Development and validation of an interpretable deep learning framework for Alzheimer's disease classification.

Alzheimer's disease is the primary cause of dementia worldwide, with an increasing morbidity burden that may outstrip diagnosis and management capacity as the population ages. Current methods integrate patient history, neuropsychological testing and MRI to identify likely cases, yet effective practices remain variably applied and lacking in sensitivity and specificity. Here we report an interpretable deep learning strategy that delineates unique Alzheimer's disease signatures from multimodal inputs of MRI, age, gender, and Mini-Mental State Examination score. Our framework linked a fully convolutional network, which constructs high resolution maps of disease probability from local brain structure to a multilayer perceptron and generates precise, intuitive visualization of individual Alzheimer's disease risk en route to accurate diagnosis. The model was trained using clinically diagnosed Alzheimer's disease and cognitively normal subjects from the Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset (n = 417) and validated on three independent cohorts: the Australian Imaging, Biomarker and Lifestyle Flagship Study of Ageing (AIBL) (n = 382), the Framingham Heart Study (n = 102), and the National Alzheimer's Coordinating Center (NACC) (n = 582). Performance of the model that used the multimodal inputs was consistent across datasets, with mean area under curve values of 0.996, 0.974, 0.876 and 0.954 for the ADNI study, AIBL, Framingham Heart Study and NACC datasets, respectively. Moreover, our approach exceeded the diagnostic performance of a multi-institutional team of practicing neurologists (n = 11), and high-risk cerebral regions predicted by the model closely tracked post-mortem histopathological findings. This framework provides a clinically adaptable strategy for using routinely available imaging techniques such as MRI to generate nuanced neuroimaging signatures for Alzheimer's disease diagnosis, as well as a generalizable approach for linking deep learning to pathophysiological processes in human disease.

Factor VIII in the setting of acute ischemic stroke among patients with suspected hypercoagulable state.

BACKGROUND: Elevation of factor VIII is associated with higher risk of large vessel arterial occlusions including stroke. METHODS: Factor VIII levels were examined in consecutive patients with acute ischemic stroke (AIS) presenting to a single center between July 2008 and May 2012. Factor VIII levels exceeding the laboratory reference range were considered elevated (>150%). RESULTS: Factor VIII level was elevated in 72.4% (84 of 116) of the patients. Elevated factor VIII level was more frequent in blacks, diabetics, and patients who were anemic. Patients with elevated factor VIII had higher median baseline National Institute of Health Stroke Scale (NIHSS; 5 vs 2, P = .0295) and twice the frequency of neuroworsening (21.4% vs 9.4%), but discharge NIHSS and modified Rankin Scale were similar in the groups. CONCLUSIONS: High factor VIII level was found in the majority of tested patients with AIS. Several baseline differences were found between patients with normal and high factor VIII levels, but no differences were identified in outcome.

The NIHSS Score and its Components can Predict Cortical Stroke.

BACKGROUND: Neuroprotective agents have the potential to reduce ischemia to penumbra of the cortex, but are time-sensitive. To quickly determine whether a cortical stroke is present without imaging, we created a scoring system based on the NIH stroke scale (NIHSS) that can accurately predict cortical damage in an acute ischemic stroke (AIS). METHODS: Patients treated with tPA for AIS were retrospectively assessed through prospectively acquired databases at two stroke centers. Stroke was classified as cortical vs. non-cortical stroke. The total NIHSS score, cortical components (gaze, visual fields, language, and neglect) and cortical score (sum of cortical components) were analyzed for site 1 and then validated for site 2 for sensitivity and positive predictive value (PPV) for a cortical stroke. RESULTS: An acute infarct was detected in 194/239 (81%) patients at site 1 and 122/174 (70%) at site 2 on diffusion-weighted MRI. Cortical involvement was found in 71% (site 1) and 75% (site 2). The median cortical score was 25% of the total NIHSS score at both sites. NIHSS ≥ 4 had the highest sensitivity; PPV was 90% for any cortical sign with ≥ 2 points. The best combination of sensitivity and PPV was cortical score/NIHSS score ≥10%. DISCUSSION: If a trial targeting cortical stroke required that the cortical score represent at least 10% of the total NIHSS score with no imaging, less than 10% of patients with cortical stroke would be missed and less than 18% of patients would be misclassified as having a cortical stroke.