MSBI Thesis & Abstracts 2017-2018
Sei Ahn
Advisor: Dr. Pratik Mukherjee, PhD
Thesis Title: Edge Density Imaging Characterization of White Matter Alterations and Overall Network Topology of 16p11.2 Deletion Carriers and Idiopathic ASD Individuals
Abstract:
David Baskin
Advisor: Dr. Stefanie Weinstein, PhD
Thesis Title: Advanced CEUS Techniques for MSK Imaging
Abstract:
Frank Chang
Advisor: Dr. Jessie Courtier, PhD
Thesis Title: Validatrion of an Augmented Reality System using MR Techniques
Abstract:
Xiao Gao
Advisor: Dr. Ashish Raj, PhD
Thesis Title: Neuroimaging-based Artificial Neural Network Predicts Conversion of Cognitive Impairment Spectrum in Alzheimer’s Disease
Abstract:
Alzheimer’s Disease (AD) represents the most frequent (60-80%) subtype of dementia and is one kind of progressive spectrum disorder without effective treatment so far. In the last decades, great efforts from all over the community have been made on the early diagnosis of AD at its preclinical stage, Mild Cognitive Impairment (MCI). Recently, a series of machine learning studies have successfully constructed several computational models in predicting conversion of cognitive impairment but seldom foresee beyond 4 years. Thanks to Alzheimer’s Disease Neuroimaging Initiative (ADNI) database, in this study we extracted cognition feature from several clinical outcomes. We then took advantage of structural MRI data and one Network Diffusion Model (NDM) raised by our group for subject-specific prediction of future cognition features. One supervised classification neural network was trained with ground-truth baseline and time-of- interest data but applied with predicted future cognition features. This established machine learning framework has demonstrated descent sensitivity and specificity in prediction of MCI-to- AD conversion (0.890 ± 0.083 and 0.923 ± 0.045) and healthy control (HC)-to-AD conversion (0.900 ± 0.074 and 0.744 ± 0.154) 5 years post baseline. To the best of our knowledge, we are the very first groups working on long-term prediction of AD spectrum conversion from both HC and MCI.
Matthew Gibbons
Advisor: Dr. Galateia Kazakia, PhD
Thesis Title: Identification of Blood Vessels in Cortical Bone Pores utilizing DCE-MRI and HR-pQCT
Abstract:
Purpose: Various diseases, such as Type II Diabetes (T2D), impact the microarchitecture of bones. DCE-MRI and HR-pQCT have been used to investigate cortical bone structural changes and to understand the role blood vessels have in the development of T2D-related pathological porosity in cortical bone. The purpose of this project was to look at the cortical bone porosity, and determine the contents of the pores (whether or not they had affiliated blood vessels), so as to better understand the mechanisms driving pathological bone porosity. To this end, image processing routines were developed to quantify bone porosity and to determine blood vessel location and volume fraction.
Methods: Results from an existing DCE-MRI and HR-pQCT image processing protocol (pipeline 1) were used in this study. As part of pipeline 1, the MRI and CT images underwent a global rigid registration. Pipeline 1 also provided masks for the CT cortical bone, the blood vessels, and the bone pores. Image pipeline 2 was developed through this thesis project. Pipeline 2 has a set of threshold and dilation/erosion steps to create a mask for the MRI cortical bone. It also performs non-rigid registration of bone masks and completes registration of vessels to pores with a piecewise rigid algorithm. The overlap of the registered vessel mask with the pore mask determines the final properties and biomarkers of the blood vessel network. Numerical and physical phantoms were used to characterize the algorithms by allowing a comparison to ground truth.
Results: In bone data artifacts are visible in the MRI bone masks. The imaging pipeline overcomes these and increases MRI to CT bone Dice coefficients from 0.8 to 0.9. Blood vessel alignment, as defined by vessel voxel overlap with pore voxels, is improved with vessel overlap increasing from 20% to 90%, and vessel voxel overlap increasing from 8% to 40%. Analysis of 12 distal and ultra-distal tibia data sets did not show a statistically significant difference between normal and T2D patients (mean ~ 0.3% for each with p = 0.4). A numerical phantom study provided metrics for pipeline 2 success. It indicated that after pipeline 1 offsets between vessels and pores should be < 15 voxels and pore densities should be < 20%. A physical phantom study showed capability to identify vessels with diameters as small as 300 um.
