Thesis (Ph.D.)--Harvard--Massachusetts Institute of Technology Division of Health Sciences and Technology, 2001. / Includes bibliographical references (p. 205-216) and index. / The broad objective of this thesis is the development of statistically robust, computationally efficient, and global image processing algorithms. Such image processing algorithms are not only useful, but in high demand within the image processing arena. Recently, curve evolution and estimation-theoretic approaches to image processing have received considerable attention. Their role in the development of novel image processing algorithms is the focus of this thesis. The main contributions of this thesis lie in the development of three different, but interrelated, image processing algorithms with strong connections to curve evolution and estimation theory. One contribution of this thesis is the development of a new class of computationally-efficient algorithms designed to solve incomplete data problems in which part of the data is not observed, or hidden. These incomplete data problems are frequently encountered in image processing and computer vision. The basis of this framework is the marriage of the expectation-maximization procedure with two powerful methodologies-optimal multiscale estimators and mean field theory. Another contribution of this thesis is the development of a new class of deformable contour models for the segmentation of images which exhibit a known number of features. The key behind this approach is the use of geometric curve evolutions which maximally separate a predetermined set of statistics within the image. In addition, by introducing a geometric constraint on the segmenting curve, we modify this segmentation algorithm to produce a geometric clustering algorithm as well. / (cont.) The final contribution of this thesis is the development of an active contour model that offers a tractable implementation of the original Mumford-Shah model to simultaneously segment and smoothly reconstruct the data within a given image in a coupled manner. By generalizing the Mumford-Shah model, we are able to apply this active contour model to problems in which data quality varies between different locations in the image and, in the limiting case, to images in which pixel measurements are missing. We then modify this active contour model to obtain a novel PDE-based approach to image magnification, yielding a new application of the Mumford-Shah paradigm. Finally, we demonstrate the utility of this thesis by applying one of the image processing methodologies that we developed to a medical application, specifically, MR guided prostate brachytherapy. / by Andy Tsai. / Ph.D.

Thesis (Ph. D.)--Harvard-MIT Program in Health Sciences and Technology, 2012. / Vita. Cataloged from PDF version of thesis. / Includes bibliographical references. / Accurate sensory perception in noisy environments requires physiological mechanisms that reduce sensory interference. In the auditory system, it has been hypothesized that attentional control of cochlear responses to sound facilitates listening in noisy environments by modulating the effects of medial olivocochlear (MOC) efferent activity in the cochlea. However, conclusive support for this hypothesis has been elusive over the past 50 years. We investigated this issue using a novel experimental paradigm in which human subjects performed auditory tasks on transient sounds presented in acoustic noise, while we recorded click-evoked otoacoustic emissions (CEOAEs) in the task ear. CEOAEs are low-level sounds that are generated in the cochlea, recorded in the ear-canal and provide a non-invasive measure of the MOC effects on cochlear mechanical responses to sound. Our results show clear evidence that attending to transient sounds in noise caused an increase in MOC activity during the auditory task. MOC activity was greater on trials with correct responses compared to trials with incorrect responses, which provides evidence that the MOC activity brought about a perceptually beneficial change in cochlear operation. In addition, the task-dependent MOC activity scaled with auditory task difficulty and varied with task instructions. These results indicate the existence of a dynamic task-dependent interaction between the cochlea and the brain that has the function of optimizing cochlear operation to enhance auditory perceptual accuracy in noisy acoustic environments. / by Nikolas A. Francis. / Ph.D.

