Faculty Bibliography

Quantitative, volumetric analysis of Magnetic Resonance Imaging (MRI) is a fundamental way researchers study the brain in a host of neurological conditions including normal maturation and aging. Despite the availability of open-source brain segmentation software, widespread clinical adoption of volumetric analysis has been hindered due to processing times and reliance on manual corrections. Here, we extend the use of deep learning models from proof-of-concept, as previously reported, to present a comprehensive segmentation of cortical and deep gray matter brain structures matching the standard regions of aseg+ aparc included in the commonly used open-source tool, Freesurfer. The work presented here provides a real-life, rapid deep learning-based brain segmentation tool to enable clinical translation as well as research application of quantitative brain segmentation. The advantages of the presented tool include short (~ 1 minute) processing time and improved segmentation quality. This is the first study to perform quick and accurate segmentation of 102 brain regions based on the surface-based protocol (DMK protocol), widely used by experts in the field. This is also the first work to include an expert reader study to assess the quality of the segmentation obtained using a deep-learning-based model. We show the superior performance of our deep-learning-based models over the traditional segmentation tool, Freesurfer. We refer to the proposed deep learning-based tool as DARTS (DenseUnet-based Automatic Rapid Tool for brain Segmentation)

In this work, we propose bag of adversarial features (BAF) for identifying mild traumatic brain injury (MTBI) patients from their diffusion magnetic resonance images (MRI) (obtained within one month of injury) by incorporating unsupervised feature learning techniques. MTBI is a growing public health problem with an estimated incidence of over 1.7 million people annually in US. Diagnosis is based on clinical history and symptoms, and accurate, concrete measures of injury are lacking. Unlike most of previous works, which use hand-crafted features extracted from different parts of brain for MTBI classification, we employ feature learning algorithms to learn more discriminative representation for this task. A major challenge in this field thus far is the relatively small number of subjects available for training. This makes it difficult to use an end-to-end convolutional neural network to directly classify a subject from MR images. To overcome this challenge, we first apply an adversarial auto-encoder (with convolutional structure) to learn patch-level features, from overlapping image patches extracted from different brain regions. We then aggregate these features through a bag-of-word approach. We perform an extensive experimental study on a dataset of 227 subjects (including 109 MTBI patients, and 118 age and sex matched healthy controls), and compare the bag-of-deep-features with several previous approaches. Our experimental results show that the BAF significantly outperforms earlier works relying on the mean values of MR metrics in selected brain regions.

BACKGROUND AND PURPOSE/OBJECTIVE:Working memory impairment is one of the most troubling and persistent symptoms after mild traumatic brain injury (MTBI). Here we investigate how working memory deficits relate to detectable WM microstructural injuries to discover robust biomarkers that allow early identification of patients with MTBI at the highest risk of working memory impairment. MATERIALS AND METHODS/METHODS:Multi-shell diffusion MR imaging was performed on a 3T scanner with 5 b-values. Diffusion metrics of fractional anisotropy, diffusivity and kurtosis (mean, radial, axial), and WM tract integrity were calculated. Auditory-verbal working memory was assessed using the Wechsler Adult Intelligence Scale, 4th ed, subtests: 1) Digit Span including Forward, Backward, and Sequencing; and 2) Letter-Number Sequencing. We studied 19 patients with MTBI within 4 weeks of injury and 20 healthy controls. Tract-Based Spatial Statistics and ROI analyses were performed to reveal possible correlations between diffusion metrics and working memory performance, with age and sex as covariates. RESULTS:= .04), mainly present in the right superior longitudinal fasciculus, which was not observed in healthy controls. Patients with MTBI also appeared to lose the normal associations typically seen in fractional anisotropy and axonal water fraction with Letter-Number Sequencing. Tract-Based Spatial Statistics results also support our findings. CONCLUSIONS:Differences between patients with MTBI and healthy controls with regard to the relationship between microstructure measures and working memory performance may relate to known axonal perturbations occurring after injury.

Mild traumatic brain injury is a growing public health problem with an estimated incidence of over 1.7 million people annually in US. Diagnosis is based on clinical history and symptoms, and accurate, concrete measures of injury are lacking. This work aims to directly use diffusion MR images obtained within one month of trauma to detect injury, by incorporating deep learning techniques. To overcome the challenge due to limited training data, we describe each brain region using the bag of word representation, which specifies the distribution of representative patch patterns. We apply a convolutional auto-encoder to learn the patch-level features, from overlapping image patches extracted from the MR images, to learn features from diffusion MR images of brain using an unsupervised approach. Our experimental results show that the bag of word representation using patch level features learnt by the auto encoder provides similar performance as that using the raw patch patterns, both significantly outperform earlier work relying on the mean values of MR metrics in selected brain regions.

