We present a new method for automatic brain extraction on structural magnetic resonance images, based on a multi-atlas registration framework.
Our method addresses fundamental challenges of ...multi-atlas approaches. To overcome the difficulties arising from the variability of imaging characteristics between studies, we propose a study-specific template selection strategy, by which we select a set of templates that best represent the anatomical variations within the data set. Against the difficulties of registering brain images with skull, we use a particularly adapted registration algorithm that is more robust to large variations between images, as it adaptively aligns different regions of the two images based not only on their similarity but also on the reliability of the matching between images. Finally, a spatially adaptive weighted voting strategy, which uses the ranking of Jacobian determinant values to measure the local similarity between the template and the target images, is applied for combining coregistered template masks.
The method is validated on three different public data sets and obtained a higher accuracy than recent state-of-the-art brain extraction methods. Also, the proposed method is successfully applied on several recent imaging studies, each containing thousands of magnetic resonance images, thus reducing the manual correction time significantly.
The new method, available as a stand-alone software package for public use, provides a robust and accurate brain extraction tool applicable for both clinical use and large population studies.
Automatic segmentation of brain anatomy has been a key processing step in quantitative neuroimaging analyses. An extensive body of literature has relied on Freesurfer segmentations. Yet, in recent ...years, the multi-atlas segmentation framework has consistently obtained results with superior accuracy in various evaluations. We compared brain anatomy segmentations from Freesurfer, which uses a single probabilistic atlas strategy, against segmentations from Multi-atlas region Segmentation utilizing Ensembles of registration algorithms and parameters and locally optimal atlas selection (MUSE), one of the leading ensemble-based methods that calculates a consensus segmentation through fusion of anatomical labels from multiple atlases and registrations. The focus of our evaluation was twofold. First, using manual ground-truth hippocampus segmentations, we found that Freesurfer segmentations showed a bias towards over-segmentation of larger hippocampi, and under-segmentation in older age. This bias was more pronounced in Freesurfer-v5.3, which has been used in multiple previous studies of aging, while the effect was mitigated in more recent Freesurfer-v6.0, albeit still present. Second, we evaluated inter-scanner segmentation stability using same day scan pairs from ADNI acquired on 1.5T and 3T scanners. We also found that MUSE obtains more consistent segmentations across scanners compared to Freesurfer, particularly in the deep structures.
Abstract
Past work on relatively small, single-site studies using regional volumetry, and more recently machine learning methods, has shown that widespread structural brain abnormalities are ...prominent in schizophrenia. However, to be clinically useful, structural imaging biomarkers must integrate high-dimensional data and provide reproducible results across clinical populations and on an individual person basis. Using advanced multi-variate analysis tools and pooled data from case–control imaging studies conducted at 5 sites (941 adult participants, including 440 patients with schizophrenia), a neuroanatomical signature of patients with schizophrenia was found, and its robustness and reproducibility across sites, populations, and scanners, was established for single-patient classification. Analyses were conducted at multiple scales, including regional volumes, voxelwise measures, and complex distributed patterns. Single-subject classification was tested for single-site, pooled-site, and leave-site-out generalizability. Regional and voxelwise analyses revealed a pattern of widespread reduced regional gray matter volume, particularly in the medial prefrontal, temporolimbic and peri-Sylvian cortex, along with ventricular and pallidum enlargement. Multivariate classification using pooled data achieved a cross-validated prediction accuracy of 76% (AUC = 0.84). Critically, the leave-site-out validation of the detected schizophrenia signature showed accuracy/AUC range of 72–77%/0.73–0.91, suggesting a robust generalizability across sites and patient cohorts. Finally, individualized patient classifications displayed significant correlations with clinical measures of negative, but not positive, symptoms. Taken together, these results emphasize the potential for structural neuroimaging data to provide a robust and reproducible imaging signature of schizophrenia. A web-accessible portal is offered to allow the community to obtain individualized classifications of magnetic resonance imaging scans using the methods described herein.
