Thesis (S.M.)--Harvard-MIT Division of Health Sciences and Technology, 2004. / Includes bibliographical references (leaves 66-67). / In order to automate data extraction from electronic medical documents, it is important to identify the correct context of the extracted information. Context in medical documents is provided by the layout of documents, which are partitioned into sections by virtue of a medical culture instilled through common practice and the training of physicians. Unfortunately, formatting and labeling is inconsistently adhered to in practice and human experts are usually required to identify sections in medical documents. A series of experiments tested the hypothesis that section identification independent of the label on sections could be achieved by using a neural network to elucidate relationships between features of sections (like size, position from start of the document) and the content characteristic of certain sections (subject-specific strings). Results showed that certain sections can be reliably identified using two different methods, and described the costs involved. The stratification of documents by document type (such as History and Physical Examination Documents or Discharge Summaries), patient diagnoses and department influenced the accuracy of identification. Future improvements suggested by the results in order to fully outline the approach were described. / by Reejis Stephen. / S.M.

Thesis (Ph.D.)--Harvard--Massachusetts Institute of Technology Division of Health Sciences and Technology, 1996. / Includes bibliographical references. / This thesis examines the problem of resource allocation in Africa to combat the HIV/AIDS epidemic. It focuses on the use of one specific technology, the anti-HIV antibody test. After describing the characteristics of the epidemic, the special problems that accompany the allocation of health resources in Africa are explored. A short description of the biological and technical aspects of HIV testing is followed by three case studies which examine different uses of the technology. (1) A model of the use of HIV testing to screen blood donors is demonstrated in several hypothetical situations to evaluate under which circumstances HIV screening is cost-effective and which is the most cost-effective of a number of testing systems. (2) Use of the HIV test is considered from a cost-effectiveness perspective for the purpose of helping to confirm the diagnosis of HIV-related disease. Possible benefits of testing (including more rapid initiation of appropriate treatment, avoidance of the cost and iatrogenic complications of inappropriate treatment, and more efficient rationing of health care resources) are compared to possible costs (including monetary costs, emotional costs, and costs associated with false test results). A detailed protocol is presented of a prospective study to evaluate the appropriate use of the HIV test in the inpatient hospital setting. (3) Serologic surveys, including procurement of samples and testing for HIV, comprise the bulk of any program to monitor and characterize the epidemiology of the HIV/AIDS epidemic. / (cont.) The descriptive data thus gathered can then be used to more effectively target interventions to prevent further HIV spread and alleviate the impact of HIV/AIDS. Two models are presented which can be used to optimize the cost-efficiency of serologic surveys by improving the selection of sample size and testing method. The conclusions draw upon the findings of the three case studies. They place the fight against AIDS in its social and economic context in Africa; outline the general rules that govern the use of HIV testing technology; and underscore the need for cost-effectiveness assessments to improve the efficiency of resource allocation by national AIDS programs. / by Stefano Michele Bertozzi. / Ph.D.

Thesis (S.M.)--Harvard-MIT Division of Health Sciences and Technology, 2005. / Includes bibliographical references (leaves 46-47). / There are various kinds of clinical documents used in a hospital or clinic setting. With the emergence of Electronic Medical Records, efforts are being made to computerize these documents in a structured fashion in order to enable decision support. With structured data entry, because each fact about the patient is stored discretely and can be retrieved separately, information can be organized and presented in different ways, depending on the needs of the user. A typical structured clinical document contains a range of findings recorded by a physician, nurse or other care. These findings can be thought of as discrete pieces of information, called observations. These observations can be grouped together to form observation sets that can be placed under relevant headers within the document. When building information systems that support structured clinical documentation, these observations and sets are created and stored in catalogs. My thesis addresses the issue of building an ontology model for clinical documentation that supports the creation and management of an observations catalog, observation sets catalog and a clinical document catalog. The ontology can be used as an organizational tool for efficient maintenance of these catalogs. By tagging observations and observation sets with relevant attributes, it is possible to generate intelligent displays of data that are more flexible and dynamic. / by Joyce George. / S.M.

