBD). Our findings supported the feasibility of deep learning-based approaches for the classification of neurodegenerative disorders on WSIs. Further investigation of this method focusing on clinicopathological correlations, is warranted.
Kim Minji, Sekiya Hiroaki, Yao Gary, Martin Nicholas B, Castanedes-Casey Monica, Dickson Dennis W, Hwang Tae Hyun, Koga Shunsuke
2023-Mar-06
Alzheimer’s disease, CLAM, Grad-CAM, Pick’s disease, corticobasal degeneration, globular glial tauopathy, neuropathology, progressive supranuclear palsy, tauopathy, weakly supervised deep learning
