Ischemic stroke is a leading cause of death worldwide, but there has been little success translating putative cerebroprotectants from preclinical trials to patients. We investigated computational image-based assessment tools for practical improvement of the quality, scalability, and outlook for large scale preclinical screening for potential therapeutic interventions. We developed, evaluated, and deployed a pipeline for image-based stroke outcome quantification for the Stroke Prelinical Assessment Network (SPAN), which is a multi-site, multi-arm, multi-stage study evaluating a suite of cerebroprotectant interventions. Our fully automated pipeline combines state-of-the-art algorithmic and data analytic approaches to assess stroke outcomes from multi-parameter MRI data collected longitudinally from a rodent model of middle cerebral artery occlusion (MCAO), including measures of infarct volume, brain atrophy, midline shift, and data quality. We tested our approach with 1,368 scans and report population level results of lesion extent and longitudinal changes from injury. We validated our system by comparison with manual annotations of coronal MRI slices and tissue sections from the same brain, using crowdsourcing from blinded stroke experts from the network. Our results demonstrate the efficacy and robustness of our image-based stroke assessments. The pipeline may provide a promising resource for ongoing preclinical studies conducted by SPAN and other networks in the future.

We developed a novel method for mapping the location, surface area, thickness, and volume of frontoinsular cortex (FI) using structural and diffusion magnetic resonance imaging. FI lies in the ventral part of anterior insular cortex and is characterized by its distinctive population von Economo neurons (VENs). Functional neuroimaging studies have revealed its involvement in affective processing, and histopathology has implicated VEN loss in behavioral-variant frontotemporal dementia and chronic alcoholism; however, structural neuroimaging of FI has been relatively limited. We delineated FI by jointly modeling cortical surface geometry and its coincident diffusion microstructure parameters. We found that neurite orientation dispersion in cortical gray matter can be used to map FI in specific individuals, and the derived measures reflect a range of behavioral factors in young adults from the Human Connectome Project (N=1052). FI volume was larger in the left hemisphere than the right (31%), and the percentage volume of FI was larger in women than men (15.3%). FI volume was associated with measures of decision-making (delay discounting, substance abuse), emotion (negative intrusive thinking and perception of hostility), and social behavior (theory of mind and working memory for faces). The common denominator is that larger FI size is related to greater self-control and social awareness.

Rodent stroke models are important for evaluating treatments and understanding the pathophysiology and behavioral changes of brain ischemia, and magnetic resonance imaging (MRI) is a valuable tool for measuring outcome in preclinical studies. Brain extraction is an essential first step in most neuroimaging pipelines; however, it can be challenging in the presence of severe pathology and when dataset quality is highly variable. Convolutional neural networks (CNNs) can improve accuracy and reduce operator time, facilitating high throughput preclinical studies. As part of an ongoing preclinical stroke imaging study, we developed a deep-learning mouse brain extraction tool by using a U-net CNN. While previous studies have evaluated U-net architectures, we sought to evaluate their practical performance across data types. We ask how performance is affected with data across: six imaging centers, two time points after experimental stroke, and across four MRI contrasts. We trained, validated, and tested a typical U-net model on 240 multimodal MRI datasets including quantitative multi-echo T2 and apparent diffusivity coefficient (ADC) maps, and performed qualitative evaluation with a large preclinical stroke database (N=1,368). We describe the design and development of this system, and report our findings linking data characteristics to segmentation performance. We consistently found high accuracy and ability of the U-net architecture to generalize performance in a range of 95-97% accuracy, with only modest reductions in performance based on lower fidelity imaging hardware and brain pathology. This work can help inform the design of future preclinical rodent imaging studies and improve their scalability and reliability.

Life satisfaction is a component of subjective well-being that reflects a global judgement of the quality of life according to an individual’s own needs and expectations. As a psychological construct, it has attracted attention due to its relationship to mental health, resilience to stress, and other factors. Neuroimaging studies have identified neurobiological correlates of life satisfaction; however, they are limited to functional connectivity and gray matter morphometry. We explored features of gray matter microstructure obtained through compartmental modeling of multi-shell diffusion MRI data, and we examined cortical microstructure in frontoinsular cortex in a cohort of 807 typical young adults scanned as part of the Human Connectome Project. Our experiments identified the orientation dispersion index (ODI), and analogously fractional anisotropy (FA), of frontoinsular cortex as a robust set of anatomically-specific lateralized diffusion MRI microstructure features that are linked to life satisfaction, independent of other demographic, socioeconomic, and behavioral factors. We further validated our findings in a secondary test-retest dataset and found high reliability of our imaging metrics and reproducibility of outcomes. In our analysis of twin and non-twin siblings, we found basic microstructure in frontoinsular cortex to be strongly genetically determined. We also found a more moderate but still very significant genetic role in determining microstructure as it relates to life satisfaction in frontoinsular cortex. Our findings suggest a potential linkage between well-being and microscopic features of frontoinsular cortex, which may reflect cellular morphology and architecture and may more broadly implicate the integrity of the homeostatic processing performed by frontoinsular cortex as an important component of an individual’s judgements of life satisfaction.

