Anomaly detection in multimodal MRI identifies rare individual phenotypes among 20,000 brains

The UK Biobank (UKB) is a large-scale epidemiological study and its imaging component focuses on the pre-symptomatic participants. Given its large sample size, rare imaging phenotypes within this unique cohort are of interest, as they are often clinically relevant and could be informative for discovering new processes and mechanisms. Identifying these rare phenotypes is often referred to as "anomaly detection", or "outlier detection". However, anomaly detection in neuroimaging has usually been applied in a supervised or semi-supervised manner for clinically defined cohorts of relatively small size. There has been much less work using anomaly detection on large unlabeled cohorts like the UKB. Here we developed a two-level anomaly screening methodology to systematically identify anomalies from ~19,000 UKB subjects. The same method was also applied to ~1,000 young healthy subjects from the Human Connectome Project (HCP). In primary screening, using ventricular, white matter, and gray matter-based imaging phenotypes derived from multimodal MRI, every subject was parameterized with an anomaly score per phenotype to quantitate the degree of abnormality. These anomaly scores were highly robust. Anomaly score distributions of the UKB cohort were all more outlier-prone than the HCP cohort of young adults. The approach enabled the assessments of test-retest reliability via the anomaly scores, which ranged from excellent reliability for ventricular volume, white matter lesion volume, and fractional anisotropy, to good reliability for mean diffusivity and cortical thickness. In secondary screening, the anomalies due to data collection/processing errors were eliminated. A subgroup of the remaining anomalies were radiologically reviewed, and a substantial percentage of them (UKB: 90.1%; HCP: 42.9%) had various brain pathologies such as masses, cysts, white matter lesions, infarcts, encephalomalacia, or prominent sulci. The remaining anomalies of the subgroup had unexplained causes and would be interesting for follow-up. Finally, we show that anomaly detection applied to resting-state functional connectivity did not identify any reliable anomalies, which was attributed to the confounding effects of brain-wide signal variation. Together, this study establishes an unsupervised framework for investigating rare individual imaging phenotypes within large heterogeneous cohorts.

Download Full-text

The relationship between white matter fiber damage and gray matter perfusion in large-scale functionally defined networks in multiple sclerosis

Multiple Sclerosis Journal ◽

10.1177/1352458517691149 ◽

2017 ◽

Vol 23 (14) ◽

pp. 1884-1892 ◽

Cited By ~ 4

Author(s):

Ashley Y Ma ◽

Rita C Vitorino ◽

Seyed-Parsa Hojjat ◽

Alannah D Mulholland ◽

Liying Zhang ◽

...

Keyword(s):

Multiple Sclerosis ◽

White Matter ◽

Gray Matter ◽

Large Scale ◽

Progressive Multiple Sclerosis ◽

Functional Networks ◽

Lesion Volume ◽

Fiber Damage ◽

Damage Extent ◽

The Relationship

Background: Recent studies utilizing perfusion as a surrogate of cortical integrity show promise for overall cognition, but the association between white matter (WM) damage and gray matter (GM) integrity in specific functional networks is not previously studied. Objective: To investigate the relationship between WM fiber integrity and GM node perfusion within six functional networks of relapsing-remitting multiple sclerosis (RRMS) and secondary progressive multiple sclerosis (SPMS) patients. Methods: Magnetic resonance imaging (MRI) and neurocognitive testing were performed on 19 healthy controls (HC), 39 RRMS, and 45 SPMS patients. WM damage extent and severity were quantified with T2-hyper/T1-hypointense (T2h/T1h) lesion volume and degree of perfusion reduction in lesional and normal-appearing white matter (NAWM), respectively. A two-step linear regression corrected for confounders was employed. Results: Cognitive impairment was present in 20/39 (51%) RRMS and 25/45 (53%) SPMS patients. GM node perfusion was associated with WM fiber damage severity (WM hypoperfusion) within each network—including both NAWM ( R2 = 0.67–0.89, p < 0.0001) and T2h ( R2 = 0.39–0.62, p < 0.0001) WM regions—but was not significantly associated ( p > 0.01) with WM fiber damage extent (i.e. T2h/T1h lesion volumes). Conclusion: Overall, GM node perfusion was associated with severity rather than extent of WM network damage, supporting a primary etiology of GM hypoperfusion.

Download Full-text

Sensory processing sensitivity and axonal microarchitecture: Identifying brain structural characteristics for behavior

10.1101/2021.11.13.468491 ◽

2021 ◽

Author(s):

Szabolcs David ◽

Lucy L Brown ◽

Anneriet M Heemskerk ◽

Elaine Aron ◽

Alexander Leemans ◽

...

