Diffusion Tensor Based White Matter Tract Atlases for Pediatric Populations

Advanced magnetic resonance neuroimaging techniques: feasibility and applications in long or post-COVID-19 syndrome - a review

Long-term or post-COVID-19 syndrome (PCS) is a condition that affects people infected with SARS‑CoV‑2, the virus that causes COVID-19. PCS is characterized by a wide range of persistent or new symptoms that last months after the initial infection, such as fatigue, shortness of breath, cognitive dysfunction, and pain. Advanced magnetic resonance (MR) neuroimaging techniques can provide valuable information on the structural and functional changes in the brain associated with PCS as well as potential biomarkers for diagnosis and prognosis. In this review, we discuss the feasibility and applications of various advanced MR neuroimaging techniques in PCS, including perfusion-weighted imaging (PWI), diffusion-weighted imaging (DWI), susceptibility-weighted imaging (SWI), functional MR imaging (fMRI), diffusion tensor imaging (DTI), and tractography. We summarize the current evidence on neuroimaging findings in PCS, the challenges and limitations of these techniques, and the future directions for research and clinical practice. Although still uncertain, advanced MRI techniques show promise for gaining insight into the pathophysiology and guiding the management of COVID-19 syndrome, pending larger validation studies.

Identifying interindividual variability of social perception and associated brain anatomical correlations in children with autism spectrum disorder using eye-tracking and diffusion tensor imaging MRI (DTI-MRI)

Even though deficits in social cognition constitute a core characteristic of autism spectrum disorders, a large heterogeneity exists regarding individual social performances and its neural basis remains poorly investigated. Here, we used eye-tracking to objectively measure interindividual variability in social perception and its correlation with white matter microstructure, measured with diffusion tensor imaging MRI, in 25 children with autism spectrum disorder (8.5 ± 3.8 years). Beyond confirming deficits in social perception in participants with autism spectrum disorder compared 24 typically developing controls (10.5 ± 2.9 years), results revealed a large interindividual variability of such behavior among individuals with autism spectrum disorder. Whole-brain analysis showed in both autism spectrum disorder and typically developing groups a positive correlation between number of fixations to the eyes and fractional anisotropy values mainly in right and left superior longitudinal tracts. In children with autism spectrum disorder a correlation was also observed in right and left inferior longitudinal tracts. Importantly, a significant interaction between group and number of fixations to the eyes was observed within the anterior portion of the right inferior longitudinal fasciculus, mainly in the right anterior temporal region. This additional correlation in a supplementary region suggests the existence of a compensatory brain mechanism, which may support enhanced performance in social perception among children with autism spectrum disorder.

Variational Autoencoders for Generating Synthetic Tractography-Based Bundle Templates in a Low-Data Setting

White matter tracts generated from whole brain tractography are often processed using automatic segmentation methods with standard atlases. Atlases are generated from hundreds of subjects, which becomes time-consuming to create and difficult to apply to all populations. In this study, we extended our prior work on using a deep generative model - a Convolutional Variational Autoencoder - to map complex and data-intensive streamlines to a low-dimensional latent space given a limited sample size of 50 subjects from the ADNI3 dataset, to generate synthetic population-specific bundle templates using Kernel Density Estimation (KDE) on streamline embeddings. We conducted a quantitative shape analysis by calculating bundle shape metrics, and found that our bundle templates better capture the shape distribution of the bundles than the atlas data used in the original segmentation derived from young healthy adults. We further demonstrated the use of our framework for direct bundle segmentation from whole-brain tractograms.

Language exposure during infancy is negatively associated with white matter microstructure in the arcuate fasciculus

Decades of research have established that the home language environment, especially quality of caregiver speech, supports language acquisition during infancy. However, the neural mechanisms behind this phenomenon remain under studied. In the current study, we examined associations between the home language environment and structural coherence of white matter tracts in 52 typically developing infants from English speaking homes in a western society. Infants participated in at least one MRI brain scan when they were 3, 6, 12, and/or 24 months old. Home language recordings were collected when infants were 9 and/or 15 months old. General linear regression models indicated that infants who heard the most adult words and participated in the most conversational turns at 9 months of age also had the lowest fractional anisotropy in the left posterior parieto-temporal arcuate fasciculus at 24 months. Similarly, infants who vocalized the most at 9 months also had the lowest fractional anisotropy in the same tract at 6 months of age. This is one of the first studies to report significant associations between caregiver speech collected in the home and white matter structural coherence in the infant brain. The results are in line with prior work showing that protracted white matter development during infancy confers a cognitive advantage.

Variational Autoencoders for Generating Synthetic Tractography-Based Bundle Templates in a Low-Data Setting

White matter tracts generated from whole brain tractography are often processed using automatic segmentation methods with standard atlases. Atlases are generated from hundreds of subjects, which becomes time-consuming to create and difficult to apply to all populations. In this study, we extended our prior work on using a deep generative model a Convolutional Variational Autoencoder - to map complex and data-intensive streamlines to a low-dimensional latent space given a limited sample size of 50 subjects from the ADNI3 dataset, to generate synthetic population-specific bundle templates using Kernel Density Estimation (KDE) on streamline embeddings. We conducted a quantitative shape analysis by calculating bundle shape metrics, and found that our bundle templates better capture the shape distribution of the bundles than the atlas data used in the original segmentation derived from young healthy adults. We further demonstrated the use of our framework for direct bundle segmentation from whole-brain tractograms.