Digital Commons Network^™

Deep Reinforcement Learning In Medical Object Detection And Segmentation, Dong Zhang

Electronic Thesis and Dissertation Repository

Medical object detection and segmentation are crucial pre-processing steps in the clinical workflow for diagnosis and therapy planning. Although deep learning methods have achieved considerable performance in this field, they impose several shortcomings, such as computational limitations, sub-optimal parameter optimization, and weak generalization. Deep reinforcement learning as the newest artificial intelligence algorithm has great potential to address the limitation of traditional deep learning methods, as well as obtaining accurate detection and segmentation results. Deep reinforcement learning has a cognitive-like process to propose the area of desirable objects, thereby facilitating accurate object detection and segmentation. In this thesis, we deploy deep …

Development Of Water-Soluble Polyesters For Tissue Engineering Applications, Trent Gordon

Electronic Thesis and Dissertation Repository

The development of tunable polymers has become increasingly important for both tissue engineering and drug delivery. This thesis investigates the development of water-soluble polyesters that contain both natural and synthetic components. These polymers offer tunable chemical structures, as well as functional groups for the conjugation of crosslinking moieties or cell signaling molecules. The first series of polymers was synthesized from poly(ethylene glycol) (PEG) and aspartic acid (Asp) via a titanium catalyzed transesterification method to provide polymers with molar masses of 12 kg/mol. After deprotection, the pendent functional groups of Asp were reacted with methacrylic, maleic, and itaconic anhydride to introduce …

Development And Validation Of Augmented Reality Training Simulator For Ultrasound Guided Percutaneous Renal Access, Yanyu Mu

Electronic Thesis and Dissertation Repository

Percutaneous renal access (PCA) is a critical step in needle-based renal procedures. Traditional PCA training relies on apprenticeship, which raises concerns about patient safety and limits training opportunities. In this thesis, we reviewed simulation-based training for PCA, described the development of a novel augmented reality (AR) simulator for ultrasound (US)-guided PCA, and evaluated its validity and efficacy as a teaching tool.

Our AR simulator allows the user to practice PCA on a silicone phantom using a tracked needle and US probe emulator under the guidance of simulated US on a tablet screen. 6 Expert and 24 novice participants were recruited …

Navigation And Patient Specific Instrumentation In Shoulder Arthroplasty, Joseph Cavanagh

Electronic Thesis and Dissertation Repository

The humeral head osteotomy influences shoulder arthroplasty component height, version and neck shaft angle and therefore outcomes. The purpose of this study is to develop and utilize navigation to execute the osteotomy.

3D printed models were used to develop a navigation technique, and then to execute planned osteotomies. Free hand, fixed angle guide, patient specific guides, and real time navigation osteotomies were completed. The height, neck shaft angle and version were recorded. Also collected, were the planes of the guides once they were placed.

Navigation had significantly less error from the planned osteotomy in neck shaft angle versus all other …

Lanthanide Nanoparticles As Contrast Agents For In Vivo Dual Energy Microcomputed Tomography Of The Mouse Vasculature, Charmainne Cruje

Electronic Thesis and Dissertation Repository

Dual energy (DE) computed tomography (CT) has the capability to influence medicine and pre-clinical research by providing quantitative information that can detect nascent lesions, identify perfusion restoration or inhomogeneities within tissues, and recognize the presence of calcium deposits. A wide variety of instrumentation techniques and scan protocols have been developed for DE CT, with a common goal of acquiring a pair of images that reports the attenuation of a given volume to two different x-ray distributions. While DE image acquisition has benefitted from technical advancements in CT, the contrast agents that are used are still predominantly composed of iodinated small …

A Simple, Realistic Walled Phantom For Intravascular And Intracardiac Applications., Hareem Nisar, John Moore, Roberta Piazza, Efthymios Maneas, Elvis C S Chen, Terry M Peters

Robarts Imaging Publications

PURPOSE: This work aims to develop a simple, anatomically and haptically realistic vascular phantom, compatible with intravascular and intracardiac ultrasound. The low-cost, dual-layered phantom bridges the gap between traditional wall-only and wall-less phantoms by showing both the vessel wall and surrounding tissue in ultrasound imaging. This phantom can better assist clinical tool training, testing of intravascular devices, blood flow studies, and validation of algorithms for intravascular and intracardiac surgical systems.