Keyword(s):

Diffusion Tensor Imaging ◽

White Matter ◽

Sensory Processing ◽

Diffusion Tensor ◽

Structural Characteristics ◽

Inferior Frontal Gyrus ◽

Mean Diffusivity ◽

Human Connectome Project ◽

Sensory Processing Sensitivity ◽

The Right

Previously, researchers used functional MRI to identify regional brain activations associated with sensory processing sensitivity (SPS), a proposed normal phenotype trait. To further validate SPS as a behavioral entity, to characterize it anatomically, and to test the usefulness in psychology of methodologies that assess axonal properties, the present study correlated SPS proxy questionnaire scores (adjusted for neuroticism) with diffusion tensor imaging measures. Participants (n=408) from the Young Adult Human Connectome Project that are free of neurologic and psychiatric disorders were investigated. We computed mean diffusivity (MD), radial diffusivity (RD), axial diffusivity (AD) and fractional anisotropy (FA). A voxelwise, exploratory analysis showed that MD and RD correlated positively with SPS proxy scores in the right and left subcallosal and anterior ventral cingulum bundle, and the right forceps minor of the corpus callosum (peak Cohens D effect size = 0.269). Further analyses showed correlations throughout the entire right and left ventromedial prefrontal cortex, including the superior longitudinal fasciculus, inferior fronto-occipital fasciculus, uncinate and arcuate fasciculus. These prefrontal regions are generally involved in emotion, reward and social processing. FA was negatively correlated with SPS proxy scores in white matter of the right premotor/motor/somatosensory/supramarginal gyrus regions, which are associated with empathy, theory of mind, primary and secondary somatosensory processing. Region of interest (ROI) analysis, based-on previous fMRI results and Freesurfer atlas-defined areas, showed small effect sizes, (+0.151 to -0.165) in white matter of the precuneus and inferior frontal gyrus. Other ROI effects were found in regions of the dorsal and ventral visual pathways and primary auditory cortex. The results reveal that in a large, diverse group of participants axonal microarchitectural differences can be identified with SPS traits that are subtle and in the range of typical behavior. The results suggest that the heightened sensory processing in people who show SPS may be influenced by the microstructure of white matter in specific neocortical regions. Although previous fMRI studies had identified most of these general neocortical regions, the DTI results put a new focus on brain areas related to attention and cognitive flexibility, empathy, emotion and low-level sensory processing, as in the primary sensory cortex. Psychological trait characterization may benefit from diffusion tensor imaging methodology by identifying influential brain systems for traits.

Download Full-text

DT MRI microstructural cortical lesion damage does not explain cognitive impairment in MS

Multiple Sclerosis Journal ◽

10.1177/1352458516689147 ◽

2017 ◽

Vol 23 (14) ◽

pp. 1918-1928 ◽

Cited By ~ 6

Author(s):

Paolo Preziosa ◽

Elisabetta Pagani ◽

Maria E Morelli ◽

Massimiliano Copetti ◽

Vittorio Martinelli ◽

...

Keyword(s):

Cognitive Impairment ◽

White Matter ◽

Diffusion Tensor ◽

Gray Matter Volume ◽

Three Dimensional ◽

Mean Diffusivity ◽

Cortical Lesion ◽

Lesion Volume ◽

Normal Appearing White Matter ◽

Tissue Abnormalities

Objective: We combined double inversion recovery (DIR) and diffusion tensor (DT) magnetic resonance imaging (MRI) to quantify the severity of cortical lesion (CL) microstructural tissue abnormalities in a large cohort of relapse-onset multiple sclerosis (MS) patients and its contribution to cognitive dysfunction. Methods: DIR, DT, dual-echo, and three-dimensional (3D) T1-weighted scans were acquired from 149 MS patients and 40 controls. Cognitively impaired (CI) patients had ⩾2 abnormal neuropsychological tests. Diffusivity values in CLs, cortex, white matter (WM) lesions, and normal-appearing (NA) WM were assessed. Predictors of cognitive impairment were identified using a random forest analysis. Results: Compared to controls, MS patients had lower normalized brain volume (NBV), gray matter volume (GMV), WM volume, lower fractional anisotropy (FA), and higher mean diffusivity in cortex and normal-appearing white matter (NAWM). A total of 44 (29.5%) patients were CI. Compared to cognitively preserved (CP), CI patients had higher T2 WM lesion volume (LV), lower NBV and GMV, and more severe diffusivity abnormalities in WM lesions, cortex, and NAWM. CL measures did not differ between CI and CP patients. Cortex FA, age, disease duration, T2 WM LV, and GMV best predicted MS-related cognitive impairment (C-statistic = 0.88). Conclusion: “Diffuse” GM and NAWM damage and WM lesions, rather than intrinsic CL damage, contribute to cognitive impairment in MS.

