Relative rate of change in cognitive score network dynamics via Bayesian hierarchical models reveal spatial patterns of neurodegeneration.

The degeneration of the human brain is a complex process, which often affects certain brain regions due to healthy aging or disease. This degeneration can be evaluated on regions of interest (ROI) in the brain through probabilistic networks and morphological estimates. Current approaches for finding such networks are limited to analyses at discrete neuropsychological stages, which cannot appropriately account for connectivity dynamics over the onset of cognitive deterioration, and morphological changes are seldom unified with connectivity networks, despite known dependencies. To overcome these limitations, a probabilistic wombling model is proposed to simultaneously estimate ROI cortical thickness and covariance networks contingent on rates of change in cognitive decline. This proposed model was applied to analyze longitudinal data from healthy control (HC) and Alzheimer's disease (AD) groups and found connection differences pertaining to regions, which play a crucial role in lasting cognitive impairment, such as the entorhinal area and temporal regions. Moreover, HC cortical thickness estimates were significantly higher than those in the AD group across all ROIs. The analyses presented in this work will help practitioners jointly analyze brain tissue atrophy at the ROI-level conditional on neuropsychological networks, which could potentially allow for more targeted therapeutic interventions.

How many streamlines are required for reliable probabilistic tractography? Solutions for microstructural measurements and neurosurgical planning.

Diffusion MRI tractography is commonly used to delineate white matter tracts. These delineations can be used for planning neurosurgery or for identifying regions of interest from which microstructural measurements can be taken. Probabilistic tractography produces different delineations each time it is run, potentially leading to microstructural measurements or anatomical delineations that are not reproducible. Generating a sufficiently large number of streamlines is required to avoid this scenario, but what constitutes "sufficient" is difficult to assess and so streamline counts are typically chosen in an arbitrary or qualitative manner. This work explores several factors influencing tractography reliability and details two methods for estimating this reliability. The first method automatically estimates the number of streamlines required to achieve reliable microstructural measurements, whilst the second estimates the number of streamlines required to achieve a reliable binarised trackmap than can be used clinically. Using these methods, we calculated the number of streamlines required to achieve a range of quantitative reproducibility criteria for three anatomical tracts in 40 Human Connectome Project datasets. Actual reproducibility was checked by repeatedly generating the tractograms with the calculated numbers of streamlines. We found that the required number of streamlines varied strongly by anatomical tract, image resolution, number of diffusion directions, the degree of reliability desired, the microstructural measurement of interest, and/or the specifics on how the tractogram was converted to a binary volume. The proposed methods consistently predicted streamline counts that achieved the target reproducibility. Implementations are made available to enable the scientific community to more-easily achieve reproducible tractography.

An efficient algorithm for estimating brain covariance networks.

Often derived from partial correlations or many pairwise analyses, covariance networks represent the inter-relationships among regions and can reveal important topological structures in brain measures from healthy and pathological subjects. However both approaches are not consistent network estimators and are sensitive to the value of the tuning parameters. Here, we propose a consistent covariance network estimator by maximising the network likelihood (MNL) which is robust to the tuning parameter. We validate the consistency of our algorithm theoretically and via a simulation study, and contrast these results against two well-known approaches: the graphical LASSO (gLASSO) and Pearson pairwise correlations (PPC) over a range of tuning parameters. The MNL algorithm had a specificity equal to and greater than 0.94 for all sample sizes in the simulation study, and the sensitivity was shown to increase as the sample size increased. The gLASSO and PPC demonstrated a specificity-sensitivity trade-off over a range of values of tuning parameters highlighting the discrepancy in the results for misspecified values. Application of the MNL algorithm to the case study data showed a loss of connections between healthy and impaired groups, and improved ability to identify between lobe connectivity in contrast to gLASSO networks. In this work, we propose the MNL algorithm as an effective approach to find covariance brain networks, which can inform the organisational features in brain-wide analyses, particularly for large sample sizes.

Comparisons of neurodegeneration over time between healthy ageing and Alzheimer's disease cohorts via Bayesian inference.

OBJECTIVES: In recent years, large-scale longitudinal neuroimaging studies have improved our understanding of healthy ageing and pathologies including Alzheimer's disease (AD). A particular focus of these studies is group differences and identification of participants at risk of deteriorating to a worse diagnosis. For this, statistical analysis using linear mixed-effects (LME) models are used to account for correlated observations from individuals measured over time. A Bayesian framework for LME models in AD is introduced in this paper to provide additional insight often not found in current LME volumetric analyses. SETTING AND PARTICIPANTS: Longitudinal neuroimaging case study of ageing was analysed in this research on 260 participants diagnosed as either healthy controls (HC), mild cognitive impaired (MCI) or AD. Bayesian LME models for the ventricle and hippocampus regions were used to: (1) estimate how the volumes of these regions change over time by diagnosis, (2) identify high-risk non-AD individuals with AD like degeneration and (3) determine probabilistic trajectories of diagnosis groups over age. RESULTS: We observed (1) large differences in the average rate of change of volume for the ventricle and hippocampus regions between diagnosis groups, (2) high-risk individuals who had progressed from HC to MCI and displayed similar rates of deterioration as AD counterparts, and (3) critical time points which indicate where deterioration of regions begins to diverge between the diagnosis groups. CONCLUSIONS: To the best of our knowledge, this is the first application of Bayesian LME models to neuroimaging data which provides inference on a population and individual level in the AD field. The application of a Bayesian LME framework allows for additional information to be extracted from longitudinal studies. This provides health professionals with valuable information of neurodegeneration stages, and a potential to provide a better understanding of disease pathology.