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.

Worldwide, there are nearly 10 million new cases of dementia annually, of which Alzheimer’s disease (AD) is the most common. New measures are needed to improve the diagnosis of individuals with cognitive impairment due to various etiologies. Here, we report a deep learning framework that accomplishes multiple diagnostic steps in successive fashion to identify persons with normal cognition (NC), mild cognitive impairment (MCI), AD, and non-AD dementias (nADD). We demonstrate a range of models capable of accepting flexible combinations of routinely collected clinical information, including demographics, medical history, neuropsychological testing, neuroimaging, and functional assessments. We then show that these frameworks compare favorably with the diagnostic accuracy of practicing neurologists and neuroradiologists. Lastly, we apply interpretability methods in computer vision to show that disease-specific patterns detected by our models track distinct patterns of degenerative changes throughout the brain and correspond closely with the presence of neuropathological lesions on autopsy. Our work demonstrates methodologies for validating computational predictions with established standards of medical diagnosis.

Objective To describe the detailed clinical characteristics, immunotherapy, and long-term outcomes of patients with anti-NMDA receptor (NMDAR) encephalitis in China. Methods A single-center, prospective study. Patients who met the diagnostic criteria were enrolled from 2011 to 2017 and followed up. The clinical features, treatment, and long-term outcomes were collected prospectively. Factors affecting the long-term prognosis were analyzed. Results The study included 220 patients. The most common clinical presentations were psychosis (82.7%) and seizures (80.9%). Of the patients, 19.5% had an underlying neoplasm; of which ovarian teratoma was 100% of tumors in females and only one male had lung cancer. Most patients (99.5%) received first-line therapy (glucocorticoids, IV immunoglobulin, or plasmapheresis alone or combined), and only 7.3% received second-line immunotherapy (rituximab, cyclophosphamide alone, or combined). Long-term immunotherapy (mycophenolate mofetil or azathioprine >1 year) was administered to 53.2% of patients. During the first 12 months, 207 (94.1%) patients experienced improvement, and 5 (2.3%) died, whereas 38 (17.3%) experienced relapses. At 12-month follow-up, 92.7% had favorable clinical outcomes (modified Rankin Scale score ≤2). Conclusions Patients in China present with psychosis and seizure frequently but have a low percentage of underlying neoplasms. Re-enforced first-line immunotherapy is effective in managing anti-NMDAR encephalitis in the acute phase. Although relapse is relatively common, with combined first-line and long-term immunotherapy, most patients reached favorable outcomes.

The positive association of choline for cognition has been reported in both animal and human studies, yet the associations of choline with the risks of incident dementia or Alzheimer's disease (AD) in humans is unclear. Our objective was to test the hypothesis that lower or higher dietary choline intake is associated with increased or decreased, respectively, risks of incident dementia and AD. Data from the Framingham Heart Study Offspring Cohort exam 5 to exam 9 were used. Participants were free of dementia and stroke, with a valid self-reported 126-item Harvard FFQ at exam 5. The intakes of total choline, its contributing compounds, and betaine were estimated based on a published nutrient database. The intakes were updated at each exam to represent the cumulative average intake across the 5 exams. The associations between dietary choline intakes and incident dementia and AD were examined in mixed-effect Cox proportional hazard models, adjusting for covariates. A total of 3224 participants (53.8% female; mean ± SD age, 54.5 ± 9.7 y) were followed up for a mean ± SD of 16.1 ± 5.1 y (1991–2011). There were 247 incident dementia cases, of which 177 were AD. Dietary choline intake showed nonlinear relationships with incident dementia and AD. After adjusting for covariates, low choline intake (defined as ≤ 219 and ≤ 215 mg/d for dementia and AD, respectively) was significantly associated with incident dementia and incident AD. Low choline intake was associated with increased risks of incident dementia and AD.

Purpose The aim of this study was to describe brain metabolic changing patterns demonstrated by serial brain FDG PET/CT scans and their relationship with the clinical course in patients with anti-N-methyl-d-aspartate receptor encephalitis (ANMDARE). Patients and Methods Eighteen serial PET scans of 8 patients with ANMDARE were reviewed. All the 18 PET scans were divided into 4 groups according to studies timing in different clinical course: group A, the acute and subacute phase; group B, early recovery phase; group C, recovery phase; and group D, relapsing phase. Antibody levels of ANMDARE of all these patients were tested at the same time. The PET images of each group were analyzed visually and also compared with 10 age- and sex-matched normal controls using voxel-wise statistical parametric mapping analysis (SPM5). Results Variable brain metabolic patterns and its association with the clinical course and the levels of NMDA antibody were demonstrated by FDG PET images. First, severe hypometabolism in bilateral occipital lobes and relatively mild hypermetabolism in the partial frontal and basal ganglia in acute and subacute phase, the level of antibody was high. Second, in early recovery phase when the symptoms was partially improved, extensive cortical hypometabolism was observed, and the level of antibody was low. Third, the patients in the recovery phase have no obvious neurological and psychiatric symptoms; PET images were nearly normal, and the antibodies tests were all negative, correspondingly. Fourth, 3 scans of relapsing phase presented heterogeneous brain metabolic abnormalities. Conclusions There existed a specific serial brain metabolic changing pattern that correlated with the clinical course and antibody level in ANMDARE.