Thesis (S.M.)--Harvard-MIT Division of Health Sciences and Technology, 2005. / Includes bibliographical references (p. 46-49). / The pars tensa of the tympanic membrane is composed of three layers: an epidermal layer, a fibrous layer, and a mucosal layer. Recent studies (Kuijpers et al, 1999; Henson and Henson, 2000; Henson et al, 2005) suggest that the fibrous layer in several mammalian species contains contractile fibers, which are located primarily within the thickened border of the pars tensa known as the annulus fibrosis. These contractile fibers resemble smooth muscle fibers. Yang and Henson (2002) studied the physiological effects of pharmacological modulators on the pars tensa of the annulus fibrosis by measuring the sound-induced cochlear response. Their results suggest a dose-dependent change in cochlear response after application of sodium orthovanadate and norepinephrine. Application of saline induced no change in cochlear response. Based on their data, Yang and Henson proposed that the pharmacological agents altered the function of the smooth muscle fibers of the annulus fibrosis to produce a mechanical change in the tympanic membrane. In this study two measurements, cochlear response and Laser Doppler Vibrometry, were used to assess the sound-induced velocity of the tympanic membrane of the gerbil before and after application of saline and varying concentrations of three smooth muscle antagonists (sodium orthovanadate, norepinephrine, and carbachol) to the pars tensa. It was demonstrated that applications of saline and varying concentrations of sodium orthovanadate were associated with both increases and decreases in the magnitude of the cochlear response in two out of three ears tested. There was no evidence of a dose-dependent change in the cochlear response. / (cont.) Applications of saline and varying concentrations of sodium orthovanadate, norepinephrine, and carbachol were associated with increases and decreases in the magnitude of the Laser Doppler Vibrometry response in eight of fourteen ears tested. Evidence of a dose-dependent change in Laser Doppler Vibrometry results was obtained in one ear. The results of this study suggest that application of any substance to the tympanic membrane may or may not be associated with an increase or decrease in the cochlear response or Laser Doppler Vibrometry response, and thus, the source of mechanical changes observed at the tympanic membrane is not necessarily the smooth muscle fibers of the annulus fibrosis. / by Amanda J. Graves. / S.M.

Thesis (Ph. D.)--Harvard-MIT Division of Health Sciences and Technology, 2008. / Includes bibliographical references (p. 131-139). / The practice of brachytherapy and brachytherapy dosimetry was investigated with emphasis on evaluations of dose distributions and shielding considerations for both photon- and neutron-emitting radionuclides. Monte Carlo simulation methods were employed to calculate dose distributions for virtual and commercial brachytherapy sources. Radionuclides studied were 103Pd, 1251, 131Cs, 137Cs, 169b, 192Ir, and 252Cf. 252Cf sources also emit neutrons from spontaneous fission. The brachytherapy dosimetry protocol recommended by the American Association of Physicists in Medicine was followed and evaluated for conditions of partial scatter (non-infinite media) and material inhomogeneities, both commonly encountered in brachytherapy treatment. Furthermore, energy-dependent characteristics of dosimetry parameters were evaluated and reference calculations performed for virtual photon and neutron sources. These findings were applied to three clinical brachytherapy cases: eye plaques using 103Pd, 125I, and 131Cs; high-dose rate 252Cf treatment; and, 2 Cf plaques for superficial lesions. For eye plaques, material heterogeneities were significant for each radionuclide with dose reduction at 5 mm of 18%, 11%, and 10% for P03pd, 125I, and 131Cs, respectively. For a proposed highdose rate 252Cf source (5mm length), relative brachytherapy dosimetry parameters were found to be similar to those obtained for a low-dose rate Applicator Tube-type source (15 mm length). Considering 252Cf plaque brachytherapy when partial scatter conditions were accounted for, central axis equivalent dose rate decreased by 11 ± 1% and 7 ± 2% for depths of 4 to 50 mm, respectively. / (cont.) The ratio of neutron dose to total physical dose was 70 ± 1% and 57 ± 2% for depths of 4 and 50 mm, respectively, while the fractional dose-equivalent due to neutrons was 93 + 1% and 89 ± 2% at these depths, respectively. Finally, shielding requirements for a clinical high-dose rate 252Cf source were explored for common shielding materials and a linear accelerator vault. Lead, polyethylene, and borated polyethylene were evaluated for neutron, primary photon, and secondary photon attenuation. Half-value layers of 0.70, 0.15, and 0.13 m were obtained for lead, polyethylene, and borated polyethylene, respectively. A linear accelerator vault was found to adequately shield up to a 5 mg 252Cf source for regular clinical use. / by Christopher S. Melhus. / Ph.D.

