Accelerated Aging after Traumatic Brain Injury: An ENIGMA Multi-Cohort Mega-Analysis.

OBJECTIVE: The long-term consequences of traumatic brain injury (TBI) on brain structure remain uncertain. Given evidence that a single significant brain injury event increases the risk of dementia, brain-age estimation could provide a novel and efficient indexing of the long-term consequences of TBI. Brain-age procedures use predictive modeling to calculate brain-age scores for an individual using structural magnetic resonance imaging (MRI) data. Complicated mild, moderate, and severe TBI (cmsTBI) is associated with a higher predicted age difference (PAD), but the progression of PAD over time remains unclear. We sought to examine whether PAD increases as a function of time since injury (TSI) and if injury severity and sex interacted to influence this progression. METHODS: Through the ENIGMA Adult Moderate and Severe (AMS)-TBI working group, we examine the largest TBI sample to date (n = 343), along with controls, for a total sample size of n = 540, to replicate and extend prior findings in the study of TBI brain age. Cross-sectional T1w-MRI data were aggregated across 7 cohorts, and brain age was established using a similar brain age algorithm to prior work in TBI. RESULTS: Findings show that PAD widens with longer TSI, and there was evidence for differences between sexes in PAD, with men showing more advanced brain age. We did not find strong evidence supporting a link between PAD and cognitive performance. INTERPRETATION: This work provides evidence that changes in brain structure after cmsTBI are dynamic, with an initial period of change, followed by relative stability in brain morphometry, eventually leading to further changes in the decades after a single cmsTBI. ANN NEUROL 2024;96:365-377.

Optimising brain age estimation through transfer learning: A suite of pre-trained foundation models for improved performance and generalisability in a clinical setting.

Estimated age from brain MRI data has emerged as a promising biomarker of neurological health. However, the absence of large, diverse, and clinically representative training datasets, along with the complexity of managing heterogeneous MRI data, presents significant barriers to the development of accurate and generalisable models appropriate for clinical use. Here, we present a deep learning framework trained on routine clinical data (N up to 18,890, age range 18-96 years). We trained five separate models for accurate brain age prediction (all with mean absolute error ≤4.0 years, R2 ≥ .86) across five different MRI sequences (T2 -weighted, T2 -FLAIR, T1 -weighted, diffusion-weighted, and gradient-recalled echo T2 *-weighted). Our trained models offer dual functionality. First, they have the potential to be directly employed on clinical data. Second, they can be used as foundation models for further refinement to accommodate a range of other MRI sequences (and therefore a range of clinical scenarios which employ such sequences). This adaptation process, enabled by transfer learning, proved effective in our study across a range of MRI sequences and scan orientations, including those which differed considerably from the original training datasets. Crucially, our findings suggest that this approach remains viable even with limited data availability (as low as N = 25 for fine-tuning), thus broadening the application of brain age estimation to more diverse clinical contexts and patient populations. By making these models publicly available, we aim to provide the scientific community with a versatile toolkit, promoting further research in brain age prediction and related areas.

The value of arterial spin labelling perfusion MRI in brain age prediction.

Current structural MRI-based brain age estimates and their difference from chronological age-the brain age gap (BAG)-are limited to late-stage pathological brain-tissue changes. The addition of physiological MRI features may detect early-stage pathological brain alterations and improve brain age prediction. This study investigated the optimal combination of structural and physiological arterial spin labelling (ASL) image features and algorithms. Healthy participants (n = 341, age 59.7 ± 14.8 years) were scanned at baseline and after 1.7 ± 0.5 years follow-up (n = 248, mean age 62.4 ± 13.3 years). From 3 T MRI, structural (T1w and FLAIR) volumetric ROI and physiological (ASL) cerebral blood flow (CBF) and spatial coefficient of variation ROI features were constructed. Multiple combinations of features and machine learning algorithms were evaluated using the Mean Absolute Error (MAE). From the best model, longitudinal BAG repeatability and feature importance were assessed. The ElasticNetCV algorithm using T1w + FLAIR+ASL performed best (MAE = 5.0 ± 0.3 years), and better compared with using T1w + FLAIR (MAE = 6.0 ± 0.4 years, p < .01). The three most important features were, in descending order, GM CBF, GM/ICV, and WM CBF. Average baseline and follow-up BAGs were similar (-1.5 ± 6.3 and - 1.1 ± 6.4 years respectively, ICC = 0.85, 95% CI: 0.8-0.9, p = .16). The addition of ASL features to structural brain age, combined with the ElasticNetCV algorithm, improved brain age prediction the most, and performed best in a cross-sectional and repeatability comparison. These findings encourage future studies to explore the value of ASL in brain age in various pathologies.