Conclusions: Positive results with bone and phantom data indicate a proof of concept for the general approach as well as the implemented algorithms. Pipeline 2 achieved high alignment fractions for blood vessels and pores. The analysis of both bone and phantom data has led to the definition of metrics and identification of specific algorithm deficiencies. Future work on these items should result in a robust image analysis pipeline for most data sets and a set of useful metrics to distinguish problematic data sets. These algorithm improvements along with analysis of more data sets will be needed to ascertain whether there are statistically significant differences between populations for cortical bone vessel densities.
Mahima Goel
Advisor: Dr. Srikantan Nagarajan, PhD
Thesis Title: Auditory Cortex Activity Modulation in Response to Sensory Feedback during Audiomotor Map Learning
Abstract:
Sensory feedback plays an important role in maintaining steady and fluent speech production. So far, a majority of research in speech has focused on auditory feedback while not a lot has been done on somatosensory feedback. Thus, the current study aims to further explore the effect of vocal tract somatosensory feedback on auditory cortex activity during the process of audiomotor map learning. Due to extensive evidence for a phenomenon known as motor-induced suppression (MIS), the current study hypothesizes that cortical activity will be reduced in subjects after the establishment of learning. Using an MEG touchscreen speech synthesizer set- up, subjects heard a target vowel sound and were asked to touch a location on the touchscreen that matched the sound they heard. With each trial, subjects were given feedback in the form of the sound that corresponded to the location on the screen they touched, resulting in them eventually learning to map out each vowel sound to a target location on the screen. This set-up allowed a paradigm to test for the effect of audiomotor map learning. Then, using the NUTMEG software and Champagne source localization algorithm, data from each subject was analyzed before and after learning, as well as any potential differences between auditory and motor feedback or left and right auditory cortices were noted. Findings across all 4 conditions (left auditory cortex with auditory feedback, right auditory cortex with auditory feedback, left auditory cortex with auditory and somatomotor feedback, right auditory cortex with auditory and somatomotor feedback) in almost all the subjects found a statistically significant decrease in cortical activity following the establishment of audiomotor map learning. The current study sets up future work in which a variety of patient populations as well as different forms of feedback can be studied using a touchscreen speech synthesizing platform.
Jeffrey Chieh Hsiao
Advisor: Dr. Michael Evans, PhD
Thesis Title: Measuring mTORC1 Signaling with Non-Invasive (89)Zr-Transferrin PET
Abstract:
The diagnosis and management of the phenotypically heterogeneous and progressive disorders tuberous sclerosis complex (TSC) and lymphangioleiomyomatosis (LAM) is a significant clinical challenge owing to a lack of disease specific biomarkers. Because the genetic mutations in TSC1 and/or TSC2 characteristic of TSC and LAM activate mTORC1 signaling, we hypothesized that disease burden could be measured with PET using 89Zr-transferrin (Tf). Toward this goal, we show that spontaneous renal cystadenomas arising in a genetically engineered mouse model heterozygous for Tsc2 were readily detectable with 89Zr-Tf PET. Moreover, subcutaneous implants of TSC and LAM cell line models consistently harbored high avidity for 89Zr-Tf in vivo. Deeper mechanistic studies showed that transferrin receptor expression and Tf biology were mTORC1 regulated in TSC and LAM models. Finally, the early treatment effects of clinically approved and experimental systemic therapies for TSC and LAM were interpretable using 89Zr-Tf PET. In summary, these data advance a translatable molecular imaging strategy that may be capable of detecting and longitudinally monitoring whole body TSC and LAM disease burden.