Brain extraction, or skull-stripping, is an essential pre-processing step in neuro-imaging that has a direct impact on the quality of all subsequent processing and analyses steps. It is also a key ...requirement in multi-institutional collaborations to comply with privacy-preserving regulations. Existing automated methods, including Deep Learning (DL) based methods that have obtained state-of-the-art results in recent years, have primarily targeted brain extraction without considering pathologically-affected brains. Accordingly, they perform sub-optimally when applied on magnetic resonance imaging (MRI) brain scans with apparent pathologies such as brain tumors. Furthermore, existing methods focus on using only T1-weighted MRI scans, even though multi-parametric MRI (mpMRI) scans are routinely acquired for patients with suspected brain tumors. In this study, we present a comprehensive performance evaluation of recent deep learning architectures for brain extraction, training models on mpMRI scans of pathologically-affected brains, with a particular focus on seeking a practically-applicable, low computational footprint approach, generalizable across multiple institutions, further facilitating collaborations. We identified a large retrospective multi-institutional dataset of n=3340 mpMRI brain tumor scans, with manually-inspected and approved gold-standard segmentations, acquired during standard clinical practice under varying acquisition protocols, both from private institutional data and public (TCIA) collections. To facilitate optimal utilization of rich mpMRI data, we further introduce and evaluate a novel ‘‘modality-agnostic training’’ technique that can be applied using any available modality, without need for model retraining. Our results indicate that the modality-agnostic approach11Publicly available source code: https://github.com/CBICA/BrainMaGe obtains accurate results, providing a generic and practical tool for brain extraction on scans with brain tumors.
•Accurate brain extraction on MRI scans in presence of diffuse gliomas is critical.•Comprehensive evaluation of prominent deep learning architectures, BET & FreeSurfer.•Multi-institutional data to test generalizability and to facilitate collaborations.•A novel “modality-agnostic” strategy to promote widespread application.
•Greater annual rates of memory decline were associated with greater volume loss in multiple temporal and occipital regions.•Decline in verbal fluency was associated with greater ventricular size and ...decline in frontal, temporal, and parietal regions.•Decline in visuospatial ability was associated with volume loss in 3 temporal and parietal regions.•Declines in Trail-Making Test-A were associated with volume loss in 4 temporal and parietal regions.•Declines in Trail-Making Test-B were associated with ventricular size and volume loss in 10 regions.
Investigation of relationships between age-related changes in regional brain volumes and changes in domain-specific cognition could provide insights into the neural underpinnings of individual differences in cognitive aging. Domain-specific cognition (memory, verbal fluency, visuospatial ability) and tests of executive function and attention (Trail-Making Test Part A and B) and 47 brain volumes of interest (VOIs) were assessed in 836 Baltimore Longitudinal Study of Aging participants with mean follow-up of 4.1 years (maximum 23.1 years). To examine the correlation between changes in domain-specific cognition and changes in brain volumes, we used bivariate linear mixed effects models with unstructured variance-covariance structure to estimate longitudinal trajectories for each variable of interest and correlations among the random effects of these measures. Higher annual rates of memory decline were associated with greater volume loss in 14 VOIs primarily within the temporal and occipital lobes. Verbal fluency decline was associated with greater ventricular enlargement and volume loss in 24 VOIs within the frontal, temporal, and parietal lobes. Decline in visuospatial ability was associated with volume loss in 3 temporal and parietal VOIs. Declines on the attentional test were associated with volume loss in 4 VOIs located within temporal and parietal lobes. Greater declines on the executive function test were associated with greater ventricular enlargement and volume loss in 10 frontal, parietal, and temporal VOIs. Our findings highlight domain-specific patterns of regional brain atrophy that may contribute to individual differences in cognitive aging.
The neuroanatomic basis for cognitive impairment in chronic kidney disease (CKD) is incompletely characterized. We performed advanced quantitative structural magnetic resonance imaging (MRI) to ...determine whether CKD affects brain structure and whether poorer neurocognitive performance in CKD is associated with structural brain differences.
Cross-sectional.
85 individuals with CKD stages 2 to 5 and 63 healthy controls, aged 8 to 25 years
CKD versus control, estimated glomerular filtration rate (eGFR), and kidney transplant status were analyzed as predictors of MRI findings. MRI volumes in 19 prespecified regions of gray matter (GM), white matter (WM), and cerebrospinal fluid were analyzed as predictors of neurocognitive performance (median z scores) in 7 prespecified domains.
19 prespecified brain regions of interest (ROIs) in 7 prespecified domains. Neurocognitive performance in 7 prespecified domains.
ROI volumes were compared in CKD versus controls using unadjusted t tests and analysis of covariance (ANCOVA). Associations of ROI volumes with eGFR and kidney transplant status in participants with CKD were analyzed using ANCOVA and linear regression. Associations of neurocognitive performance and ROI volumes were analyzed by linear regression.
