MRI radiomics for hamstring strain injury identification and return to sport classification: a pilot study.

OBJECTIVE: To determine if MRI-based radiomics from hamstring muscles are related to injury and if the features could be used to perform a time to return to sport (RTS) classification. We hypothesize that radiomics from hamstring muscles, especially T2-weighted and diffusion tensor imaging-based features, are related to injury and can be used for RTS classification. SUBJECTS AND METHODS: MRI data from 32 athletes at the University of Wisconsin-Madison that sustained a hamstring strain injury were collected. Diffusion tensor imaging and T1- and T2-weighted images were processed, and diffusion maps were calculated. Radiomics features were extracted from the four hamstring muscles in each limb and for each MRI modality, individually. Feature selection was performed and multiple support vector classifiers were cross-validated to differentiate between involved and uninvolved limbs and perform binary (≤ or > 25 days) and multiclass (< 14 vs. 14-42 vs. > 42 days) classification of RTS. RESULT: The combination of radiomics features from all diffusion tensor imaging and T2-weighted images provided the most accurate differentiation between involved and uninvolved limbs (AUC ≈ 0.84 ± 0.16). For the binary RTS classification, the combination of all extracted radiomics offered the most accurate classification (AUC ≈ 0.95 ± 0.15). While for the multiclass RTS classification, the combination of features from all the diffusion tensor imaging maps provided the most accurate classification (weighted one vs. rest AUC ≈ 0.81 ± 0.16). CONCLUSION: This pilot study demonstrated that radiomics features from hamstring muscles are related to injury and have the potential to predict RTS.

Application and Construction of Deep Learning Networks in Medical Imaging.

Deep learning (DL) approaches are part of the machine learning (ML) subfield concerned with the development of computational models to train artificial intelligence systems. DL models are characterized by automatically extracting high-level features from the input data to learn the relationship between matching datasets. Thus, its implementation offers an advantage over common ML methods that often require the practitioner to have some domain knowledge of the input data to select the best latent representation. As a result of this advantage, DL has been successfully applied within the medical imaging field to address problems, such as disease classification and tumor segmentation for which it is difficult or impossible to determine which image features are relevant. Therefore, taking into consideration the positive impact of DL on the medical imaging field, this article reviews the key concepts associated with its evolution and implementation. The sections of this review summarize the milestones related to the development of the DL field, followed by a description of the elements of deep neural network and an overview of its application within the medical imaging field. Subsequently, the key steps necessary to implement a supervised DL application are defined, and associated limitations are discussed.

Convergent brain microstructure across multiple genetic models of schizophrenia and autism spectrum disorder: A feasibility study.

Neuroimaging studies of psychiatric illness have revealed a broad spectrum of structural and functional perturbations that have been attributed in part to the complex genetic heterogeneity underpinning these disorders. These perturbations have been identified in both preclinical genetic models and in patients when compared to control populations, but recent work has also demonstrated strong evidence for genetic, molecular, and structural convergence of several psychiatric diseases. We explored potential similarities in neural microstructure in preclinical genetic models of ASD (Fmr1, Nrxn1, Pten) and schizophrenia (Disc1 svΔ2) and in age- and sex-matched control animals with diffusion tensor imaging (DTI) and neurite orientation dispersion and density imaging (NODDI). Our findings demonstrate a convergence in brain microstructure across these four genetic models with both tract-based and region-of-interest based analyses, which continues to buttress an emerging understanding of converging neural microstructure in psychiatric disease.

Refractory diet-dependent changes in neural microstructure: Implications for microstructural endophenotypes of neurologic and psychiatric disease.

Alterations in gut microbiome populations via dietary manipulation have been shown to induce diet-dependent changes in white matter microstructure. The purpose of this study is to examine the durability of these diet-induced microstructural alterations. We implemented a crossover experimental design where post-weaned male rats were assigned to one of four experimental diets. Following the administration of experimental diets and again following crossover and resumption of a normal diet, brains were imaged ex-vivo with diffusion tensor imaging. Following standard image preprocessing, tract-based spatial statistics and region-of-interest measurements were then calculated for all diffusion tensor indices. Voxel-wise differences in FA were identified in the high fat diet group when compared to animals receiving a control diet. Following crossover, there were new voxel-wise changes in both FA and TR that do not correspond to the regions previously identified. Animals crossed over from the high fiber diet demonstrate widespread and global changes in the diffusion tensor that stand in stark contrast to the minimal changes identified before crossover. While no significant differences between any of the diffusion metrics were identified in the high protein group before crossover, statistically significant decreased RD values were observed following resumption of a normal diet. Diet-induced changes in neural microstructure are durable changes that are unrecoverable following the resumption of a normal diet. We further show that in certain experimental diets, resumption of a normal diet can lead to further marked and unanticipated changes in white matter microstructure.

Exercise ameliorates deficits in neural microstructure in a Disc1 model of psychiatric illness.