General dimensions of human brain morphometry inferred from genome-wide association data.

Understanding the neurodegenerative mechanisms underlying cognitive decline in the general population may facilitate early detection of adverse health outcomes in late life. This study investigates genetic links between brain morphometry, ageing and cognitive ability. We develop Genomic Principal Components Analysis (Genomic PCA) to model general dimensions of brain-wide morphometry at the level of their underlying genetic architecture. Genomic PCA is applied to genome-wide association data for 83 brain-wide volumes (36,778 UK Biobank participants) and we extract genomic principal components (PCs) to capture global dimensions of genetic covariance across brain regions (unlike ancestral PCs that index genetic similarity between participants). Using linkage disequilibrium score regression, we estimate genetic overlap between those general brain dimensions and cognitive ageing. The first genetic PCs underlying the morphometric organisation of 83 brain-wide regions accounted for substantial genetic variance (R2  = 40%) with the pattern of component loadings corresponding closely to those obtained from phenotypic analyses. Genetically more central regions to overall brain structure - specifically frontal and parietal volumes thought to be part of the central executive network - tended to be somewhat more susceptible towards age (r = -0.27). We demonstrate the moderate genetic overlap between the first PC underlying each of several structural brain networks and general cognitive ability (rg  = 0.17-0.21), which was not specific to a particular subset of the canonical networks examined. We provide a multivariate framework integrating covariance across multiple brain regions and the genome, revealing moderate shared genetic etiology between brain-wide morphometry and cognitive ageing.

Brain age predicted using graph convolutional neural network explains neurodevelopmental trajectory in preterm neonates.

OBJECTIVES: Dramatic brain morphological changes occur throughout the third trimester of gestation. In this study, we investigated whether the predicted brain age (PBA) derived from graph convolutional network (GCN) that accounts for cortical morphometrics in third trimester is associated with postnatal abnormalities and neurodevelopmental outcome. METHODS: In total, 577 T1 MRI scans of preterm neonates from two different datasets were analyzed; the NEOCIVET pipeline generated cortical surfaces and morphological features, which were then fed to the GCN to predict brain age. The brain age index (BAI; PBA minus chronological age) was used to determine the relationships among preterm birth (i.e., birthweight and birth age), perinatal brain injuries, postnatal events/clinical conditions, BAI at postnatal scan, and neurodevelopmental scores at 30 months. RESULTS: Brain morphology and GCN-based age prediction of preterm neonates without brain lesions (mean absolute error [MAE]: 0.96 weeks) outperformed conventional machine learning methods using no topological information. Structural equation models (SEM) showed that BAI mediated the influence of preterm birth and postnatal clinical factors, but not perinatal brain injuries, on neurodevelopmental outcome at 30 months of age. CONCLUSIONS: Brain morphology may be clinically meaningful in measuring brain age, as it relates to postnatal factors, and predicting neurodevelopmental outcome. CLINICAL RELEVANCE STATEMENT: Understanding the neurodevelopmental trajectory of preterm neonates through the prediction of brain age using a graph convolutional neural network may allow for earlier detection of potential developmental abnormalities and improved interventions, consequently enhancing the prognosis and quality of life in this vulnerable population. KEY POINTS: •Brain age in preterm neonates predicted using a graph convolutional network with brain morphological changes mediates the pre-scan risk factors and post-scan neurodevelopmental outcomes. •Predicted brain age oriented from conventional deep learning approaches, which indicates the neurodevelopmental status in neonates, shows a lack of sensitivity to perinatal risk factors and predicting neurodevelopmental outcomes. •The new brain age index based on brain morphology and graph convolutional network enhances the accuracy and clinical interpretation of predicted brain age for neonates.