Download Full-text

Neurobiological underpinnings of rapid white matter plasticity during intensive reading instruction

10.1101/2020.05.28.122499 ◽

2020 ◽

Author(s):

Elizabeth Huber ◽

Aviv Mezer ◽

Jason D. Yeatman

Keyword(s):

White Matter ◽

Reading Instruction ◽

Diffusion Mri ◽

Large Scale ◽

Mean Diffusivity ◽

Quantitative Mri ◽

Diffusion Kurtosis Imaging ◽

Control Group ◽

Intensive Reading ◽

Diffusion Kurtosis

AbstractHuman white matter is remarkably plastic. Yet it is challenging to infer the biological underpinnings of this plasticity using non-invasive measurements like diffusion MRI. Here we capitalize on metrics derived from diffusion kurtosis imaging (DKI) to interpret previously reported changes in mean diffusivity throughout the white matter during an 8-week, intensive reading intervention. We then use an independent quantitative MRI measurement of R1 (1/T1 relaxation time) in the same white matter regions; since R1 closely tracks variation in myelin content, it provides complementary information about white matter microstructure. Behavioral measures, multi-shell diffusion MRI data, and quantitative T1 data were collected at regular intervals during the intervention in a group of 33 children with reading difficulties (7-12 years old), and over the same period in an age-matched non-intervention control group. Changes in DKI parameters modeled over the intervention were consistent with increased hindrance in the extra-axonal space, rather than a large-scale change in axon density and/or myelination. Supporting this interpretation, analysis of R1 values did not suggest a change in myelin, although R1 estimates were correlated with individual differences in reading skill. Together, these results suggest that large-scale changes in diffusivity observed over a short timescale during an intensive educational experience are most likely to reflect changes occurring in the extra-axonal space, in line with recent work highlighting the role of glial cells in experience-dependent plasticity and learning.

Download Full-text

A Measure of Whole-Brain White Matter Topography

10.1101/783274 ◽

2019 ◽

Author(s):

Junyan Wang ◽

Yonggang Shi

Keyword(s):

White Matter ◽

Diffusion Mri ◽

Large Scale ◽

Uk Biobank ◽

Reliable Measure ◽

Whole Brain ◽

Brain White Matter ◽

Human Connectome Project ◽

Natural Coordinate System ◽

Natural Coordinate

The unprecedentedly high-quality large-scale brain imaging datasets, from such as the Human Connectome Project (HCP) and UK-Biobank, provide a unique opportunity for measuring the white matter topography of the human brain. By leveraging the multi-shell diffusion MRI data from the original young adult HCP, we systematically develop a reliable measure of the whole-brain white matter topography, and we coin it topographic vector. As the main result, we find that the three most dominant dimensions of the topographic vectors strongly and linearly correlate with the coordinates of the corresponding streamlines of the whole-brain tractograms. Our results support the earlier prescient hypothesis that brain development follows a “base-plan” established by three (main) chemotactic gradients of early embryogenesis, and they implicate that the whole brain white matter tracts can be represented by vectors of a natural coordinate system.

Download Full-text

Functional and diffusion MRI reveal the functional and structural basis of infants’ noxious-evoked brain activity

10.1101/2020.04.28.065730 ◽

2020 ◽

Author(s):

Luke Baxter ◽

Fiona Moultrie ◽

Sean Fitzgibbon ◽

Marianne Aspbury ◽

Roshni Mansfield ◽

...

Keyword(s):

White Matter ◽

Pain Management ◽

Resting State ◽

Diffusion Mri ◽

Brain Activity ◽

Mean Diffusivity ◽

Structural Basis ◽

Noxious Stimulation ◽

State Activity ◽

Human Connectome Project

AbstractUnderstanding the neurophysiology underlying pain perception in infants is central to improving early life pain management. In this multimodal MRI study, we use resting-state functional and white matter diffusion MRI to investigate individual variability in infants’ noxious-evoked brain activity. In an 18-infant nociception-paradigm dataset, we show it is possible to predict infants’ cerebral haemodynamic responses to experimental noxious stimulation using their resting-state activity across nine networks from a separate stimulus-free scan. In an independent 215-infant Developing Human Connectome Project dataset, we use this resting-state-based prediction model to generate noxious responses. We identify a significant correlation between these predicted noxious responses and infants’ white matter mean diffusivity, and this relationship is subsequently confirmed within our nociception-paradigm dataset. These findings reveal that a newborn infant’s pain-related brain activity is tightly coupled to both their spontaneous resting-state activity and underlying white matter microstructure. This work provides proof-of-concept that knowledge of an infant’s functional and structural brain architecture could be used to predict pain responses, informing infant pain management strategies and facilitating evidence-based personalisation of care.