Participants with CKD had lower whole-brain, cortical, and left parietal GM volumes than controls in unadjusted analyses, but no differences were found in adjusted analysis. In participants with CKD, lower eGFR was associated with higher WM volume in whole-brain (P=0.05) and frontal (P=0.04) ROIs, but differences were not significant after multiple comparisons correction. Kidney transplant recipients had lower GM volumes in whole-brain (P=0.01; Q=0.06), frontal (P=0.02; Q=0.08), and left and right parietal (P=0.01; Q=0.06; and P=0.03; Q=0.1) ROIs and higher whole-brain WM volume (P=0.04; Q=0.1). Neurocognitive performance in the CKD group was not associated with ROI volumes.
Unable to assess changes in brain structure and kidney function over time; analysis limited to prespecified ROIs and neurocognitive domains.
CKD in children and young adults may be associated with lower GM and higher WM volumes in some ROIs. Differences were relatively subtle in the CKD group as a whole, but were more prominent in recipients of a kidney transplant. However, neurocognitive performance was not explained by differences in brain ROI volumes, suggesting a functional rather than structural basis for neurocognitive impairment in CKD.
While Alzheimer disease (AD) and vascular dementia (VaD) may be accelerated by hypercholesterolemia, the mechanisms underlying this association are unclear. We tested whether dysregulation of ...cholesterol catabolism, through its conversion to primary bile acids (BAs), was associated with dementia pathogenesis.
We used a 3-step study design to examine the role of the primary BAs, cholic acid (CA), and chenodeoxycholic acid (CDCA) as well as their principal biosynthetic precursor, 7α-hydroxycholesterol (7α-OHC), in dementia. In Step 1, we tested whether serum markers of cholesterol catabolism were associated with brain amyloid accumulation, white matter lesions (WMLs), and brain atrophy. In Step 2, we tested whether exposure to bile acid sequestrants (BAS) was associated with risk of dementia. In Step 3, we examined plausible mechanisms underlying these findings by testing whether brain levels of primary BAs and gene expression of their principal receptors are altered in AD. Step 1: We assayed serum concentrations CA, CDCA, and 7α-OHC and used linear regression and mixed effects models to test their associations with brain amyloid accumulation (N = 141), WMLs, and brain atrophy (N = 134) in the Baltimore Longitudinal Study of Aging (BLSA). The BLSA is an ongoing, community-based cohort study that began in 1958. Participants in the BLSA neuroimaging sample were approximately 46% male with a mean age of 76 years; longitudinal analyses included an average of 2.5 follow-up magnetic resonance imaging (MRI) visits. We used the Alzheimer's Disease Neuroimaging Initiative (ADNI) (N = 1,666) to validate longitudinal neuroimaging results in BLSA. ADNI is an ongoing, community-based cohort study that began in 2003. Participants were approximately 55% male with a mean age of 74 years; longitudinal analyses included an average of 5.2 follow-up MRI visits. Lower serum concentrations of 7α-OHC, CA, and CDCA were associated with higher brain amyloid deposition (p = 0.041), faster WML accumulation (p = 0.050), and faster brain atrophy mainly (false discovery rate FDR p = <0.001-0.013) in males in BLSA. In ADNI, we found a modest sex-specific effect indicating that lower serum concentrations of CA and CDCA were associated with faster brain atrophy (FDR p = 0.049) in males.Step 2: In the Clinical Practice Research Datalink (CPRD) dataset, covering >4 million registrants from general practice clinics in the United Kingdom, we tested whether patients using BAS (BAS users; 3,208 with ≥2 prescriptions), which reduce circulating BAs and increase cholesterol catabolism, had altered dementia risk compared to those on non-statin lipid-modifying therapies (LMT users; 23,483 with ≥2 prescriptions). Patients in the study (BAS/LMT) were approximately 34%/38% male and with a mean age of 65/68 years; follow-up time was 4.7/5.7 years. We found that BAS use was not significantly associated with risk of all-cause dementia (hazard ratio (HR) = 1.03, 95% confidence interval (CI) = 0.72-1.46, p = 0.88) or its subtypes. We found a significant difference between the risk of VaD in males compared to females (p = 0.040) and a significant dose-response relationship between BAS use and risk of VaD (p-trend = 0.045) in males.Step 3: We assayed brain tissue concentrations of CA and CDCA comparing AD and control (CON) samples in the BLSA autopsy cohort (N = 29). Participants in the BLSA autopsy cohort (AD/CON) were approximately 50%/77% male with a mean age of 87/82 years. We analyzed single-cell RNA sequencing (scRNA-Seq) data to compare brain BA receptor gene expression between AD and CON samples from the Religious Orders Study and Memory and Aging Project (ROSMAP) cohort (N = 46). ROSMAP is an ongoing, community-based cohort study that began in 1994. Participants (AD/CON) were approximately 56%/36% male with a mean age of 85/85 years. In BLSA, we found that CA and CDCA were detectable in postmortem brain tissue samples and were marginally higher in AD samples compared to CON. In ROSMAP, we found sex-specific differences in altered neuronal gene expression of BA receptors in AD. Study limitations include the small sample sizes in the BLSA cohort and likely inaccuracies in the clinical diagnosis of dementia subtypes in primary care settings.