Thesis (Ph. D.)--Harvard University--MIT Division of Health Sciences and Technology, 2005. / Vita. / Includes bibliographical references. / From brainstem to cortex, sound is processed in centers that are functionally and structurally distinct. In animals, invasive electrophysiology and histology has revealed these distinctions and, consequently, organizational principles behind sound processing. In humans, however, comparable demonstrations are sparse. This thesis presents three MRI studies that provide new information regarding structural and functional distinctions between auditory centers in living humans. The first study compared the effect of a fundamental acoustic variable, sound level, on the population neural activity of auditory brainstem, thalamus and cortex. Brainstem and cortex exhibited contrasting sensitivities to sound level (growth in activation followed by saturation in brainstem vs. plateau then growth in cortex), with thalamus showing intermediate properties. The second study identified functional distinctions between cortical areas by spatially mapping the temporal properties of fMRI responses. Using a continuous noise stimulus, we found sustained responses on Heschl's gyrus flanked medially and laterally by more phasic activity. This pattern suggests that transient activity marking the beginning and end of a sound is most pronounced in non-primary areas of auditory cortex. The region of sustained responses may correspond to primary and primary-like areas. Thus, it may present a physiological marker for these areas in neuroimaging studies, something that has long been needed in the auditory neuroimaging field. The third study examined whether auditory cortical areas can be distinguished - in the living human brain - based on classical features of cortical gray matter previously resolvable only in postmortem tissue. / (cont.) By mapping the imaging parameter R1, we identified regions of heavily myelinated gray matter that may correspond to primary auditory cortex. We further found greater gray matter myelination of the left temporal lobe, which may be a substrate for higher fidelity temporal processing on the left, and for left-hemispheric speech and language specializations. Being able to resolve gray matter structure in-vivo opens the way to relating cortical physiology and structure directly in living humans in ways previously possible only in animals. / by Irina S. Sigalovsky. / Ph.D.

Thesis (Ph. D.)--Harvard-MIT Division of Health Sciences and Technology, 2011. / Cataloged from PDF version of thesis. / Includes bibliographical references (p. 126-134). / Anaphora in language is defined as an expression that refers to another expression. Hankamer & Sag 1976 and Sag & Hankamer 1984 proposed that anaphors can be divided into deep anaphors, which are resolved using a non-linguistic discourse-level interpretation of the antecedent, and surface anaphors, which are resolved by accessing their antecedents at a linguistic level which is highly determined by surface syntactic structure. Previous behavioral studies of the differences between deep and surface anaphors have conflicting and inconsistent results. Additionally, no neuroimaging studies have previously been conducted on deep and surface anaphors or on verb-phrase anaphora in general. Using two sets of materials which differed in whether they used a surface or a deep anaphor, the behavioral and neural responses as a function of anaphor type were determined. One set of materials was used to examine the effect of placing an intervening sentence between the antecedent and anaphor (distance materials), and one set was used to examine the effect of shifted word order (particle shift materials), both of which were expected to affect surface anaphors more than deep anaphors. Behavioral responses were measured using naturalness ratings and self-paced reading times, and neural responses were measured using blood-oxygenlevel dependent (BOLD) signal differences obtained with functional magnetic resonance imaging (fMRI). Increasing the distance between the antecedent and anaphor affected surface anaphors more than deep anaphors in naturalness ratings, question response times, and BOLD signal, while altering word order had similar or insignificant effects on surface and deep anaphors. These results are consistent with Sag and Hankamer's idea that surface and deep anaphora are distinct categories that are processed differently, but are not consistent with the exact level of access of surface anaphora proposed by Sag and Hankamer. Instead, the results suggest that surface anaphors are more dependent on syntactic information that decays over distance than deep anaphors, but do not differ from deep anaphors in terms of accessing exact surface word order information. / by Rebecca R. Woodbury. / Ph.D.