Recent studies have investigated the effectiveness of aerobic exercise to improve physical and mental health outcomes in schizophrenia; however, few have explicitly explored the impact of aerobic exercise on neural microstructure, which is hypothesized to mediate the behavioral changes observed. Neural microstructure is influenced by numerous genetic factors including DISC1, which is a major molecular scaffold protein that interacts with partners like GSK3ß, NDEL1, and PDE4. DISC1 has been shown to play a role in neurogenesis, neuronal migration, neuronal maturation, and synaptic signaling. As with other genetic variants that present an increased risk for disease, mutations of the DISC1 gene have been implicated in the molecular intersection of schizophrenia and numerous other major psychiatric illnesses. This study investigated whether short-term exercise recovers deficits in neural microstructure in a novel genetic Disc1 svΔ2 rat model. Disc1 svΔ2 animals and age- and sex-matched controls were subjected to a treadmill exercise protocol. Subsequent ex-vivo diffusion tensor imaging (DTI) and neurite orientation dispersion and density imaging (NODDI) compared neural microstructure in regions of interest (ROI) between sedentary and exercise wild-type animals and between sedentary and exercise Disc1 svΔ2 animals. Short-term exercise uncovered no significant differences in neural microstructure between sedentary and exercise control animals but did lead to significant differences between sedentary and exercise Disc1 svΔ2 animals in neocortex, basal ganglia, corpus callosum, and external capsule, suggesting a positive benefit derived from a short-term exercise regimen. Our findings suggest that Disc1 svΔ2 animals are more sensitive to the effects of short-term exercise and highlight the ameliorating potential of positive treatment interventions such as exercise on neural microstructure in genetic backgrounds of psychiatric disease susceptibility.

Detecting Microglial Density With Quantitative Multi-Compartment Diffusion MRI.

Neuroinflammation plays a central role in the neuropathogenesis of a wide-spectrum of neurologic and psychiatric disease, but current neuroimaging methods to detect and characterize neuroinflammation are limited. We explored the sensitivity of quantitative multi-compartment diffusion MRI, and specifically neurite orientation dispersion and density imaging (NODDI), to detect changes in microglial density in the brain. Monte Carlo simulations of water diffusion using a NODDI acquisition scheme were performed to measure changes in a virtual MRI signal following modeled cellular changes within the extra-neurite space. 12-week-old C57BL/6J male mice (n = 48; 24 control, 24 treated with colony stimulating factor 1 receptor (CSF1R) inhibitor, PLX5622) were sacrificed at 0, 1, 3, and 7 days following withdrawal of CSF1R inhibition and were imaged ex-vivo to obtain measures of the orientation dispersion index (ODI). Following imaging, all brains were immunostained with Iba-1, NeuN, and GFAP for quantitative fluorescence microscopy. Cell populations were calculated with the ImageJ particle analyzer tool; correlation between microglial density and mean ODI values were calculated with Kendall's tau. Monte Carlo simulations demonstrate the sensitivity and positive correlation of ODI to increased occupancy in the extra-neurite space. Commensurate with our simulation data, ex-vivo NODDI imaging demonstrates an increase in ODI as microglia repopulate the brain following the withdrawal of CSF1R inhibition. Quantitative immunofluorescence of microglial density reveals that microglial density is positively correlated with ODI and greater hindered diffusion in the extra-neurite space (τ = 0.386, p < 0.05). Our results demonstrate that clinically feasible multi-compartment diffusion weighted imaging techniques such as NODDI are sensitive to microglial density and the cellular changes associated with microglial activation and highlights its potential to improve clinical diagnostic accuracy, patient risk stratification, and therapeutic monitoring of neuroinflammation in neurologic and psychiatric disease.

Sex-specific deficits in neurite density and white matter integrity are associated with targeted disruption of exon 2 of the Disc1 gene in the rat.

Diffusion tensor imaging (DTI) has provided remarkable insight into our understanding of white matter microstructure and brain connectivity across a broad spectrum of psychiatric disease. While DTI and other diffusion weighted magnetic resonance imaging (MRI) methods have clarified the axonal contribution to the disconnectivity seen in numerous psychiatric diseases, absent from these studies are quantitative indices of neurite density and orientation that are especially important features in regions of high synaptic density that would capture the synaptic contribution to the psychiatric disease state. Here we report the application of neurite orientation dispersion and density imaging (NODDI), an emerging microstructure imaging technique, to a novel Disc1 svΔ2 rat model of psychiatric illness and demonstrate the complementary and more specific indices of tissue microstructure found in NODDI than those reported by DTI. Our results demonstrate global and sex-specific changes in white matter microstructural integrity and deficits in neurite density as a consequence of the Disc1 svΔ2 genetic variation and highlight the application of NODDI and quantitative measures of neurite density and neurite dispersion in psychiatric disease.