SynthSR: A public AI tool to turn heterogeneous clinical brain scans into high-resolution T1-weighted images for 3D morphometry.

Every year, millions of brain magnetic resonance imaging (MRI) scans are acquired in hospitals across the world. These have the potential to revolutionize our understanding of many neurological diseases, but their morphometric analysis has not yet been possible due to their anisotropic resolution. We present an artificial intelligence technique, "SynthSR," that takes clinical brain MRI scans with any MR contrast (T1, T2, etc.), orientation (axial/coronal/sagittal), and resolution and turns them into high-resolution T1 scans that are usable by virtually all existing human neuroimaging tools. We present results on segmentation, registration, and atlasing of >10,000 scans of controls and patients with brain tumors, strokes, and Alzheimer's disease. SynthSR yields morphometric results that are very highly correlated with what one would have obtained with high-resolution T1 scans. SynthSR allows sample sizes that have the potential to overcome the power limitations of prospective research studies and shed new light on the healthy and diseased human brain.

Early-Onset Dementia in War Veterans: Brain Polypathology and Clinicopathologic Complexity.

The neuropathology associated with cognitive decline in military personnel exposed to traumatic brain injury (TBI) and chronic stress is incompletely understood. Few studies have examined clinicopathologic correlations between phosphorylated-tau neurofibrillary tangles, ß-amyloid neuritic plaques, neuroinflammation, or white matter (WM) lesions, and neuropsychiatric disorders in veterans. We describe clinicopathologic findings in 4 military veterans with early-onset dementia (EOD) who had varying histories of blunt- and blast-TBI, cognitive decline, behavioral abnormalities, post-traumatic stress disorder, suicidal ideation, and suicide. We found that pathologic lesions in these military-EOD cases could not be categorized as classic Alzheimer's disease (AD), chronic traumatic encephalopathy, traumatic axonal injury, or other well-characterized clinicopathologic entities. Rather, we observed a mixture of polypathology with unusual patterns compared with pathologies found in AD or other dementias. Also, ultrahigh resolution ex vivo MRI in 2 of these 4 brains revealed unusual patterns of periventricular WM injury. These findings suggest that military-EOD cases are associated with atypical combinations of brain lesions and distribution rarely seen in nonmilitary populations. Future prospective studies that acquire neuropsychiatric data before and after deployments, as well as genetic and environmental exposure data, are needed to further elucidate clinicopathologic correlations in military-EOD.

Regionally specific TSC1 and TSC2 gene expression in tuberous sclerosis complex.

Tuberous sclerosis complex (TSC), a heritable neurodevelopmental disorder, is caused by mutations in the TSC1 or TSC2 genes. To date, there has been little work to elucidate regional TSC1 and TSC2 gene expression within the human brain, how it changes with age, and how it may influence disease. Using a publicly available microarray dataset, we found that TSC1 and TSC2 gene expression was highest within the adult neo-cerebellum and that this pattern of increased cerebellar expression was maintained throughout postnatal development. During mid-gestational fetal development, however, TSC1 and TSC2 expression was highest in the cortical plate. Using a bioinformatics approach to explore protein and genetic interactions, we confirmed extensive connections between TSC1/TSC2 and the other genes that comprise the mammalian target of rapamycin (mTOR) pathway, and show that the mTOR pathway genes with the highest connectivity are also selectively expressed within the cerebellum. Finally, compared to age-matched controls, we found increased cerebellar volumes in pediatric TSC patients without current exposure to antiepileptic drugs. Considered together, these findings suggest that the cerebellum may play a central role in TSC pathogenesis and may contribute to the cognitive impairment, including the high incidence of autism spectrum disorder, observed in the TSC population.

A Fast Approach to Automatic Detection of Brain Lesions.