Thesis (Ph. D. in Medical Engineering)--Harvard-MIT Program in Health Sciences and Technology, 2013. / Cataloged from PDF version of thesis. Vita. / Includes bibliographical references. / Cell-based therapies, which rely on transplanted cells to restore function to damaged tissues, are currently under investigation in clinical trials. Stem and progenitor cells, including mesenchymal stem cells (MSCs), have shown potential in pre-clinical models to treat diseases ranging from connective tissue defects, through differentiating into bone or cartilage forming cells, to inflammatory conditions, through suppressing activated immune cells. While the ability of stem cells to differentiate into multiple lineages, secrete trophic factors, and modulate inflammatory processes has made them applicable to many diseases, these diverse functions also pose challenges in controlling their phenotype. In this thesis a new platform technology to influence the phenotype of cells is described and used to solve three critical challenges in MSCbased therapies, controlling MSC differentiation, tracking cells, and enhancing MSC's immunomodulatory potency. MSCs were found to efficiently and stably internalize micron-sized biodegradable particles. The platform can be tuned to specific applications through incorporation of phenotype altering drugs or other payloads into particles. In the first study, particles were loaded with a small molecule drug, dexamethasone (DEX), that induces MSC osteogenic differentiation. Modification of MSCs with DEX-particles resulted in differentiation of particle-laden cells to the same extent as those grown in osteogenic media. Furthermore, DEX was released from the cells in sufficient quantities to influence neighboring and distant cells demonstrating the particle platform can influence both the modified cell and its microenvironment. Next, the platform was adapted to address the need for longitudinal tracking of MSCs. Loading iron oxide nanoparticles in the microparticles resulted in enhanced tracking of MSCs by MRI from 6 days with nanoparticles alone to beyond 12 days with iron oxide microparticles. Finally, the novel discovery that glucocorticoid steroids significantly increase the immunomodulatory potency of MSCs by up-regulating expression of indoleamine-2,3- dioxygenase (IDO) is reported. Loading MSCs with particles containing the glucocorticoid steroid, budesonide, doubled their potency in suppressing activated peripheral blood mononuclear cell co-cultures in an IDO dependent manner. While the platform presented here was used to control, track, and augment MSCs, it can easily be tailored to control the function of other therapeutically relevant cells to develop next-generation cell-based therapies. / by James Allen Ankrum. / Ph.D.in Medical Engineering

Thesis (Ph. D.)--Harvard-MIT Division of Health Sciences and Technology, 2005. / Includes bibliographical references (leaves 145-155). / Solid tumors require blood vessels for growth, and the goal of anti-angiogenic therapy is to destroy the tumor vasculature. Recent findings suggest that anti-angiogenic therapy enhances radiation and chemotherapy responses. These findings seem paradoxical, since anti-angiogenic therapy prunes tumor vasculature while chemotherapy and radiation therapy rely on the vasculature to transport cancer drugs and oxygen, respectively, to cancer cells. To resolve this paradox, we propose that anti-angiogenic therapy can "normalize" the tumor vasculature transiently, resulting in a more efficient delivery of drugs and oxygen to cancer cells. We first show that DC101, a monoclonal antibody targeting Vascular Endothelial Growth Factor Receptor 2 (VEGFR2), prunes immature blood vessels, reduces vascular diameter and improves pericyte and basement membrane coverages. Functionally, the vascular permeability to macromolecules and interstitial fluid pressure are reduced. By lowering interstitial fluid pressure while maintaining microvascular pressure, DC101 induces a hydrostatic pressure gradient across the vascular wall, which leads to enhanced penetration of macromolecules in tumors. Tumor hypoxia is also reduced, and it is associated with the increased red blood cell velocity after DC101 treatment. Using gene array, real time PCR and Western blot analyses, changes in angiopoietin-2 level during DC101 treatment are identified. To test if similar effects happen in clinical setting, we obtained tumor biopsy samples from rectal adenocarcinoma patients treated with bevacizumab, an anti-VEGF monoclonal antibody. / (cont.) Our analysis shows that after bevacizumab treatment, microvascular density of the tumors decreases while pericyte coverage increases. The level of angiopoietin-2 also decreases, similar to the pre-clinical data. Thus our work shows a potential mechanism that explains the synergism between anti-angiogenic therapy and conventional therapies. These findings should facilitate the design of optimal dose and schedule of anti-angiogenic therapy. / by Ricky T. Tong. / Ph.D.