Interpretable machine learning for dementia: A systematic review.

INTRODUCTION: Machine learning research into automated dementia diagnosis is becoming increasingly popular but so far has had limited clinical impact. A key challenge is building robust and generalizable models that generate decisions that can be reliably explained. Some models are designed to be inherently "interpretable," whereas post hoc "explainability" methods can be used for other models. METHODS: Here we sought to summarize the state-of-the-art of interpretable machine learning for dementia. RESULTS: We identified 92 studies using PubMed, Web of Science, and Scopus. Studies demonstrate promising classification performance but vary in their validation procedures and reporting standards and rely heavily on popular data sets. DISCUSSION: Future work should incorporate clinicians to validate explanation methods and make conclusive inferences about dementia-related disease pathology. Critically analyzing model explanations also requires an understanding of the interpretability methods itself. Patient-specific explanations are also required to demonstrate the benefit of interpretable machine learning in clinical practice.

Distinct patterns of neurodegeneration after TBI and in Alzheimer's disease.

INTRODUCTION: Traumatic brain injury (TBI) is a dementia risk factor, with Alzheimer's disease (AD) more common following injury. Patterns of neurodegeneration produced by TBI can be compared to AD and aging using volumetric MRI. METHODS: A total of 55 patients after moderate to severe TBI (median age 40), 45 with AD (median age 69), and 61 healthy volunteers underwent magnetic resonance imaging over 2 years. Atrophy patterns were compared. RESULTS: AD patients had markedly lower baseline volumes. TBI was associated with increased white matter (WM) atrophy, particularly involving corticospinal tracts and callosum, whereas AD rates were increased across white and gray matter (GM). Subcortical WM loss was shared in AD/TBI, but deep WM atrophy was TBI-specific and cortical atrophy AD-specific. Post-TBI atrophy patterns were distinct from aging, which resembled AD. DISCUSSION: Post-traumatic neurodegeneration 1.9-4.0 years (median) following moderate-severe TBI is distinct from aging/AD, predominantly involving central WM. This likely reflects distributions of axonal injury, a neurodegeneration trigger. HIGHLIGHTS: We compared patterns of brain atrophy longitudinally after moderate to severe TBI in late-onset AD and healthy aging. Patients after TBI had abnormal brain atrophy involving the corpus callosum and other WM tracts, including corticospinal tracts, in a pattern that was specific and distinct from AD and aging. This pattern is reminiscent of axonal injury following TBI, and atrophy rates were predicted by the extent of axonal injury on diffusion tensor imaging, supporting a relationship between early axonal damage and chronic neurodegeneration.

Accurate brain-age models for routine clinical MRI examinations.

Convolutional neural networks (CNN) can accurately predict chronological age in healthy individuals from structural MRI brain scans. Potentially, these models could be applied during routine clinical examinations to detect deviations from healthy ageing, including early-stage neurodegeneration. This could have important implications for patient care, drug development, and optimising MRI data collection. However, existing brain-age models are typically optimised for scans which are not part of routine examinations (e.g., volumetric T1-weighted scans), generalise poorly (e.g., to data from different scanner vendors and hospitals etc.), or rely on computationally expensive pre-processing steps which limit real-time clinical utility. Here, we sought to develop a brain-age framework suitable for use during routine clinical head MRI examinations. Using a deep learning-based neuroradiology report classifier, we generated a dataset of 23,302 'radiologically normal for age' head MRI examinations from two large UK hospitals for model training and testing (age range = 18-95 years), and demonstrate fast (< 5 s), accurate (mean absolute error [MAE] < 4 years) age prediction from clinical-grade, minimally processed axial T2-weighted and axial diffusion-weighted scans, with generalisability between hospitals and scanner vendors (Δ MAE < 1 year). The clinical relevance of these brain-age predictions was tested using 228 patients whose MRIs were reported independently by neuroradiologists as showing atrophy 'excessive for age'. These patients had systematically higher brain-predicted age than chronological age (mean predicted age difference = +5.89 years, 'radiologically normal for age' mean predicted age difference = +0.05 years, p < 0.0001). Our brain-age framework demonstrates feasibility for use as a screening tool during routine hospital examinations to automatically detect older-appearing brains in real-time, with relevance for clinical decision-making and optimising patient pathways.