Molly Kadlec
Advisor: Dr. Duygu Tosun-Turgut, PhD
Thesis Title: Functional Magnetic Resonance Imaging Signature of Pain Anticipation and Clinical Depressive Symptomatology
Abstract:
Using functional magnetic resonance imaging (fMRI), this study tested the hypothesis that individuals diagnosed with major depressive disorder (MDD) are more likely to anticipate negative outcomes. The participants included thirty-one (15 females) unmedicated adults diagnosed with MDD and twenty-two (11 females) healthy control subjects with no history of MDD. fMRI data were collected during an event-related pain-anticipation paradigm, during which participants were cued to anticipate painful heat stimuli. Stimuli were delivered at high (moderately painful sensation) or low (mild painful sensation) intensity, and all cues were either known (high and low pain cues), or unknown (50% probability of high or low pain, which was not known to the participant), with a total of fourteen known and fourteen unknown anticipation trials. Based on prior findings regarding the importance of the insula cortex during pain anticipation, twelve insular regions (six on each side) were examined. Single-subject multi-voxel pattern analysis (MVPA) was used to create subject-specific activation maps for each anticipation condition. The mean activation (beta coefficient) was extracted within each insular region and was subjected to linear regression by way of a least absolute shrinkage and selection operator (LASSO). Following LASSO regression, prediction analysis was conducted on the test set (activation maps from fourteen unknown anticipation trials), based on the correlation with the training set (activation maps from fourteen known anticipation trials). Across each unknown anticipation trial, participants’ anticipation was classified as either high pain (HP) or low pain (LP). Several results were observed. First, we found that, within the chosen insular regions, neurobiological signatures of high and low pain anticipation were distinguishable at a high sensitivity. Second, the anterior short gyrus on the right side showed the highest deterministic beta coefficient in anticipatory predictions in our study. Third, we found that across both groups more of the unknown anticipatory conditions were labeled as HP. Fourth, in the current sample, we found no significant relationship with overall depressive symptoms severity and anticipatory labeling. Nevertheless, as hypothesized, significantly more cases of unknown anticipation were labeled as HP in the depressed cohort than in the control group (chi= 3.9; p < 0.05).
I-Ju Eric Lee
Advisor: Dr. Rong Wang, PhD
Thesis Title: Cerebral vascular and hemodynamic imaging with two-photon microscopy
Abstract:
A functional brain vascular supply is crucial for delivery of life-essential nutrients and removal of metabolites. Abnormal vascular development and hemodynamics can result in pathologies of many vascular diseases. The enlarged high-flow blood vessels that shunt blood from arteries to veins can cause arteriovenous malformation (AVM) and it can lead to life-threatening ruptures in the brain. The ability to correlate the relationship between blood flow with vascular structure at cell levels in living animals would foster our knowledge of the disease. Conventional wide-field microscopy is powerful in imaging at cell levels; however, light penetration depth is the limitation in deep tissue imaging. Near-infrared fluorescence imaging system can achieve deep tissue imaging but the low spatial resolution makes it difficult to map vascular structure and blood flow. In this research, we use two-photon laser scanning microscopy that can achieve deep tissue imaging and high spatial resolution to do cerebral vascular imaging in the genetic mutant mice that showing the phenotype of AVM. By using the line-scan imaging and 3D scan, we are able to analyze the blood velocity and diameter of the blood vessel that give the dynamic information of blood flow and vascular structure. We hope the findings through “5D” two-photon microscopy imaging that include high spatial vascular structure (3D) and blood velocity (4th dimension) over a period of time (5th dimension) can improve insights for the mechanism of AVM formation in the brain.
