Semisupervised white matter hyperintensities segmentation on MRI.

This study proposed a semisupervised loss function named level-set loss (LSLoss) for cerebral white matter hyperintensities (WMHs) segmentation on fluid-attenuated inversion recovery images. The training procedure did not require manually labeled WMH masks. Our image preprocessing steps included biased field correction, skull stripping, and white matter segmentation. With the proposed LSLoss, we trained a V-Net using the MRI images from both local and public databases. Local databases were the small vessel disease cohort (HKU-SVD, n = 360) and the multiple sclerosis cohort (HKU-MS, n = 20) from our institutional imaging center. Public databases were the Medical Image Computing Computer-assisted Intervention (MICCAI) WMH challenge database (MICCAI-WMH, n = 60) and the normal control cohort of the Alzheimer's Disease Neuroimaging Initiative database (ADNI-CN, n = 15). We achieved an overall dice similarity coefficient (DSC) of 0.81 on the HKU-SVD testing set (n = 20), DSC = 0.77 on the HKU-MS testing set (n = 5), and DSC = 0.78 on MICCAI-WMH testing set (n = 30). The segmentation results obtained by our semisupervised V-Net were comparable with the supervised methods and outperformed the unsupervised methods in the literature.

A rare variant in MLKL confers susceptibility to ApoE ɛ4-negative Alzheimer's disease in Hong Kong Chinese population.

Alzheimer's disease (AD) is the most common neurodegenerative disorders in the elderly. To identify rare genetic factors other than apolipoprotein E ɛ4 allele (ApoE ɛ4) contributing to the pathogenesis of late-onset AD (LOAD), we conducted a whole-exome analysis of 246 ApoE ɛ4-negative LOAD cases and 172 matched controls in Hong Kong Chinese population. LOAD patients showed a significantly higher burden of rare loss-of-function variants in genes related to immune function than healthy controls. Among the genes involved in immune function, we identified a rare stop-gain variant (p.Q48X) in mixed lineage kinase domain like pseudokinase (MLKL) gene present exclusively in 6 LOAD cases. MLKL is expressed in neurons, and the its expression levels in the p.Q48X carriers were significantly lower than that in age-matched wild-type controls. The ratio of Aß42 to Aß40 significantly increased in MLKL knockdown cells compared to scramble controls. MLKL loss-of-function mutation might contribute to late-onset ApoE ɛ4-negative AD in the Hong Kong Chinese population.

Multimodal MRI of the hippocampus in Parkinson's disease with visual hallucinations.

Visual hallucinations carry poor prognosis in Parkinson's disease. Here we tested the hypothesis that the hippocampus and visuospatial memory impairment play a central role in the pathology of PD with visual hallucinations. Multimodal magnetic resonance imaging of the brain was carried out in 12 people with PD and visual hallucinations; 15 PD individuals without hallucinations; and 14 healthy controls. Age, gender, cognitive ability, and education level were matched across the three groups. PD patients were taking dopaminergic medication. Hippocampal volume, shape, mean diffusivity (MD), and functional connectivity within the whole brain were examined. Visuospatial memory was compared between groups, and correlations with hippocampal MD, functional connectivity, and the severity of hallucinations were explored. There were no macrostructural differences across groups, but individuals with hallucinations had higher diffusivity in posterior hippocampus than the other two groups. Visuospatial memory was poorer in both PD groups compared to controls, and was correlated with hallucinations. Finally, hippocampal functional connectivity in the visual cortices was lower in those with hallucinations than other groups, and this correlated with visuospatial memory impairment. In contrast, functional connectivity between the hippocampus and default mode network regions and frontal regions was greater in the PD hallucinators compared to other groups. We suggest that hippocampal pathology, which disrupts visuospatial memory, makes a key contribution to visual hallucinations in PD. These findings may pave the way for future studies of imaging biomarkers to measure treatment response in those with PD who are most at risk of poor outcomes.

The default mode network is disrupted in Parkinson's disease with visual hallucinations.

BACKGROUND: Visual hallucinations (VH) are one of the most striking nonmotor symptoms in Parkinson's disease (PD), and predict dementia and mortality. Aberrant default mode network (DMN) is associated with other psychoses. Here, we tested the hypothesis that DMN dysfunction contributes to VH in PD. METHODS: Resting state functional data was acquired from individuals with PD with VH (PDVH) and without VH (PDnonVH), matched for levodopa drug equivalent dose, and a healthy control group (HC). Independent component analysis was used to investigate group differences in functional connectivity within the DMN. In addition, we investigated whether the functional changes associated with hallucinations were accompanied by differences in cortical thickness. RESULTS: There were no group differences in cortical thickness but functional coactivation within components of the DMN was significantly lower in both PDVH and PDnonVH groups compared to HC. Functional coactivation within the DMN was found to be greater in PDVH group relative to PDnonVH group. CONCLUSION: Our study demonstrates, for the first time that, within a functionally abnormal DMN in PD, relatively higher "connectivity" is associated with VH. We postulate that aberrant connectivity in a large scale network affects sensory information processing and perception, and contributes to "positive" symptom generation in PD.