Sex differences in predictors and regional patterns of brain age gap estimates.

The brain-age-gap estimate (brainAGE) quantifies the difference between chronological age and age predicted by applying machine-learning models to neuroimaging data and is considered a biomarker of brain health. Understanding sex differences in brainAGE is a significant step toward precision medicine. Global and local brainAGE (G-brainAGE and L-brainAGE, respectively) were computed by applying machine learning algorithms to brain structural magnetic resonance imaging data from 1113 healthy young adults (54.45% females; age range: 22-37 years) participating in the Human Connectome Project. Sex differences were determined in G-brainAGE and L-brainAGE. Random forest regression was used to determine sex-specific associations between G-brainAGE and non-imaging measures pertaining to sociodemographic characteristics and mental, physical, and cognitive functions. L-brainAGE showed sex-specific differences; in females, compared to males, L-brainAGE was higher in the cerebellum and brainstem and lower in the prefrontal cortex and insula. Although sex differences in G-brainAGE were minimal, associations between G-brainAGE and non-imaging measures differed between sexes with the exception of poor sleep quality, which was common to both. While univariate relationships were small, the most important predictor of higher G-brainAGE was self-identification as non-white in males and systolic blood pressure in females. The results demonstrate the value of applying sex-specific analyses and machine learning methods to advance our understanding of sex-related differences in factors that influence the rate of brain aging and provide a foundation for targeted interventions.

Mind the gap: Performance metric evaluation in brain-age prediction.

Estimating age based on neuroimaging-derived data has become a popular approach to developing markers for brain integrity and health. While a variety of machine-learning algorithms can provide accurate predictions of age based on brain characteristics, there is significant variation in model accuracy reported across studies. We predicted age in two population-based datasets, and assessed the effects of age range, sample size and age-bias correction on the model performance metrics Pearson's correlation coefficient (r), the coefficient of determination (R2 ), Root Mean Squared Error (RMSE) and Mean Absolute Error (MAE). The results showed that these metrics vary considerably depending on cohort age range; r and R2 values are lower when measured in samples with a narrower age range. RMSE and MAE are also lower in samples with a narrower age range due to smaller errors/brain age delta values when predictions are closer to the mean age of the group. Across subsets with different age ranges, performance metrics improve with increasing sample size. Performance metrics further vary depending on prediction variance as well as mean age difference between training and test sets, and age-bias corrected metrics indicate high accuracy-also for models showing poor initial performance. In conclusion, performance metrics used for evaluating age prediction models depend on cohort and study-specific data characteristics, and cannot be directly compared across different studies. Since age-bias corrected metrics generally indicate high accuracy, even for poorly performing models, inspection of uncorrected model results provides important information about underlying model attributes such as prediction variance.

Rapid processing and quantitative evaluation of structural brain scans for adaptive multimodal imaging.

Current neuroimaging acquisition and processing approaches tend to be optimised for quality rather than speed. However, rapid acquisition and processing of neuroimaging data can lead to novel neuroimaging paradigms, such as adaptive acquisition, where rapidly processed data is used to inform subsequent image acquisition steps. Here we first evaluate the impact of several processing steps on the processing time and quality of registration of manually labelled T1 -weighted MRI scans. Subsequently, we apply the selected rapid processing pipeline both to rapidly acquired multicontrast EPImix scans of 95 participants (which include T1 -FLAIR, T2 , T2 *, T2 -FLAIR, DWI and ADC contrasts, acquired in ~1 min), as well as to slower, more standard single-contrast T1 -weighted scans of a subset of 66 participants. We quantify the correspondence between EPImix T1 -FLAIR and single-contrast T1 -weighted scans, using correlations between voxels and regions of interest across participants, measures of within- and between-participant identifiability as well as regional structural covariance networks. Furthermore, we explore the use of EPImix for the rapid construction of morphometric similarity networks. Finally, we quantify the reliability of EPImix-derived data using test-retest scans of 10 participants. Our results demonstrate that quantitative information can be derived from a neuroimaging scan acquired and processed within minutes, which could further be used to implement adaptive multimodal imaging and tailor neuroimaging examinations to individual patients.