We mapped the connections of the insular von Economo neuron (VEN) area in ex vivo brains of a bonobo, an orangutan and two gorillas with high angular resolution diffusion MRI imaging acquired in 36 h imaging sessions for each brain. The apes died of natural causes without neurological disorders. The localization of the insular VEN area was based on cresyl violet-stained histological sections from each brain that were coregistered with structural and diffusion images from the same individuals. Diffusion MRI tractography showed that the insular VEN area is connected with olfactory, gustatory, visual and other sensory systems, as well as systems for the mediation of appetite, reward, aversion and motivation. The insular VEN area in apes is most strongly connected with frontopolar cortex, which could support their capacity to choose voluntarily among alternative courses of action particularly in exploring for food resources. The frontopolar cortex may also support their capacity to take note of potential resources for harvesting in the future (prospective memory). All of these faculties may support insight and volitional choice when contemplating courses of action as opposed to rule-based decision-making.

We propose a novel approach for processing diffusion MRI tractography datasets using the sparse closest point transform (SCPT). Tractography enables the 3D geometry of white matter pathways to be reconstructed; however, algorithms for processing them are often highly customized, and thus, do not leverage the existing wealth of machine learning (ML) algorithms. We investigated a vector-space tractography representation that aims to bridge this gap by using the SCPT, which consists of two steps: first, extracting sparse and representative landmarks from a tractography dataset, and second transforming curves relative to these landmarks with a closest point transform. We explore its use in three typical tasks: fiber bundle clustering, simplification, and selection across a population. The clustering algorithm groups fibers from single whole-brain datasets using a non-parametric k-means clustering algorithm, with performance compared with three alternative methods and across four datasets. The simplification algorithm removes redundant curves to improve interactive visualization, with performance gauged relative to random subsampling. The selection algorithm extracts bundles across a population using a one-class Gaussian classifier derived from an atlas prototype, with performance gauged by scan- rescan reliability and sensitivity to normal aging, as compared to manual mask-based selection. Our results demonstrate how the SCPT enables the novel application of existing vector-space ML algorithms to create effective and efficient tools for tractography processing. Our experimental data is available online, and our software implementation is available in the Quantitative Imaging Toolkit.

The endocannabinoid system serves a critical role in homeostatic regulation through its influence on processes underlying appetite, pain, reward, and stress, and cannabis has long been used for the related modulatory effects it provides through tetrahydrocannabinol (THC). We investigated how THC exposure relates to tissue microstructure of the cerebral cortex and subcortical nuclei using computational modeling of diffusion magnetic resonance imaging data in a large cohort of young adults from the Human Connectome Project. We report strong associations between biospecimen-defined THC exposure and microstructure parameters in discrete gray matter brain areas, including frontoinsular cortex, ventromedial prefrontal cortex, and the lateral amygdala subfields, with independent effects in behavioral measures of memory performance, negative intrusive thinking, and paternal substance abuse. These results shed new light on the relationship between THC exposure and microstructure variation in brain areas related to salience processing, emotion regulation, and decision making. The absence of effects in some other cannabinoid-receptor-rich brain areas prompts the consideration of cellular and molecular mechanisms that we discuss. Further studies are needed to characterize the nature of these effects across the lifespan and to investigate the mechanistic neurobiological factors connecting THC exposure and microstructural parameters.

Epilepsy is a debilitating neurological disorder that directly impacts millions of people and exerts a tremendous economic burden on society at large. While traumatic brain injury (TBI) is a common cause, there remain many open questions regarding its pathological mechanism. The goal of the Epilepsy Bioinformatics Study for Antiepileptogenic Therapy (EpiBioS4Rx) is to identify epileptogenic biomarkers through a comprehensive project spanning multiple species, modalities, and research institutions; in particular, diffusion magnetic resonance imaging (MRI) is a critical component, as it probes tissue microstructure and structural connectivity. The project includes in vivo imaging of a rodent fluid-percussion model of TBI, and we developed a computational diffusion MRI framework for EpiBioS4Rx which employs advanced techniques for preprocessing, modeling, spatial normalization, region analysis, and tractography to derive imaging metrics at group and individual levels. We describe the system’s design, present characteristic results from a longitudinal cohort, and discuss its role in biomarker discovery and further studies.