Thesis (Ph. D.)--Harvard-MIT Division of Health Sciences and Technology, February 2007. / Includes bibliographical references. / While being able to balance is something most of us take for granted, each year approximately 400,000 Americans are diagnosed with a balance disorder. In order to prevent fall-related injuries due to postural instability, it is important to create both diagnosis techniques so that therapy can be applied before a fall occurs and devices which can aid the balance-impaired population. The aims of this research are twofold: 1) to develop metrics that quantify the locomotor stability of individuals with reduced vestibular function and 2) to assess the capability of a noninvasive vibrotactile balance prosthesis for improving postural and gait stability. The clinical standards of practice for assessing vestibular deficiency include testing postural stability while standing but not during locomotion. This research examines one prospective locomotor-based technique involving the analysis of postural recovery from controlled surface perturbations. The research also investigates the use of a novel wearable vibrotactile sensory substitution device for enhanced postural and locomotor stability. The balance prosthesis is composed of an inertial motion-sensing system mounted on the lower back, a vibrotactile display worn around the torso, and a computer controller. / (cont.) It can serve as a permanent or temporary replacement of motion cues, a tool for vestibular rehabilitation, or an additional sensory channel for military troops, pilots, and astronauts. This research demonstrates that well-compensated vestibulopathic patients can be differentiated from young and age-matched controls during over ground locomotion based on step width variability. Prior to this research, unilateral and bilateral vestibulopathic patients donning the vibrotactile balance prosthesis have demonstrated increased postural stability during single-axis support surface perturbations using single-axis sway information. This work shows that multi-directional vibrotactile tilt feedback reduces postural sway during multi-directional support surface perturbations, and has both short- and long-term effects on increasing postural stability. Finally, this research demonstrates for the first time that medial-lateral (M/L) tilt feedback can be used by balance-deficient subjects to reduce factors associated with fall risk (M/L tilt and M/L step width variability) during various locomotor tasks. / by Kathleen H. Sienko. / Ph.D.

Thesis (Ph. D.)--Harvard--Massachusetts Institute of Technology Division of Health Sciences and Technology, 2000. / Includes bibliographical references (leaves 145-156). / Diagnostic imaging technologies for the detection of cancer include CT, MRI, ultrasonography, and endoscopy. However, many early neoplastic changes remain beyond their detection limits. A modality capable of imaging at or near the cellular level could detect disease at earlier stages than currently possible and thus improve patient prognosis. Optical coherence tomography (OCT) can achieve resolutions in the cellular and subcellular range (1-15 [mu]m) and could improve the diagnostic range of clinical imaging procedures. The research described in this thesis includes the ex vivo imaging surveys, technology development and clinical studies necessary to demonstrate the feasibility of OCT imaging for the in situ diagnosis of premalignant and neoplastic lesions. The first step in this endeavor was to evaluate the performance of OCT using ex vivo tissue specimens and identify the features in the OCT images which can be used for the diagnosis of disease. The development of the technologies necessary to introduce OCT to clinical settings, including a portable system and imaging devices, ensued. Enhancements in the post processing and visualization algorithms were introduced and the feasibility of ultra-high resolution and spectroscopic imaging was evaluated. In vivo OCT imaging, to assess the performance of the technology in clinical scenarios, followed in two systems, the cervical and oral mucosa. OCT can function as a type of optical biopsy to yield image information with resolutions approaching that of conventional histopathology in situ, without the need for excision and processing. OCT can be integrated to a wide range of clinical instruments including endoscopes, catheters, laparascopes, and surgical probes. A role for OCT is envisioned in clinical scenarios such as the follow-up management of patients with cervical intraepithelial neoplasia, on tamoxifen, with gastro-esophageal reflux or Barrett's and also guidance of surgical procedures in sensitive areas. Further improvements of the OCT technology will most likely enhance the abilities of the technique and may eventually lead to its establishment as a a powerful imaging modality for the diagnosis and management of cancer. / by Constantinos Pitris. / Ph.D.