Deep learning to automate the labelling of head MRI datasets for computer vision applications.

OBJECTIVES: The purpose of this study was to build a deep learning model to derive labels from neuroradiology reports and assign these to the corresponding examinations, overcoming a bottleneck to computer vision model development. METHODS: Reference-standard labels were generated by a team of neuroradiologists for model training and evaluation. Three thousand examinations were labelled for the presence or absence of any abnormality by manually scrutinising the corresponding radiology reports ('reference-standard report labels'); a subset of these examinations (n = 250) were assigned 'reference-standard image labels' by interrogating the actual images. Separately, 2000 reports were labelled for the presence or absence of 7 specialised categories of abnormality (acute stroke, mass, atrophy, vascular abnormality, small vessel disease, white matter inflammation, encephalomalacia), with a subset of these examinations (n = 700) also assigned reference-standard image labels. A deep learning model was trained using labelled reports and validated in two ways: comparing predicted labels to (i) reference-standard report labels and (ii) reference-standard image labels. The area under the receiver operating characteristic curve (AUC-ROC) was used to quantify model performance. Accuracy, sensitivity, specificity, and F1 score were also calculated. RESULTS: Accurate classification (AUC-ROC > 0.95) was achieved for all categories when tested against reference-standard report labels. A drop in performance (ΔAUC-ROC > 0.02) was seen for three categories (atrophy, encephalomalacia, vascular) when tested against reference-standard image labels, highlighting discrepancies in the original reports. Once trained, the model assigned labels to 121,556 examinations in under 30 min. CONCLUSIONS: Our model accurately classifies head MRI examinations, enabling automated dataset labelling for downstream computer vision applications. KEY POINTS: • Deep learning is poised to revolutionise image recognition tasks in radiology; however, a barrier to clinical adoption is the difficulty of obtaining large labelled datasets for model training. • We demonstrate a deep learning model which can derive labels from neuroradiology reports and assign these to the corresponding examinations at scale, facilitating the development of downstream computer vision models. • We rigorously tested our model by comparing labels predicted on the basis of neuroradiology reports with two sets of reference-standard labels: (1) labels derived by manually scrutinising each radiology report and (2) labels derived by interrogating the actual images.

Beyond the average patient: how neuroimaging models can address heterogeneity in dementia.

Dementia is a highly heterogeneous condition, with pronounced individual differences in age of onset, clinical presentation, progression rates and neuropathological hallmarks, even within a specific diagnostic group. However, the most common statistical designs used in dementia research studies and clinical trials overlook this heterogeneity, instead relying on comparisons of group average differences (e.g. patient versus control or treatment versus placebo), implicitly assuming within-group homogeneity. This one-size-fits-all approach potentially limits our understanding of dementia aetiology, hindering the identification of effective treatments. Neuroimaging has enabled the characterization of the average neuroanatomical substrates of dementias; however, the increasing availability of large open neuroimaging datasets provides the opportunity to examine patterns of neuroanatomical variability in individual patients. In this update, we outline the causes and consequences of heterogeneity in dementia and discuss recent research that aims to tackle heterogeneity directly, rather than assuming that dementia affects everyone in the same way. We introduce spatial normative modelling as an emerging data-driven technique, which can be applied to dementia data to model neuroanatomical variation, capturing individualized neurobiological 'fingerprints'. Such methods have the potential to detect clinically relevant subtypes, track an individual's disease progression or evaluate treatment responses, with the goal of moving towards precision medicine for dementia.