Jacob Mallott
Advisor: Dr. Pratik Mukherjee, PhD
Thesis Title: Diffusion Tensor Imaging Analysis of mTBI in Scholastic Athletes
Abstract:
Mild traumatic brain injury (mTBI) is a major public health concern, linked with post-concussive syndrome and chronic traumatic encephalopathy. At present, standard clinical imaging fails to reliably detect traumatic axonal injury associated with mTBI and post-concussive symptoms. Diffusion tensor imaging (DTI) is an MR imaging technique that is sensitive to changes in white matter microstructure. Prior studies using DTI to investigate mTBI did not separate contact sport athletes, a population at high risk for mTBI and subconcussive head traumas, and there has been a dearth of longitudinal studies of mTBI patients. In this study, we used Tract-Based Spatial Statistics to perform cross-sectional and longitudinal analysis describing changes in DTI scalar parameters in emergency room (ER) patients and in scholastic contact sport athletes. In the acute post-injury period, athletes demonstrated an elevated rate of regional decreases in axial diffusivity compared to controls. These decreases were especially pronounced in the cerebellar peduncles, and were more pronounced in contact sport athletes compared to the ER patient population. These results lend credence to the hypothesis that post-concussive symptoms are caused by shearing of axons of an attention network in the brain with timing mediated by the cerebellum, and warrant further study of the correlation between cerebellar DTI findings and clinical outcomes in mTBI patients.
Alex Nguyen
Advisor: Dr. Ashish Raj, PhD
Thesis Title: Applying Deep Leaming Models on the Brain Connectome for Improving Cognitive Predictions of Parkinson's Disease
Abstract:
Parkinson’s disease (PD) is one of the most prevalent progressive, neurodegenerative disorders that affects motor and cognitive function. It is characterized by tremors, rigidity and bradykinesia and eventually progresses to cognitive decline in late stages. Currently, there is no cure for PD. The standard therapy for treatment merely slows progression, to an extent, and provides temporary symptomatic relief. This is largely due to the lack of clinical biomarkers to successfully identify PD in early stages, resulting in a huge gap of knowledge surrounding the progression and disease stage. Recent advancements in MR clinical imaging have provided substantial anatomical datasets for subsets of populations affected by PD. In addition, there has been increasing focus on the implementation of artificial intelligence within the study of the brain network. In 2012, Raj et. al proposed the network diffusion model (NDM) based on the diffusion equation that provides analytical projections on baseline atrophy rate. In this study, we applied a deep learning model, autoencoder, to predict future motor and cognitive states using features acquired at baseline. We found that by implementing the autocoder with the NDM, prediction accuracy was improved when compared to using stepwise linear regression alone. This novel and innovative approach to neurodegenerative diseases, such as PD, has great potential for enhancing statistical power within the clinic, by providing clinicians to make more informed therapy decisions for PD patients. Not only does it reduce subjectivity, but it allows the clinician to assess motor and cognitive states at any given time point in the future.
Guangzhong Su
Advisor: Dr. Janine Lupo, PhD
Thesis Title: Incorporating Metabolic and Physiologic Imaging into Radiation Therapy Treatment Planning of Patients with Glioblastoma
Abstract:
Glioblastoma (GBM) remains the most aggressive cancer of the brain. Typical survival in patients with GBM is around 12-18 months, due to local or distant recurrence that occurs even after treatment with radiation and chemotherapy. Although there have been improvements in modern imaging, the radiation planning protocols are still based purely on a 2 cm geometric expansion of conventional post-contrast T1-weighted and T2-weighted FLAIR anatomic sequences. As a result, about 60% of tissue within the high dose treatment field can be normal brain tissue, which can damage to healthy brain function, while microscopic disease farther from the primary tumor bed is untreated. The main goal of this project was to develop a pipeline for the integration of probability maps derived from metabolic and diffusion-weighted MRI into the clinical workflow for radiation treatment planning. 24 patients with newly-diagnosed glioblastoma, who had undergone RT and chemotherapy consisting of an anti-angiogenic agent, were scanned at baseline prior to therapy and had serial follow-up imaging every 2 months until progression. Four patients were excluded because of either a poor initial model fit that resulted in inaccurate probability maps or poor image quality at the time of progression, leaving a total of 20 patients for evaluating our automatic contouring algorithm. First, we determined the optimal threshold for each patient’s probability map based on ROC analysis of the overlap of probability map with the progressed lesion. This value was then projected back on the histogram to automatically calculate each patient’s threshold based on the individual patient’s histogram and a maximum distance cutoff of 5 cm based on standard clinical procedures of high dose delivery during RT planning. Our results show that we were able to develop an automated contouring routine for integration of metabolic and physiology imaging into clinical workflow for radiation treatment planning. Incorporating a maximum distance from the original lesion allowed the automatic selection of a threshold from a consistent position on the histogram of the probability maps that optimized the overlap with the progressed lesion.