Thesis (Ph. D.)--Harvard-MIT Division of Health Sciences and Technology, 2009. / Cataloged from PDF version of thesis. / Includes bibliographical references. / Vestibular Evoked Myogenic Potentials (VEMPs) are electrical signals recorded from the skin overlying skeletal muscles of the head and neck in response to high-intensity acoustic stimuli. VEMPs have been observed in stimulus locked averages of the electromyogram in a majority of human subjects, and are thought to originate in the otolith organs of the inner ear, which are balance organs responsible for sensing acceleration and orientation with respect to gravity. Otolith reflexes interact with the motor drive to a contracted muscle to give rise to the VEMP signal. In the last few years these signals have been used in the clinic as an indicator of peripheral vestibular function and a test based on VEMP from neck muscles (cervical, or cVEMP) is currently the only clinically feasible means of assessing the functioning of the saccule and its innervation. However, the usefulness of the test is limited by the inter-subject and test-retest variability of the response, and the unclear relationship between specific response features and vestibular pathophysiology. In this thesis, our goal is to measure VEMP variability, assess the influence of non-vestibular factors on the VEMP, and to develop a signal processing strategy to estimate response parameters that are both statistically stable, and physiologically meaningful. In the first part of the thesis, we systematically measure VEMPs from a small clinically normal population, and quantify the variability of the response, particularly the dependence on contraction effort. We also assess approaches to normalizing the response by estimates of the effort. / (cont.) In the second part of the thesis, we develop a computational model of VEMP physiology, and use the model to separate external sources of variability from internal noise. The model outputs are also used to define a statistical measure, the inverse coefficient of variation (iCOV), which correlates with altered vestibular sensitivity, but is relatively robust to other changes. When applied to the experimental data, the iCOV is found to yield estimates of vestibular sensitivity that are more stable than the conventional VEMP amplitude. This measure also reveals a diversity of response threshold and growth characteristics within the clinically normal population. These findings suggest that the proposed approach could lead to the development of an improved clinical tool, but could also yield new insights into the physiological mechanisms of vestibular pathology. / by Srinivasamurthy Ravi Prakash. / Ph.D.

Thesis (Ph. D.)--Harvard-MIT Division of Health Sciences and Technology, February 2004. / Includes bibliographical references (p. 138-142). / This electronic version was submitted by the student author. The certified thesis is available in the Institute Archives and Special Collections. / Normal-hearing listeners have a remarkable ability to hear in noisy environments, while hearing-impaired listeners and automatic speech-recognition systems often have difficulty in noise. With the ultimate goal of improving hearing aids and speech-recognition systems, we study the neural mechanisms involved in one aspect of noisy-environment listening, "spatial release from masking," which is the observation that a signal is more easily detected when its source is spatially separated from a masking-noise source. We use neurophysiology, computational modeling, and psychoacoustics to investigate the neural mechanisms of spatial release from masking, and we focus on low frequencies, which are important for speech recognition and are often spared in hearing-impaired listeners. Previous studies suggest that at low frequencies, listeners use interaural time differences (ITDs) to improve signal detection when signals and maskers are spatially separated in azimuth. To determine how individual neurons respond to spatially separated signals and maskers, we record in anesthetized cats from low-frequency, ITD-sensitive neurons in the inferior colliculus (IC), a major center of converging auditory pathways in the midbrain. We develop a computational model of the neuron responses, which incorporates both interaural cross-correlation (as used in existing binaural models) and amplitude-modulation sensitivity. The need for modulation sensitivity to predict the neural responses indicates that binaural and temporal processing are interacting in signal detection, rather than acting independently as is often assumed. This modification is especially important because most natural sounds, including speech, have pronounced envelope fluctuations that / (cont.) previous models of binaural detection have not utilized. To relate these neurophysiological results to human behavioral thresholds, we define population thresholds based on the most sensitive neurons in the population. The neural population thresholds are similar to human behavioral thresholds, indicating that low-frequency, ITD-sensitive neurons in the IC may be necessary for low-frequency spatial release from masking in humans. Both interaural correlation and modulation sensitivity seem to be required for the model population thresholds to predict human behavioral thresholds. Overall, our findings suggest that considering the auditory system's modulation sensitivity and interaural cross-correlation in the design of hearing aids and speech-recognition systems may improve these devices' performance in noise. / by Courtney C. Lane. / Ph.D.