Associations between alcohol use and accelerated biological ageing.

Harmful alcohol use is a leading cause of premature death and is associated with age-related disease. Biological ageing is highly variable between individuals and may deviate from chronological ageing, suggesting that biomarkers of biological ageing (derived from DNA methylation or brain structural measures) may be clinically relevant. Here, we investigated the relationships between alcohol phenotypes and both brain and DNA methylation age estimates. First, using data from UK Biobank and Generation Scotland, we tested the association between alcohol consumption (units/week) or hazardous use (Alcohol Use Disorders Identification Test [AUDIT] scores) and accelerated brain and epigenetic ageing in 20,258 and 8051 individuals, respectively. Second, we used Mendelian randomisation (MR) to test for a causal effect of alcohol consumption levels and alcohol use disorder (AUD) on biological ageing. Alcohol use showed a consistent positive association with higher predicted brain age (AUDIT-C: ß = 0.053, p = 3.16 × 10-13 ; AUDIT-P: ß = 0.052, p = 1.6 × 10-13 ; total AUDIT score: ß = 0.062, p = 5.52 × 10-16 ; units/week: ß = 0.078, p = 2.20 × 10-16 ), and two DNA methylation-based estimates of ageing, GrimAge (units/week: ß = 0.053, p = 1.48 × 10-7 ) and PhenoAge (units/week: ß = 0.077, p = 2.18x10-10 ). MR analyses revealed limited evidence for a causal effect of AUD on accelerated brain ageing (ß = 0.118, p = 0.044). However, this result should be interpreted cautiously as the significant effect was driven by a single genetic variant. We found no evidence for a causal effect of alcohol consumption levels on accelerated biological ageing. Future studies investigating the mechanisms associating alcohol use with accelerated biological ageing are warranted.

Prediction of brain age and cognitive age: Quantifying brain and cognitive maintenance in aging.

The concept of brain maintenance refers to the preservation of brain integrity in older age, while cognitive reserve refers to the capacity to maintain cognition in the presence of neurodegeneration or aging-related brain changes. While both mechanisms are thought to contribute to individual differences in cognitive function among older adults, there is currently no "gold standard" for measuring these constructs. Using machine-learning methods, we estimated brain and cognitive age based on deviations from normative aging patterns in the Whitehall II MRI substudy cohort (N = 537, age range = 60.34-82.76), and tested the degree of correspondence between these constructs, as well as their associations with premorbid IQ, education, and lifestyle trajectories. In line with established literature highlighting IQ as a proxy for cognitive reserve, higher premorbid IQ was linked to lower cognitive age independent of brain age. No strong evidence was found for associations between brain or cognitive age and lifestyle trajectories from midlife to late life based on latent class growth analyses. However, post hoc analyses revealed a relationship between cumulative lifestyle measures and brain age independent of cognitive age. In conclusion, we present a novel approach to characterizing brain and cognitive maintenance in aging, which may be useful for future studies seeking to identify factors that contribute to brain preservation and cognitive reserve mechanisms in older age.

Longitudinal Assessment of Multiple Sclerosis with the Brain-Age Paradigm.