Download Full-text

Distinct white matter alterations following severe stroke

Neurology ◽

10.1212/wnl.0000000000003843 ◽

2017 ◽

Vol 88 (16) ◽

pp. 1546-1555 ◽

Cited By ~ 19

Author(s):

Roza M. Umarova ◽

Lena Beume ◽

Marco Reisert ◽

Christoph P. Kaller ◽

Stefan Klöppel ◽

...

Keyword(s):

White Matter ◽

Network Architecture ◽

Large Scale ◽

Diffusion Tensor ◽

Neuropsychological Testing ◽

Mean Diffusivity ◽

Spatial Neglect ◽

Association Cortex ◽

Structural Reorganization ◽

Domain Specific

Objective:To distinguish white matter remodeling directly induced by stroke lesion from that evoked by remote network dysfunction, using spatial neglect as a model.Methods:We examined 24 visual neglect/extinction patients and 17 control patients combining comprehensive analyses of diffusion tensor metrics and global fiber tracking with neuropsychological testing in the acute (6.3 ± 0.5 days poststroke) and chronic (134 ± 7 days poststroke) stroke phases.Results:Compared to stroke controls, patients with spatial neglect/extinction displayed longitudinal white matter alterations with 2 defining signatures: (1) perilesional degenerative changes characterized by congruently reduced fractional anisotropy and increased radial diffusivity (RD), axial diffusivity, and mean diffusivity, all suggestive of direct axonal damage by lesion and therefore nonspecific for impaired attention network and (2) transneuronal changes characterized by an increased RD in contralesional frontoparietal and bilateral occipital connections, suggestive of primary periaxonal involvement; these changes were distinctly related to the degree of unrecovered neglect symptoms in chronic stroke, hence emerging as network-specific alterations.Conclusions:The present data show how stroke entails global alterations of lesion-spared network architecture over time. Sufficiently large lesions of widely interconnected association cortex induce distinct, large-scale structural reorganization in domain-specific network connections. Besides their relevance to unrecovered domain-specific symptoms, these effects might also explain mechanisms of domain-general deficits in stroke patients, pointing to potential targets for therapeutic intervention.

Download Full-text

Abstract TP441: Fornix Microstructure Correlates with Post-stroke Memory Impairment Independently of Lesion Volume

Stroke ◽

10.1161/str.48.suppl_1.tp441 ◽

2017 ◽

Vol 48 (suppl_1) ◽

Author(s):

Paul Wright ◽

Nicola J Ray ◽

Michael J O’Sullivan

Keyword(s):

White Matter ◽

Recognition Memory ◽

Temporal Lobe ◽

Memory Impairment ◽

Unmet Need ◽

Mean Diffusivity ◽

Cognitive Outcome ◽

First Year ◽

Lesion Volume ◽

Post Stroke

Memory impairment is common and a cause of unmet need after stroke. One third of patients recover spontaneously over the first year. Others develop later decline. The mechanism of recovery and the contribution of comorbid neurodegenerative pathology are not well understood. Data from older adults and Mild Cognitive Impairment suggest preserved memory depends on reorganisation of function between the fornix and other temporal lobe white matter tracts. The STRATEGIC study investigates memory over the first year post-stroke. The current analysis uses baseline data to investigate the contributions of initial infarct and white matter status to initial memory impairment. Patients (n = 21; 5 lacunar, 12 MCA, 4 PCA; 11 left hemisphere, 10 right) performed a cognitive battery and had diffusion-weighted MRI at 30-95 days after stroke onset (median = 64 days). Healthy controls (n = 33) provided the same measures. We reconstructed the fornix using HARDI-based deterministic tractography. Lesions were drawn manually on FLAIR images and direct injury of the fornix was excluded. We tested free recall using a delayed verbal memory test and recognition memory using a face recognition test with trialwise confidence ratings. Recall was markedly lower in patients than controls (50% vs 71%, p < 0.001) but did not correlate with age, lesion volume or fornix integrity in patients. Recognition memory was unimpaired in patients. For confidently remembered items, performance correlated with fornix mean diffusivity in both controls (r = -0.344, p < 0.05) and patients (r = -0.544, p < 0.05). Controlling for age eliminated the correlation in controls but in patients there was a correlation independent of age and lesion volume (r = -0.531, p < 0.05). These findings demonstrate that preserved recognition memory after stroke depends on the status of the fornix and is independent of infarct volume. Compromise of fornix structure from pre-existing neurodegeneration may be a predictor of poor cognitive outcome after stroke. Verbal recall in patients was independent of fornix integrity. Investigation of other temporal lobe pathways may reveal reorganisation underpinning partially preserved recall in some patients.