METHODS: Polyvinyl alcohol cryogel (PVA-c) incorporating a scattering agent was used to obtain vessel and tissue-mimicking materials. Our specific design targeted the inferior vena cava and renal bifurcations which were …

Simulation Approaches To X-Ray C-Arm-Based Interventions, Daniel R. Allen

Electronic Thesis and Dissertation Repository

Mobile C-Arm systems have enabled interventional spine procedures, such as facet joint injections, to be performed minimally-invasively under X-ray or fluoroscopy guidance. The downside to these procedures is the radiation exposure the patient and medical staff are subject to, which can vary greatly depending on the procedure as well as the skill and experience of the team. Standard training methods for these procedures involve the use of a physical C-Arm with real X-rays training on either cadavers or via an apprenticeship-based program. Many guidance systems have been proposed in the literature which aim to reduce the amount of radiation exposure …

Computational Modeling Of The Human Brain For Mtbi Prediction And Diagnosis, Yanir Levy

Electronic Thesis and Dissertation Repository

Sports related concussions and mild traumatic brain injuries have seen an increase in frequency over the past decade. The creation of highly biofidelic computational head models is an important step in understanding the mechanisms of these mild brain injuries and preventing them. Hence, the purpose of this research is to combine state-of-the-art computational models, brain imaging modalities and traditional head injury assessment protocols to simulate and predict the brains responses during traumatic head impacts. A novel, atlas-based, parcellated axon fiber embedded head model was developed which allows for in-depth analysis of the brain’s structural connectome tracts for injury diagnosis and …

A Biomechanical Investigation Into The Effect Of Experimental Design On Wrist Biomechanics And Contact Mechanics, Clare E. Padmore

Electronic Thesis and Dissertation Repository

The wrist is one of the most commonly injured joints, and injury can have serious sequelae if pathological healing ensues. Strides have been made to understand normal and pathological wrist biomechanics using experimental approaches, which has contributed to improved patient care. The present work advances our understanding of the influence of experimental techniques and joint motion measurement techniques on in-vitro wrist biomechanical cadaveric studies, and applies the knowledge learned to a common clinical entity of scapholunate insufficiency.

First, the relative contributions of the carpal rows to wrist motion were assessed, in addition to the identification of limitations of current biomechanical …

Development Of An Active Infection Monitoring Knee Spacer For Two-Stage Revision, Michael K. Lavdas

Electronic Thesis and Dissertation Repository

Infections requiring surgical revision occur in ~3% of total knee arthroplasty (TKA) procedures. The standard treatment for TKA infection in North America is referred to as the two-stage revision process. Unfortunately, the success rate of the two-stage revision is low, at roughly 85%. There are indications that the two-stage revision may fail to clear the infection due to unsatisfactory diagnostic tools. This thesis depicts the development and evaluation of a novel telemetric sensing package designed for integration into the two-stage revision process. Studies within evaluate its performance through sensor validation, thermal insulation testing, radio frequency penetration evaluation, a cadaveric loading …

Detecting Command-Driven Brain Activity In Patients With Disorders Of Consciousness Using Tr-Fnirs, Androu Abdalmalak

Electronic Thesis and Dissertation Repository

Vegetative state (VS) is a disorder of consciousness often referred to as “wakefulness without awareness”. Patients in this condition experience normal sleep-wake cycles, but lack all awareness of themselves and their surroundings. Clinically, assessing consciousness relies on behavioural tests to determine a patient’s ability to follow commands. This subjective approach often leads to a high rate of misdiagnosis (~40%) where patients who retain residual awareness are misdiagnosed as being in a VS. Recently, functional neuroimaging techniques such as functional magnetic resonance imaging (fMRI), has allowed researchers to use command-driven brain activity to infer consciousness. Although promising, the cost and accessibility …

Cardiac Modelling Techniques To Predict Future Heart Function And New Biomarkers In Acute Myocardial Infarction, Sergio C. H. Dempsey