OBJECTIVE: During the natural course of multiple sclerosis (MS), the brain is exposed to aging as well as disease effects. Brain aging can be modeled statistically; the so-called "brain-age" paradigm. Here, we evaluated whether brain-predicted age difference (brain-PAD) was sensitive to the presence of MS, clinical progression, and future outcomes. METHODS: In a longitudinal, multicenter sample of 3,565 magnetic resonance imaging (MRI) scans, in 1,204 patients with MS and clinically isolated syndrome (CIS) and 150 healthy controls (mean follow-up time: patients 3.41 years, healthy controls 1.97 years), we measured "brain-predicted age" using T1-weighted MRI. We compared brain-PAD among patients with MS and patients with CIS and healthy controls, and between disease subtypes. Relationships between brain-PAD and Expanded Disability Status Scale (EDSS) were explored. RESULTS: Patients with MS had markedly higher brain-PAD than healthy controls (mean brain-PAD +10.3 years; 95% confidence interval [CI] = 8.5-12.1] versus 4.3 years; 95% CI = 2.1 to 6.4; p < 0.001). The highest brain-PADs were in secondary-progressive MS (+13.3 years; 95% CI = 11.3-15.3). Brain-PAD at study entry predicted time-to-disability progression (hazard ratio 1.02; 95% CI = 1.01-1.03; p < 0.001); although normalized brain volume was a stronger predictor. Greater annualized brain-PAD increases were associated with greater annualized EDSS score (r = 0.26; p < 0.001). INTERPRETATION: The brain-age paradigm is sensitive to MS-related atrophy and clinical progression. A higher brain-PAD at baseline was associated with more rapid disability progression and the rate of change in brain-PAD related to worsening disability. Potentially, "brain-age" could be used as a prognostic biomarker in early-stage MS, to track disease progression or stratify patients for clinical trial enrollment. ANN NEUROL 2020 ANN NEUROL 2020;88:93-105.

Diffuse axonal injury predicts neurodegeneration after moderate-severe traumatic brain injury.

Traumatic brain injury is associated with elevated rates of neurodegenerative diseases such as Alzheimer's disease and chronic traumatic encephalopathy. In experimental models, diffuse axonal injury triggers post-traumatic neurodegeneration, with axonal damage leading to Wallerian degeneration and toxic proteinopathies of amyloid and hyperphosphorylated tau. However, in humans the link between diffuse axonal injury and subsequent neurodegeneration has yet to be established. Here we test the hypothesis that the severity and location of diffuse axonal injury predicts the degree of progressive post-traumatic neurodegeneration. We investigated longitudinal changes in 55 patients in the chronic phase after moderate-severe traumatic brain injury and 19 healthy control subjects. Fractional anisotropy was calculated from diffusion tensor imaging as a measure of diffuse axonal injury. Jacobian determinant atrophy rates were calculated from serial volumetric T1 scans as a measure of measure post-traumatic neurodegeneration. We explored a range of potential predictors of longitudinal post-traumatic neurodegeneration and compared the variance in brain atrophy that they explained. Patients showed widespread evidence of diffuse axonal injury, with reductions of fractional anisotropy at baseline and follow-up in large parts of the white matter. No significant changes in fractional anisotropy over time were observed. In contrast, abnormally high rates of brain atrophy were seen in both the grey and white matter. The location and extent of diffuse axonal injury predicted the degree of brain atrophy: fractional anisotropy predicted progressive atrophy in both whole-brain and voxelwise analyses. The strongest relationships were seen in central white matter tracts, including the body of the corpus callosum, which are most commonly affected by diffuse axonal injury. Diffuse axonal injury predicted substantially more variability in white matter atrophy than other putative clinical or imaging measures, including baseline brain volume, age, clinical measures of injury severity and microbleeds (>50% for fractional anisotropy versus <5% for other measures). Grey matter atrophy was not predicted by diffuse axonal injury at baseline. In summary, diffusion MRI measures of diffuse axonal injury are a strong predictor of post-traumatic neurodegeneration. This supports a causal link between axonal injury and the progressive neurodegeneration that is commonly seen after moderate/severe traumatic brain injury but has been of uncertain aetiology. The assessment of diffuse axonal injury with diffusion MRI is likely to improve prognostic accuracy and help identify those at greatest neurodegenerative risk for inclusion in clinical treatment trials.

Multimodal brain-age prediction and cardiovascular risk: The Whitehall II MRI sub-study.