Thesis (Ph. D.)--Harvard-MIT Division of Health Sciences and Technology, 2008. / Includes bibliographical references. / Developing a consensus on how to model neural and behavioral responses and to quantify important response properties is a challenging signal processing problem because models do not always adequately capture the data and different methods often yield different estimates of the same response property. The threshold, the first stimulus level for which a difference between baseline activity and stimulus-driven activity exists, is an example of such a response property for both neural and behavioral responses.In the first and second sections of this work, we show how the state-space model framework can be used to represent neural and behavioral responses to auditory stimuli with a high degree of model goodness-of-fit. In the first section, we use likelihood methods to develop a state-space generalized linear model and estimate maximum likelihood parameters for neural data. In the second section, we develop the alternative Bayesian state-space model for behavioral data. Based on the estimated joint density, we then illustrate how important response properties, such as the neural and behavioral threshold, can be estimated, leading to lower threshold estimates than current methods by at least 2 dB. Our methods provide greater sensitivity, obviation of the hypothesis testing framework, and a more accurate description of the data.Formulating appropriate models to describe neural data in response to natural sound stimulation is another problem that currently represents a challenge. In the third section of the thesis, we develop a generalized linear model for responses to natural sound stimuli and estimate maximum likelihood parameters. Our methodology has the advantage of describing neural responses as point processes, capturing aspects of the stimulus response such as past spiking history and estimating the contributions of the various response covariates, resulting in a high degree of model goodness-of-fit. / (cont) Using our model parameter estimates, we illustrate that decoding of the natural sound stimulus in our model framework produces neural discrimination performance on par with behavioral data.These findings have important implications for developing theoretically-sound and practical definitions of the neural response properties, for understanding information transmission within the auditory system and for design of auditory prostheses. / by Anna A. Dreyer. / Ph.D.

Thesis (S.M.)--Harvard-MIT Division of Health Sciences and Technology, 2010. / Cataloged from PDF version of thesis. / Includes bibliographical references (p. 78-84). / Autism spectrum disorder (ASD) is characterized by many behavioral symptoms, including delays in social and communicative development. A cluster of symptoms concentrate on speech and language development, especially manipulation of non-verbal information conveyed in prosody. It is largely unknown whether this is due to functional or structural differences in the brain regions involved in auditory and speech processing, although recent studies have shown that ASD individuals do exhibit different activation patterns in various brain regions in response to speech stimuli. This study investigated responses in regions of the auditory pathway to short recorded and synthesized vowel stimuli. These regions were the Inferior Colliculus, the Left Thalamus, the left Posterior Insula, the Auditory Cortex, Wernicke's area, and Broca's area. The stimuli were parametrized so as to target different signal processing capabilities associated with each region. They were presented to ASD and typically developing (TD) subjects while the salient regions were subject to a functional magnetic resonance imaging (fMRI). The results suggest that there were not gross differences in how ASD individuals responded from TD individuals in the subcortical regions. Results from the Auditory Cortex, however, showed a significant hemisphere dominance in TD subjects with more temporally complex stimuli that did not appear in ASD subjects. Moreover, the results showed that it was temporally-measured periodicities in the signal that were responsible for this difference. The results also show slightly different activation patterns in cortical regions which could have implications for attentiveness, and semantic and emotional processing. These results suggest that deficiencies in the temporal processing capabilities of the left Auditory Cortex play a major role in ASD speech processing. / byBennett Bullock. / S.M.

Thesis (Ph. D.)--Harvard-MIT Division of Health Sciences and Technology, 2003. / Includes bibliographical references (p. 119-122). / In conventional optical microscopy, the numerical aperture (NA) of the objective lens intrinsically links the resolution, depth of field (DOF), working distance (WD) and field of view (FOV). In a diffraction-limited system, changing the NA to improve resolution is not possible without degrading the other optical performances. This linkage, which represents an important limitation in conventional microscopy, has now been broken. In the new method, a target is illuminated by a sequence of finely textured light patterns generated by interference of multiple coherent beams that converge in a cone. The corresponding sequence of brightness values, measured by a single photodetector (e.g., a single pixel of a CCD), encodes the target's sub-pixel contrast pattern. Fourier domain components at spatial frequencies contained in the probing illumination patterns can be recovered from the pixel brightness sequence by solving a set of over-determined linear equations. For a given wavelength, the resolution of the reconstructed image is primarily determined by the NA of the cone of beams, rather than the NA of the microscope objective. A low NA objective, with large DOF, long WD, and large FOV, can be used without compromising resolution. A cone of 31 coherent beams with NA of 0.98 is produced from a single source beam ([lambda] = 488 nm) using an acousto-optic deflector (AOD) and an all-reflective beam delivery system. The target is placed where the beams overlap. Fluoresced light from the target is collected with a 0.2 NA objective for 930 different interference patterns. Brightness sequences are decoded to reconstruct an image with resolution comparable to what would be obtained / using a conventional system with a 0.98 NA objective. Further, changing the NA of the actual objective from 0.2 to 0.1 causes negligible change in resolution, demonstrating that resolution is primarily determined by the illumination produced by the cone of beams. Another restriction of conventional systems is removed by the use of reflective elements. The illumination system, which determines resolution, can be constructed using only reflective elements. Ultraviolet or x-ray illumination can be used with fluorescing targets to obtain resolution beyond what is possible using visible light. The well-known problems of refractive optics at short wavelengths can be avoided. / by Jekwan Ryu. / Ph.D.