Electronic Thesis and Dissertation Repository

Fundamental to treatment planning for patients that have suffered myocardial infarction are predictive biomarkers and risk factors. Important among these in terms of a patient’s treatment plan or prognosis are the contractility of the damaged myofibers, final infarct volume, and poor infarct healing rate. Proposed and developed in this thesis are techniques to predict these biomarkers and risk factors using cardiac biomechanical modelling. One of the developed techniques was a CT compatible shape optimization technique which can predict the contraction force of healthy, and stunned myofibers within 6.3% and the distribution of potentially necrotic myofibers within 10% accuracy. The second …

3d Culture Strategies For The Dynamic Expansion And Preconditioning Of Adipose-Derived Stromal Cells On Decellularized Adipose Tissue Bioscaffolds, Tim Tian Han

Electronic Thesis and Dissertation Repository

Adipose tissue engineering holds promise for the development of therapeutic strategies for subcutaneous adipose tissue regeneration to treat defects resulting from congenital birth defects, invasive surgical procedures and traumatic injuries. Decellularized adipose tissue (DAT) scaffolds represent a potential off-the-shelf tissue substitute for volume augmentation. Seeding the DAT with adipose-derived stromal cells (ASCs) has been shown to enhance adipose tissue regeneration in immunocompetent animals in vivo. Although promising, this strategy is limited by low cell attachment on the DAT. As such, this thesis focused on the development of bioreactor strategies to enhance the capacity of human ASCs to stimulate angiogenesis …

Video Processing For The Evaluation Of Vascular Dynamics In Neurovascular Interventions, Reid Vassallo

Electronic Thesis and Dissertation Repository

An arteriovenous malformation (AVM) is an abnormal collection of blood vessels which causes blood to travel from arteries to veins through an abnormal twisted network of vessels. This network has an elevated risk of rupture, which can lead to permanent disability and death if the rupture occurs in the brain. The gold standard treatment for AVM is surgical resection, and it is crucial to know which vessels are bringing blood towards and away from the AVM. Unfortunately, it is almost impossible to know this by looking at the surgical scene. The primary limitations of current methods to address this are …

Structure-Function Relationships In The Brain: Applications In Neurosurgery, Daiana-Roxana Pur

Electronic Thesis and Dissertation Repository

Multimodal brain imaging allows the study of structure-function relationships of the brain at the individual level, a key subject in basic neuroscience with important applications in neurosurgery. The current thesis aims to better understand these relationships by (1) examining how cortical morphology metrics influence measures of brain function, (2) their visualization in augmented reality (AR), and (3) their application in neurosurgical planning. To achieve these objectives, we made use of multimodal magnetic resonance imaging (MRI) data: diffusion weighted imaging, resting-state functional MRI (rs-fMRI), task-based fMRI, and T1-weighted images. Various metrics were calculated: cortical thickness (CT), blood oxygen level dependent signal …

Machine Learning Towards General Medical Image Segmentation, Clara Tam

Electronic Thesis and Dissertation Repository

The quality of patient care associated with diagnostic radiology is proportionate to a physician's workload. Segmentation is a fundamental limiting precursor to diagnostic and therapeutic procedures. Advances in machine learning aims to increase diagnostic efficiency to replace single applications with generalized algorithms. We approached segmentation as a multitask shape regression problem, simultaneously predicting coordinates on an object's contour while jointly capturing global shape information. Shape regression models inherent point correlations to recover ambiguous boundaries not supported by clear edges and region homogeneity. Its capabilities was investigated using multi-output support vector regression (MSVR) on head and neck (HaN) CT images. Subsequently, …

Pulmonary Imaging Of Chronic Obstructive Pulmonary Disease Using Multi-Parametric Response Maps, Jonathan Macneil

Electronic Thesis and Dissertation Repository

Chronic obstructive pulmonary disease (COPD) is characterized by irreversible airflow obstruction caused by airway remodelling and parenchymal destruction. Clinically, observation of COPD is performed using spirometry, but this technique only provides a global measure of lung health. To supplement these clinical measurements, thoracic computed tomography (CT) and hyperpolarized gas magnetic resonance imaging (MRI) have been used to measure regional structure and function abnormalities. Although CT and MRI have been used to research COPD, combination of both modalities into an interrelated image has never been performed. Therefore, we developed an image processing pipeline to combine MRI-CT information into a multi-parametric response …