We develop and evaluate a novel hybrid tractography algorithm for improved segmentation of complex fiber bundles from diffusion magnetic resonance imaging datasets. We propose an approach inspired by reinforcement learning that combines the strengths of both probabilistic and deterministic tractography to better resolve pathways dominated by crossing fibers. Given a fiber bundle query, our approach first explores an array of possible pathways probabilistically, and then exploits this information with streamline tractography using globally optimal fiber compartment assignment in a conditional random field. We quantitatively evaluated our approach in comparison with deterministic and probabilistic approaches using a realistic phantom with Tractometer and 88 test-retest scans from the Human Connectome Project. We found that the proposed hybrid method offers improved accuracy with phantom data and more biologically plausible topographic organization and higher reliability with in vivo data. This demonstrates the benefits of combining tractography approaches and indicates opportunities for integrating reinforcement learning strategies into tractography algorithms.

Computational tools are increasingly important to MR imaging research, as they can make experiments more reproducible, improve our ability to share our findings and methods, and facilitate hypothesis generation. We aim to contribute a software package to the research community named the Quantitative Imaging Toolkit (QIT). QIT was developed to provide tools for 3D visualization, data exploration, and computational analysis of neuroimaging datasets. While meant to be generally useful for neuroimaging, the tools have extensively developed features for analyzing diffusion MRI data, running large population imaging analyses, and developing new algorithms.

This paper presents a comparative evaluation of methods for automated voxel-based spatial mapping in diffusion tensor imaging studies. Such methods are an essential step in computational pipelines and provide anatomically comparable measurements across a population in atlas-based studies. To better understand their strengths and weaknesses, we tested a total of eight methods for voxel-based spatial mapping in two types of diffusion tensor templates. The methods were evaluated with respect to scan-rescan reliability and an application to normal aging. The methods included voxel-based analysis with and without smoothing, two types of region-based analysis, and combinations thereof with skeletonization. The templates included a study-specific template created with DTI-TK and the IIT template serving as a standard template. To control for other factors in the pipeline, the experiments used a common dataset, acquired at 1.5T with a single shell high angular resolution diffusion MR imaging protocol, and tensor-based spatial normalization with DTI-TK. Scan-rescan reliability was assessed using the coefficient of variation (CV) and intraclass correlation (ICC) in eight subjects with three scans each. Sensitivity to normal aging was assessed in a population of 80 subjects aged 25-65 years old, and methods were compared with respect to the anatomical agreement of significant findings and the R2 of the associated models of fractional anisotropy. The results show that reliability depended greatly on the method used for spatial mapping. The largest differences in reliability were found when adding smoothing and comparing voxel-based and region-based analyses. Skeletonization and template type were found to have either a small or negligible effect on reliability. The aging results showed agreement among the methods in nine brain areas, with some methods showing more sensitivity than others. Skeletonization and smoothing were not major factors affecting sensitivity to aging, but the standard template showed higher R2 in several conditions. A structural comparison of the templates showed that large deformations between them may be related to observed differences in patterns of significant voxels. Most areas showed significantly higher R2 with voxel-based analysis, particularly when clusters were smaller than the available regions-of-interest. Looking forward, these results can potentially help to interpret results from existing white matter imaging studies, as well as provide a resource to help in planning future studies to maximize reliability and sensitivity with regard to the scientific goals at hand.

We present and evaluate a method for kernel regression estimation of fiber orientations and associated volume fractions for diffusion MR tractography and population-based atlas construction in clinical imaging studies of brain white matter. This is a model-based image processing technique in which representative fiber models are estimated from collections of component fiber models in model-valued image data. This extends prior work in nonparametric image processing and multi-compartment processing to provide computational tools for image interpolation, smoothing, and fusion with fiber orientation mixtures. In contrast to related work on multi-compartment processing, this approach is based on directional measures of divergence and includes data-adaptive extensions for model selection and bilateral filtering. This is useful for reconstructing complex anatomical features in clinical datasets analyzed with the ball-and-sticks model, and our framework’s data-adaptive extensions are potentially useful for general multi-compartment image processing. We experimentally evaluate our approach with both synthetic data from computational phantoms and in vivo clinical data from human subjects. With synthetic data experiments, we evaluate performance based on errors in fiber orientation, volume fraction, compartment count, and tractography-based connectivity. With in vivo data experiments, we first show improved scan-rescan reproducibility and reliability of quantitative fiber bundle metrics, including mean length, volume, streamline count, and mean volume fraction. We then demonstrate the creation of a multi-fiber tractography atlas from a population of 80 human subjects. In comparison to single tensor atlasing, our multi-fiber atlas shows more complete features of known fiber bundles and includes reconstructions of the lateral projections of the corpus callosum and complex fronto-parietal connections of the superior longitudinal fasciculus I, II, and III.