Thesis (Ph. D.)--Harvard-MIT Division of Health Sciences and Technology, 2011. / Cataloged from PDF version of thesis. / Includes bibliographical references. / Magnetic resonance imaging (MRI) has demonstrated the potential for non-invasive mapping of structure and function (fMRI) in the human brain. In this thesis, we propose a series of methodological developments towards improved fMRI of auditory processes. First, the inefficiency of standard fMRI that acquires brain volumes one slice at a time is addressed. The proposed single-shot method is capable, for the first time, of imaging the entire brain in a single-acquisition while still maintaining adequate spatial resolution for fMRI. This method dramatically increases the temporal resolution of fMRI (20 fold) and improves sampling efficiency as well as the ability to discriminate against detrimental physiological noise. To accomplish this it exploits highly accelerated parallel imaging techniques and MRI signal detection with a large number of coil elements. We then address a major problem in the application of fMVIRI to auditory studies. In standard fMRI, loud acoustic noise is generated by the rapid switching of the gradient magnetic fields required for image encoding, which interferes with auditory stimuli and enforces inefficient and slow sampling strategies. We demonstrate a fMRI method that uses parallel imaging and redesigned gradient waveforms to both minimize and slow down the gradient switching to substantially reduce acoustic noise while still enabling rapid acquisitions for fMRI. Conventional fMRI is based on a hemodynamic response that is secondary to the underlying neuronal activation. In the final contribution of this thesis, a novel image contrast is introduced that is aimed at the direct observation of neuronal magnetic fields associated with functional activation. Early feasibility studies indicate that the imaging is sensitive to oscillating magnetic fields at amplitudes similar to those observed by magnetoencephalography. / by Thomas Witzel. / Ph.D.

Thesis (Ph.D.)--Harvard--Massachusetts Institute of Technology Division of Health Sciences and Technology, 2002. / Includes bibliographical references (p. 169-184). / This electronic version was submitted by the student author. The certified thesis is available in the Institute Archives and Special Collections. / Cochlear implants are prosthetic devices that seek to restore hearing in profoundly deaf patients by electrically stimulating the auditory-nerve (AN). With current implants, the representation of the sound waveform in temporal discharge patterns of the auditory nerve is severely distorted. The distortion is particularly significant at higher (>600 Hz) frequencies, for which the period of the electric stimulus is near the AN refractory period. For example, in response to a 1000-Hz pulse train, most AN fibers may fire on every other stimulus cycle, so that the AN population would represent half of the stimulus frequency rather than the actual frequency. Rubinstein et al. [Hearing. Res. 127, 108] proposed that the coding of electric waveforms in cochlear implants can be improved if a sustained, electric high-rate (5 kpps) desynchronizing pulse train (DPT) is presented in addition to the information-carrying electric stimulus. The DPT may amplify the inherent noise in ANfibers so as to produce ongoing, stochastic discharges similar to the spontaneous activity in a healthy hear. We tested this hypothesis by recording responses of ANfibers of deafened cats to sustained electric pulse trains. For most fibers, responses to the DPT showed adaptation during the first 2 minutes, followed by a sustained response for the remainder of the 10-minute stimulus. These sustained responses partially resembled spontaneous activity in terms of discharge rate and interspike interval distributions. AN fibers were extremely sensitive to modulations of the DPT, responding tomodulations as small as 0.5%. / (cont.) Responses to sinusoidal modulations resembled AN responses to pure tones over a 15-25 dB range of modulation depths. Responses to complex modulations simultaneously represented several spectral components of the modulator in their temporal discharge patterns. However, for modulation depths above 10%, the representation of both sinusoidal and complex modulators was more distorted. These results demonstrate that strategies that incorporate a DPT, and that use low modulation depths to encode sounds, may evoke AN responses that more accurately represent the modulator in their temporal discharge patterns. If the central nervous system can utilize this information, then these strategies may substantially improve performance enjoyed by cochlear implant users. / by Leonid Litvak. / Ph.D.