Brain age is becoming a widely applied imaging-based biomarker of neural aging and potential proxy for brain integrity and health. We estimated multimodal and modality-specific brain age in the Whitehall II (WHII) MRI cohort using machine learning and imaging-derived measures of gray matter (GM) morphology, white matter microstructure (WM), and resting state functional connectivity (FC). The results showed that the prediction accuracy improved when multiple imaging modalities were included in the model (R2 = 0.30, 95% CI [0.24, 0.36]). The modality-specific GM and WM models showed similar performance (R2 = 0.22 [0.16, 0.27] and R2 = 0.24 [0.18, 0.30], respectively), while the FC model showed the lowest prediction accuracy (R2 = 0.002 [-0.005, 0.008]), indicating that the FC features were less related to chronological age compared to structural measures. Follow-up analyses showed that FC predictions were similarly low in a matched sub-sample from UK Biobank, and although FC predictions were consistently lower than GM predictions, the accuracy improved with increasing sample size and age range. Cardiovascular risk factors, including high blood pressure, alcohol intake, and stroke risk score, were each associated with brain aging in the WHII cohort. Blood pressure showed a stronger association with white matter compared to gray matter, while no differences in the associations of alcohol intake and stroke risk with these modalities were observed. In conclusion, machine-learning based brain age prediction can reduce the dimensionality of neuroimaging data to provide meaningful biomarkers of individual brain aging. However, model performance depends on study-specific characteristics including sample size and age range, which may cause discrepancies in findings across studies.

Nonlinear biomarker interactions in conversion from mild cognitive impairment to Alzheimer's disease.

Multiple biomarkers can capture different facets of Alzheimer's disease. However, statistical models of biomarkers to predict outcomes in Alzheimer's rarely model nonlinear interactions between these measures. Here, we used Gaussian Processes to address this, modelling nonlinear interactions to predict progression from mild cognitive impairment (MCI) to Alzheimer's over 3 years, using Alzheimer's Disease Neuroimaging Initiative (ADNI) data. Measures included: demographics, APOE4 genotype, CSF (amyloid-ß42, total tau, phosphorylated tau), [18F ]florbetapir, hippocampal volume and brain-age. We examined: (a) the independent value of each biomarker; and (b) whether modelling nonlinear interactions between biomarkers improved predictions. Each measured added complementary information when predicting conversion to Alzheimer's. A linear model classifying stable from progressive MCI explained over half the variance (R2 = 0.51, p < .001); the strongest independently contributing biomarker was hippocampal volume (R2 = 0.13). When comparing sensitivity of different models to progressive MCI (independent biomarker models, additive models, nonlinear interaction models), we observed a significant improvement (p < .001) for various two-way interaction models. The best performing model included an interaction between amyloid-ß-PET and P-tau, while accounting for hippocampal volume (sensitivity = 0.77, AUC = 0.826). Closely related biomarkers contributed uniquely to predict conversion to Alzheimer's. Nonlinear biomarker interactions were also implicated, and results showed that although for some patients adding additional biomarkers may add little value (i.e., when hippocampal volume is high), for others (i.e., with low hippocampal volume) further invasive and expensive examination may be warranted. Our framework enables visualisation of these interactions, in individual patient biomarker 'space', providing information for personalised or stratified healthcare or clinical trial design.

An automated machine learning approach to predict brain age from cortical anatomical measures.

The use of machine learning (ML) algorithms has significantly increased in neuroscience. However, from the vast extent of possible ML algorithms, which one is the optimal model to predict the target variable? What are the hyperparameters for such a model? Given the plethora of possible answers to these questions, in the last years, automated ML (autoML) has been gaining attention. Here, we apply an autoML library called Tree-based Pipeline Optimisation Tool (TPOT) which uses a tree-based representation of ML pipelines and conducts a genetic programming-based approach to find the model and its hyperparameters that more closely predicts the subject's true age. To explore autoML and evaluate its efficacy within neuroimaging data sets, we chose a problem that has been the focus of previous extensive study: brain age prediction. Without any prior knowledge, TPOT was able to scan through the model space and create pipelines that outperformed the state-of-the-art accuracy for Freesurfer-based models using only thickness and volume information for anatomical structure. In particular, we compared the performance of TPOT (mean absolute error [MAE]: 4.612 ± .124 years) and a relevance vector regression (MAE 5.474 ± .140 years). TPOT also suggested interesting combinations of models that do not match the current most used models for brain prediction but generalise well to unseen data. AutoML showed promising results as a data-driven approach to find optimal models for neuroimaging applications.