We seek to elucidate the underlying pathophysiology of injury sustained after mild traumatic brain injury (MTBI) using multi-shell diffusion MRI, deriving compartment-specific WM tract integrity (WMTI) metrics. WMTI allows a more biophysical interpretation of WM changes by describing microstructural characteristics in both intra- and extra-axonal environments. Thirty-two patients with MTBI within 30 days of injury and twenty-one age- and sex-matched controls were imaged on a 3T MR scanner. Multi-shell diffusion acquisition was performed with 5 b-values (250 - 2500 s/mm²) along 6 - 60 diffusion encoding directions. Tract-based spatial statistics (TBSS) was used with family-wise error (FWE) correction for multiple comparisons. TBSS results demonstrate focally lower intra-axonal diffusivity (D_axon) in MTBI patients in the splenium of the corpus callosum (sCC) (p < 0.05, FWE-corrected). The Area Under the Curve (AUC)-value for was 0.76 with low sensitivity of 46.9%, but 100% specificity. These results indicate that D_axon may be a useful imaging biomarker highly specific for MTBI-related WM injury. The observed decrease in D_axon suggests restriction of the diffusion along the axons occurring shortly after injury.

Working memory is a complex cognitive process at the intersection of sensory processing, learning, and short-term memory and also has a general executive attention component. Impaired working memory is associated with a range of neurological and psychiatric disorders, but very little is known about how working memory relates to underlying white matter (WM) microstructure. In this study, we investigate the association between WM microstructure and performance on working memory tasks in healthy adults (right-handed, native English speakers). We combine compartment specific WM tract integrity (WMTI) metrics derived from multi-shell diffusion MRI as well as diffusion tensor/kurtosis imaging (DTI/DKI) metrics with Wechsler Adult Intelligence Scale-Fourth Edition (WAIS-IV) subtests tapping auditory working memory. WMTI is a novel tool that helps us describe the microstructural characteristics in both the intra- and extra-axonal environments of WM such as axonal water fraction (AWF), intra-axonal diffusivity, extra-axonal axial and radial diffusivities, allowing a more biophysical interpretation of WM changes. We demonstrate significant positive correlations between AWF and letter-number sequencing (LNS), suggesting that higher AWF with better performance on complex, more demanding auditory working memory tasks goes along with greater axonal volume and greater myelination in specific regions, causing efficient and faster information process.

Remarkable advances have been made in the last decade in the use of diffusion MR imaging to study mild traumatic brain injury (mTBI). Diffusion imaging shows differences between mTBI patients and healthy control groups in multiple different metrics using a variety of techniques, supporting the notion that there are microstructural injuries in mTBI patients that radiologists have been insensitive to. Future areas of discovery in diffusion MR imaging and mTBI include larger longitudinal studies to better understand the evolution of the injury and unravel the biophysical meaning that the detected changes in diffusion MR imaging represent.

The effect of T1 signal on FSL voxel-based morphometry modulated GM density and FreeSurfer cortical thickness is explored. The techniques rely on different analyses, but both are commonly used to detect spatial changes in GM. Standard pipelines show FSL voxel-based morphometry is sensitive to T1 signal alterations within a physiologic range, and results can appear discordant between FSL voxel-based morphometry and FreeSurfer cortical thickness. Care should be taken in extrapolating results to the effect on brain volume.

BACKGROUND AND PURPOSE: Obtaining high-resolution brain MR imaging in patients with a previously implanted deep brain stimulator has been challenging and avoided by many centers due to safety concerns relating to implantable devices. We present our experience with a practical clinical protocol at 1.5T by using 2 magnet systems capable of achieving presurgical quality imaging in patients undergoing bilateral, staged deep brain stimulator insertion. MATERIALS AND METHODS: Protocol optimization was performed to minimize the specific absorption rate while providing image quality necessary for adequate surgical planning of the second electrode placement. We reviewed MR imaging studies performed with a minimal specific absorption rate protocol in patients with a deep brain stimulator in place at our institution between February 1, 2012, and August 1, 2015. Images were reviewed by a neuroradiologist and a functional neurosurgeon. Image quality was qualitatively graded, and the presence of artifacts was noted. RESULTS: Twenty-nine patients (22 with Parkinson disease, 6 with dystonia, 1 with essential tremor) were imaged with at least 1 neuromodulation implant in situ. All patients were imaged under general anesthesia. There were 25 subthalamic and 4 globus pallidus implants. Nineteen patients were preoperative for the second stage of bilateral deep brain stimulator placement; 10 patients had bilateral electrodes in situ and were being imaged for other neurologic indications, including lead positioning. No adverse events occurred during or after imaging. Mild device-related local susceptibility artifacts were present in all studies, but they were not judged to affect overall image quality. Minimal aliasing artifacts were seen in 7, and moderate motion, in 4 cases on T1WI only. All preoperative studies were adequate for guidance of a second deep brain stimulator placement. CONCLUSIONS: An optimized MR imaging protocol that minimizes the specific absorption rate can be used to safely obtain high-quality images in patients with previously implanted deep brain stimulators, and these images are adequate for surgical guidance.