Kai Qiao Tiong
Advisor: Dr. Myriam Chaumeil, PhD
Thesis Title: Exploration of Early Metabolic Changes in Alzheimer's Disease as a Biomarker using Hyperpolarized C-13
Abstract:
Guillaume Trusz
Advisor: Dr. Henry Vanbrocklin, PhD
Thesis Title: Pharmacokinetics and Biodistribution of 89Zr-Df-PEG40 Nanoparticles in Xenograft Tumor Models using micro PET/CT
Abstract:
One of the challenges that physicians face is the uncertainty that their therapy of choice will be effective. A reason why therapeutic treatments often fall short in efficacy remains their limited presence and accumulation at the intended site of action. This challenge may be overcome by utilizing non-invasive molecular imaging. Attaching a radiolabel to the therapeutic, one may assess whether or not a drug accumulates at its target site, which may then allow one to more effectively predict whether patients may benefit from the treatment, thus offering patients a form of personalized medicine. Methods: Using a Zr-89 radionuclide and µPET/CT imaging technology, we assessed the pharmacokinetics (PK) of a novel chemotherapy tagged nanoparticle within various murine models engrafted with cell-line derived xenografts and patient-derived xenograft (PDX) tumors. Two other, non-therapeutic, versions of the nanoparticle were also radiolabeled and imaged so as to determine their potential as surrogate imaging probes for the therapeutic version. Tumor bearing mice were injected with 150 – 200 Ci of the Zr-89 labeled nanoparticles. Serial images were taken at 1, 24, 48, 72, 96, and 216 hours post-injection. Tumor and organ accumulation of the tracer were determined from the mouse µPET images. After the last imaging time-point, mice were euthanized and tumor, blood, and organs were weighted and counted. The percent injected dose per gram (%ID/g) of tissue was determined. Results: Nanoparticle radiolabeling was greater than 90% yield and 99% purity. µPET/CT time activity curves (TACs) showed very similar PK trends between the therapeutic and non-therapeutic nanoparticles. A steady accumulation followed by a plateauing of the nanoparticle concentrations within the growing tumors, as well as a clearance from all of the major organs (brain, heart, liver, and kidneys) was noted. Tumor maximum %ID/mL (5.81 – 18.14) occurred between the 72 and 96 hour time points, and ex vivo BioD experiments confirmed all of our in vivo PK findings (P > 0.05). Conclusion: Similar PK and BioD patterns between the therapeutic and non-therapeutic versions of the nanoparticle have validated that the imaging probes can in fact be used to predict the uptake and accumulation of the therapeutic nanoparticle.
Dyana Vega
Advisor: Dr. Benjamin Yeh, PhD
Thesis Title: Development of an anthropomorphic colon phantom to test the performance of dual energy computed tomography (DECT)
Abstract:
Colorectal cancer (CRC) is a very common malignancy that affects 1 in 20 Americans. Screening has been shown to reduce both incidence and mortality. When screening for CRC, identifying the polyp masses along the colon is key for the appropriate follow-up and treatment. To screen for CRC, Computer Tomographic Colonography (CTC) can be performed with a conventional 120kVp CT scan. However, a potentially sensitive approach to CTC, using dual energy CT (DECT) is emerging. The focus of this study was to create a colon phantom and use it to compare the performance of DECT to conventional CT in CTC. In creating the colon phantom, the attenuation spectral properties for DECT were matched to the real human abdomen, which included fat, iodine tagged stool and soft tissue (similar to polyps). The abdomen geometry was also simulated. The final colon phantom was scanned in conventional CT mode and in DECT at 40 keV, and the sensitivity, specificity and reading confidence were compared. Overall, DECT showed to have a higher sensitivity and increased reading confidence compared to conventional CT.