Download Full-text

Microstructural damage of normal-appearing white matter in subcortical ischemic vascular dementia is associated with Montreal Cognitive Assessment scores

Journal of International Medical Research ◽

10.1177/0300060519863520 ◽

2019 ◽

Vol 47 (11) ◽

pp. 5723-5731

Author(s):

Zhangyuan Liao ◽

Chun Dang ◽

Meijie Li ◽

Yali Bu ◽

Ranran Han ◽

...

Keyword(s):

White Matter ◽

Vascular Dementia ◽

Cognitive Assessment ◽

Mean Diffusivity ◽

Montreal Cognitive Assessment ◽

Lesion Volume ◽

Clinical Dementia Rating ◽

Assessment Scores ◽

Normal Appearing White Matter ◽

Brain White Matter

Objectives This study was performed to determine whether multimodal biomarkers are more strongly associated with the Montreal Cognitive Assessment (MoCA) scores compared with the Mini-Mental State Examination (MMSE) scores, and whether they are correlated with the Clinical Dementia Rating (CDR) in patients with subcortical ischemic vascular dementia (SIVD). Methods Patients diagnosed with SIVD were enrolled. Peripheral blood hypersensitive C-reactive protein, white matter lesion volume (WMLV), fractional anisotropy (FA)/mean diffusivity (MD) of whole brain white matter (WBWM), and normal-appearing white matter (NAWM) were measured and correlated with MMSE, MoCA, and CDR scores. Results Bivariate analyses of data from 48 included patients revealed that both MoCA and MMSE were significantly associated with age, education, and FA of NAWM. Only MD of NAWM was correlated with MoCA scores. In partial correlation analysis adjusted for demographic and neuroimaging variables, MD/FA of NAWM and the MoCA scores were significantly correlated. FA/MD of NAWM had a modest trend toward a correlation with the CDR, but it was not significant. Conclusions In the patients with SIVD, FA/MD of NAWM were more strongly related to MoCA scores compared with MMSE scores.

Download Full-text

Reducing variability in along-tract analysis with diffusion profile realignment

10.1101/541920 ◽

2019 ◽

Author(s):

Samuel St-Jean ◽

Maxime Chamberland ◽

Max A. Viergever ◽

Alexander Leemans

Keyword(s):

White Matter ◽

Fractional Anisotropy ◽

Mean Diffusivity ◽

Region Of Interest ◽

Diffusion Profile ◽

Fixed Number ◽

Fiber Density ◽

Human Connectome Project ◽

Student’S T

AbstractDiffusion weighted magnetic resonance imaging (dMRI) provides a non invasive virtual reconstruction of the brain’s white matter structures through tractography. Analyzing dMRI measures along the trajectory of white matter bundles can provide a more specific investigation than considering a region of interest or tract-averaged measurements. However, performing group analyses with this along-tract strategy requires correspondence between points of tract pathways across subjects. This is usually achieved by creating a new common space where the representative streamlines from every subject are resampled to the same number of points. If the underlying anatomy of some subjects was altered due to, e.g., disease or developmental changes, such information might be lost by resampling to a fixed number of points. In this work, we propose to address the issue of possible misalignment, which might be present even after resampling, by realigning the representative streamline of each subject in this 1D space with a new method, coined diffusion profile realignment (DPR). Experiments on synthetic datasets show that DPR reduces the coefficient of variation for the mean diffusivity, fractional anisotropy and apparent fiber density when compared to the unaligned case. Using 100 in vivo datasets from the human connectome project, we simulated changes in mean diffusivity, fractional anisotropy and apparent fiber density. Pairwise Student’s t-tests between these altered subjects and the original subjects indicate that regional changes are identified after realignment with the DPR algorithm, while preserving differences previously detected in the unaligned case. This new correction strategy contributes to revealing effects of interest which might be hidden by misalignment and has the potential to improve the specificity in longitudinal population studies beyond the traditional region of interest based analysis and along-tract analysis workflows.

Download Full-text