Brain extraction is an important step in the analysis of brain images. The variability in brain morphology and the difference in intensity characteristics due to imaging sequences make the development of a general purpose brain extraction algorithm challenging. To address this issue, we propose a new robust method (BEaST) dedicated to produce consistent and accurate brain extraction. This method is based on nonlocal segmentation embedded in a multi-resolution framework. A library of 80 priors is semi-automatically constructed from the NIH-sponsored MRI study of normal brain development, the International Consortium for Brain Mapping, and the Alzheimer's Disease Neuroimaging Initiative databases. In testing, a mean Dice similarity coefficient of 0.9834±0.0053 was obtained when performing leave-one-out cross validation selecting only 20 priors from the library. Validation using the online Segmentation Validation Engine resulted in a top ranking position with a mean Dice coefficient of 0.9781±0.0047. Robustness of BEaST is demonstrated on all baseline ADNI data, resulting in a very low failure rate. The segmentation accuracy of the method is better than two widely used publicly available methods and recent state-of-the-art hybrid approaches. BEaST provides results comparable to a recent label fusion approach, while being 40 times faster and requiring a much smaller library of priors.

Algorithms for computer-aided diagnosis of dementia based on structural MRI have demonstrated high performance in the literature, but are difficult to compare as different data sets and methodology were used for evaluation. In addition, it is unclear how the algorithms would perform on previously unseen data, and thus, how they would perform in clinical practice when there is no real opportunity to adapt the algorithm to the data at hand. To address these comparability, generalizability and clinical applicability issues, we organized a grand challenge that aimed to objectively compare algorithms based on a clinically representative multi-center data set. Using clinical practice as the starting point, the goal was to reproduce the clinical diagnosis. Therefore, we evaluated algorithms for multi-class classification of three diagnostic groups: patients with probable Alzheimer's disease, patients with mild cognitive impairment and healthy controls. The diagnosis based on clinical criteria was used as reference standard, as it was the best available reference despite its known limitations. For evaluation, a previously unseen test set was used consisting of 354 T1-weighted MRI scans with the diagnoses blinded. Fifteen research teams participated with a total of 29 algorithms. The algorithms were trained on a small training set (n=30) and optionally on data from other sources (e.g., the Alzheimer's Disease Neuroimaging Initiative, the Australian Imaging Biomarkers and Lifestyle flagship study of aging). The best performing algorithm yielded an accuracy of 63.0% and an area under the receiver-operating-characteristic curve (AUC) of 78.8%. In general, the best performances were achieved using feature extraction based on voxel-based morphometry or a combination of features that included volume, cortical thickness, shape and intensity. The challenge is open for new submissions via the web-based framework: http://caddementia.grand-challenge.org.

Background: The aim of this longitudinal study was to assess with positron emission tomography (PET) the relationship between levels of inflammation and the loads of aggregated β-amyloid and tau at baseline and again after 2 years in prodromal Alzheimer's disease.Methods: Forty-three subjects with mild cognitive impairment (MCI) had serial¹¹C-PK11195 PET over 2 years to measure inflammation changes, and¹¹C-PiB PET to determine β-amyloid fibril load; 22 also had serial¹⁸F-Flortaucipir PET to determine tau tangle load. Cortical surface statistical mapping was used to localise areas showing significant changes in tracer binding over time and to interrogate correlations between tracer binding of the tracers at baseline and after 2 years.Results: Those MCI subjects with high¹¹C-PiB uptake at baseline (classified as prodromal Alzheimer’s disease) had raised inflammation levels which significantly declined across cortical regions over 2 years although their β-amyloid levels continued to rise. Those MCI cases who had low/normal¹¹C-PiB uptake at baseline but their levels then rose over 2 years were classified as prodromal AD with low Thal phase 1-2 amyloid deposition at baseline. They showed levels of cortical inflammation which correlated with their rising β-amyloid load. Those MCI cases with baseline low¹¹C-PiB uptake that remained stable were classified as non-AD, and they showed no correlated inflammation levels. Finally, MCI cases which showed both high¹¹C-PiB and¹⁸F-Flortaucipir uptake at baseline (MCI due to AD) showed a further rise in their tau tangle load over 2 years with a correlated rise in levels of inflammation.Conclusions: Our baseline and 2-year imaging findings are compatible with a biphasic trajectory of inflammation in Alzheimer’s disease: MCI cases with low baseline but subsequently rising β-amyloid load show correlated levels of microglial activation which then later decline when the β-amyloid load approaches AD levels. Later, as tau tangles form in β-amyloid positive MCI cases with prodromal AD, the rising tau load is associated with higher levels of inflammation.

See Kreisl (doi: 10.1093/awx151 ) for a scientific commentary on this article. Subjects with mild cognitive impairment associated with cortical amyloid-β have a greatly increased risk of progressing to Alzheimer’s disease. We hypothesized that neuroinflammation occurs early in Alzheimer’s disease and would be present in most amyloid-positive mild cognitive impairment cases. ¹¹C-Pittsburgh compound B and ¹¹C-(R)-PK11195 positron emission tomography was used to determine the amyloid load and detect the extent of neuroinflammation (microglial activation) in 42 mild cognitive impairment cases. Twelve age-matched healthy control subjects had ¹¹C-Pittsburgh compound B and 10 healthy control subjects had ¹¹C-(R)-PK11195 positron emission tomography for comparison. Amyloid-positivity was defined as ¹¹C-Pittsburgh compound B target-to-cerebellar ratio above 1.5 within a composite cortical volume of interest. Supervised cluster analysis was used to generate parametric maps of ¹¹C-(R)-PK11195 binding potential. Levels of ¹¹C-(R)-PK11195 binding potential were measured in a selection of cortical volumes of interest and at a voxel level. Twenty-six (62%) of 42 mild cognitive impairment cases showed a raised cortical amyloid load compared to healthy controls. Twenty-two (85%) of the 26 amyloid-positive mild cognitive impairment cases showed clusters of increased cortical microglial activation accompanying the amyloid. There was a positive correlation between levels of amyloid load and ¹¹C-(R)-PK11195 binding potentials at a voxel level within subregions of frontal, parietal and temporal cortices. ¹¹C-(R)-PK11195 positron emission tomography reveals increased inflammation in a majority of amyloid positive mild cognitive impairment cases, its cortical distribution overlapping that of amyloid deposition.