We combined targeted metabolomics in serum and amyloid positron emission tomography (PET) and MRI of the brain with pharmacoepidemiologic analysis to implicate dysregulation of cholesterol catabolism in dementia pathogenesis. We observed that lower serum BA concentration mainly in males is associated with neuroimaging markers of dementia, and pharmacological lowering of BA levels may be associated with higher risk of VaD in males. We hypothesize that dysregulation of BA signaling pathways in the brain may represent a plausible biologic mechanism underlying these results. Together, our observations suggest a novel mechanism relating abnormalities in cholesterol catabolism to risk of dementia.
•Multiscale functional connectivity pattern of the aging brain were learned from a large-scale multisite fMRI datasets.•A machine learning model built on the multiscale functional connectivity ...measures achieved accurate brain age prediction.•Functional connectivity measures at multiple scales were more informative than those at any single scale for the brain age prediction.•Data harmonization significantly improved the brain age prediction performance.
To learn multiscale functional connectivity patterns of the aging brain, we built a brain age prediction model of functional connectivity measures at seven scales on a large fMRI dataset, consisting of resting-state fMRI scans of 4186 individuals with a wide age range (22 to 97 years, with an average of 63) from five cohorts. We computed multiscale functional connectivity measures of individual subjects using a personalized functional network computational method, harmonized the functional connectivity measures of subjects from multiple datasets in order to build a functional brain age model, and finally evaluated how functional brain age gap correlated with cognitive measures of individual subjects. Our study has revealed that functional connectivity measures at multiple scales were more informative than those at any single scale for the brain age prediction, the data harmonization significantly improved the brain age prediction performance, and the data harmonization in the functional connectivity measures' tangent space worked better than in their original space. Moreover, brain age gap scores of individual subjects derived from the brain age prediction model were significantly correlated with clinical and cognitive measures. Overall, these results demonstrated that multiscale functional connectivity patterns learned from a large-scale multi-site rsfMRI dataset were informative for characterizing the aging brain and the derived brain age gap was associated with cognitive and clinical measures.
CKD has been linked with cognitive deficits and affective disorders in multiple studies. Analysis of structural and functional neuroimaging in adults and children with kidney disease may provide ...additional important insights into the pathobiology of this relationship. This paper comprehensively reviews neuroimaging studies in both children and adults. Major databases (PsychLit, MEDLINE, WorldCat, ArticleFirst, PubMed, Ovid MEDLINE) were searched using consistent search terms, and studies published between 1975 and 2012 were included if their samples focused on CKD as the primary disease process. Exclusion criteria included case reports, chapters, and review articles. This systematic process yielded 43 studies for inclusion (30 in adults, 13 in children). Findings from this review identified several clear trends: (1) presence of cerebral atrophy and cerebral density changes in patients with CKD; (2) cerebral vascular disease, including deep white matter hyperintensities, white matter lesions, cerebral microbleeds, silent cerebral infarction, and cortical infarction, in patients with CKD; and (3) similarities in regional cerebral blood flow between patients with CKD and those with affective disorders. These findings document the importance of neuroimaging procedures in understanding the effect of CKD on brain structure, function, and associated behaviors. Results provide a developmental linkage between childhood and adulthood, with respect to the effect of CKD on brain functioning across the lifespan, with strong implications for a cerebrovascular mechanism contributing to this developmental linkage. Use of neuroimaging methods to corroborate manifest neuropsychological deficits or perhaps to indicate preventive actions may prove useful to individuals with CKD.