Template matching is a popular approach to computer-aided detection of brain lesions from magnetic resonance (MR) images. The outcomes are often sufficient for localizing lesions and assisting clinicians in diagnosis. However, processing large MR volumes with three-dimensional (3D) templates is demanding in terms of computational resources, hence the importance of the reduction of computational complexity of template matching, particularly in situations in which time is crucial (e.g. emergent stroke). In view of this, we make use of 3D Gaussian templates with varying radii and propose a new method to compute the normalized cross-correlation coefficient as a similarity metric between the MR volume and the template to detect brain lesions. Contrary to the conventional fast Fourier transform (FFT) based approach, whose runtime grows as O(N logN) with the number of voxels, the proposed method computes the cross-correlation in O(N). We show through our experiments that the proposed method outperforms the FFT approach in terms of computational time, and retains comparable accuracy.

Multi-modal robust inverse-consistent linear registration.

Registration performance can significantly deteriorate when image regions do not comply with model assumptions. Robust estimation improves registration accuracy by reducing or ignoring the contribution of voxels with large intensity differences, but existing approaches are limited to monomodal registration. In this work, we propose a robust and inverse-consistent technique for cross-modal, affine image registration. The algorithm is derived from a contextual framework of image registration. The key idea is to use a modality invariant representation of images based on local entropy estimation, and to incorporate a heteroskedastic noise model. This noise model allows us to draw the analogy to iteratively reweighted least squares estimation and to leverage existing weighting functions to account for differences in local information content in multimodal registration. Furthermore, we use the nonparametric windows density estimator to reliably calculate entropy of small image patches. Finally, we derive the Gauss-Newton update and show that it is equivalent to the efficient second-order minimization for the fully symmetric registration approach. We illustrate excellent performance of the proposed methods on datasets containing outliers for alignment of brain tumor, full head, and histology images.

Impact of MRI head placement on glioma response assessment.

Diagnosis of progressive disease or (partial) response during tumor treatment is based on manual size estimates of enhancing tumor area: an expert measures two perpendicular diameters of the enhancing tumor region in a single MRI slice with the largest enhancing area. This paper analyzes the reliability of the area measure with respect to head placement in the MRI scanner and compares it with 3D volume measures in a dataset of eight subjects (5-7 follow-up scans each) with high-grade glioma. We show that the manual area measure is highly sensitive to head position changes, with a root mean squared error of 22%, compared to volume estimates with less than 5% error. In our simulated study using the 2D manual measurements, the majority of subjects would have been incorrectly diagnosed with progressive disease without any true anatomical changes. These results highlight the urgent need for revised and more reliable response assessment criteria, for example, based on increased slice resolution, 3D volume analysis and percent change computation with respect to an average of patient specific longitudinal measurements instead of a single measurement to define progression or response.

Changes in cerebral cortex of children treated for medulloblastoma.

PURPOSE: Children with medulloblastoma undergo surgery, radiotherapy, and chemotherapy. After treatment, these children have numerous structural abnormalities. Using high-resolution magnetic resonance imaging, we measured the thickness of the cerebral cortex in a group of medulloblastoma patients and a group of normally developing children. METHODS AND MATERIALS: We obtained magnetic resonance imaging scans and measured the cortical thickness in 9 children after treatment of medulloblastoma. The measurements from these children were compared with the measurements from age- and gender-matched normally developing children previously scanned. For additional comparison, the pattern of thickness change was compared with the cortical thickness maps from a larger group of 65 normally developing children. RESULTS: In the left hemisphere, relatively thinner cortex was found in the perirolandic region and the parieto-occipital lobe. In the right hemisphere, relatively thinner cortex was found in the parietal lobe, posterior superior temporal gyrus, and lateral temporal lobe. These regions of cortical thinning overlapped with the regions of cortex that undergo normal age-related thinning. CONCLUSION: The spatial distribution of cortical thinning suggested that the areas of cortex that are undergoing development are more sensitive to the effects of treatment of medulloblastoma. Such quantitative methods may improve our understanding of the biologic effects that treatment has on the cerebral development and their neuropsychological implications.