Community health assessment (CHA) professionals who use information technology need a complete system that is capable of supporting numerical-spatial problem solving. On-Line Analytical Processing (OLAP) is a multidimensional data warehouse technique that is commonly used as a decision support system in standard industry. Coupling OLAP with Geospatial Information System (GIS) offers the potential for a very powerful system. For this work, OLAP and GIS were combined to develop the Spatial OLAP Visualization and Analysis Tool (SOVAT) for numerical-spatial problem solving. In addition to the development of this system, this dissertation describes three studies in relation to this work: a usability study, a CHA survey, and a summative evaluation. The purpose of the usability study was to identify human-computer interaction issues. Fifteen participants took part in the study. Three participants per round used the system to complete typical numerical-spatial tasks. Objective and subjective results were analyzed after each round and system modifications were implemented. The result of this study was a novel OLAP-GIS system streamlined for the purposes of numerical-spatial problem solving. The online CHA survey aimed to identify the information technology currently used for numerical-spatial problem solving. The survey was sent to CHA professionals and allowed for them to record the individual technologies they used during specific steps of a numerical-spatial routine. In total, 27 participants completed the survey. Results favored SPSS for numerical-related steps and GIS for spatial-related steps. Next, a summative within-subjects crossover design compared SOVAT to the combined use of SPSS and GIS (termed SPSS-GIS) for numerical-spatial problem solving. Twelve individuals from the health sciences at the University of Pittsburgh participated. Half were randomly selected to use SOVAT first, while the other half used SPSS-GIS first. In the second session, they used the alternate application. Objective and subjective results favored SOVAT over SPSS-GIS. Inferential statistics were analyzed using linear mixed model analysis. At the .01 level, SOVAT was statistically significant from SPSS-GIS for satisfaction and time (p < .002). The results demonstrate the potential for OLAP-GIS in CHA analysis. Future work will explore the impact of an OLAP-GIS system in other areas of public health.

<p>Breast cancer is one of the leading causes of death every year in the United States for women. Breast conserving surgery (BCS) is one treatment option for these patients where achieving tumor-free surgical margins is desired to avoid local recurrence [1, 2]. Unfortunately, as many as 17.7-72% of patients undergoing BCS require repeat surgeries due to a close or positive surgical margin diagnosed post-operatively [3-11]. Histopathology is the current gold standard for determining surgical margin status; however, given the large volumes of resected breast tissue it is not feasible to section the entire specimen. High re-excision rates and limited histopathological sampling of the tissue represent a significant unmet clinical need for margin assessment for both the surgeon and pathologist. Quantitative diffuse reflectance (DR) spectral imaging has been shown to be a promising tool for interrogating tumor margins but patient and surgical factors have to be accounted for in order to fully exploit the discriminatory capability of the technology. The objective of this work was to characterize an instrument for margin assessment and to evaluate the effects of inter-patient variability and surgical and excisional factors on quantitative tissue optical properties, to devise strategies to exploit optical contrast for the detection of positive (<2mm) tumor margins. In addition, the performance of the spectral imaging platform was evaluated.</p><p>The DR spectral imaging device utilized in these studies consisted of a Xenon lamp, a multi-channel imaging fiber-optic probe, an imaging spectrograph, and a 2D charge-coupled device (CCD) [12]. The instrument was found to extract quantitative optical parameters related to tissue micro-morphology with <15% error. Cross-talk at the tissue surface was <1% when the spacing between adjacent channels was 10mm and the sensing depth of each channel was found to be 0.5-2.2mm, an appropriate depth for identifying close and positive tumor margins. Reproducibility of the imaging protocol was best when the probe was interfaced with lumpectomy specimens from the side; this methodology was maintained for all measurements from lumpectomies in this dissertation.</p><p>DR spectral images were acquired from lumpectomy margins and converted into composition maps of quantitative optical parameters. Mammographic breast density was found to have the greatest impact on the optical data with &#946;-carotene concentration ([&#946;-carotene]) and the ratio of [&#946;-carotene] to the wavelength-averaged reduced scattering coefficient from 450-600nm (<&#956;s'>) being significantly higher in the negative margins of high-density patients (p=0.017 and p=0.038, respectively). We originally hypothesized that increased [&#946;-carotene] would be associated with an increase in fat; however the significant increase in [&#946;-carotene] cannot be attributed to differences in the percentage of adipose tissue since low-density patients should theoretically have higher percentages of this tissue type. Hematoxylin and eosin analysis of the adipose sites (n=25) showed increased [&#946;-carotene] (p=0.066), increased adipocyte density (p=0.034), and smaller adipocyte sizes (p=0.051) in the adipose tissues (where &#946;-carotene is stored) of high-density patients. This analysis suggests that increased [&#946;-carotene] is associated with smaller adipocytes and that high-density breasts overall have smaller adipocytes, thus affecting optical contrast. This increase in [&#946;-carotene] actually served to increase contrast between negative and positive margins which resulted in better classification accuracy in the high-density patients with a conditional inference tree model (77% in low-density and 80% in high-density). </p><p>If the purpose of the spectroscopy tool is to provide a differential diagnosis of benign versus malignant tissue, there must be an understanding of how excision of the tissue affects the optical properties over time, and how differences in surgical techniques affect optical properties. DR spectra were acquired 17±4 minutes post-excision from 12 incised mastectomies and from the surface of 10 lumpectomies 7±3 minutes post-excision. A linear longitudinal model was used to fit the data and obtain a rate of change for the tissue parameters. In lumpectomies, [&#946;-carotene], <&#956;s'>, and [&#946;-carotene]/<&#956;s'> had the lowest percent change (<14%) over 30 minutes; total hemoglobin concentration ([THb]) and [THb]/<&#956;s'> had higher percent changes (>40%) over 30 minutes; hemoglobin saturation (HbSat) showed non-linear changes making it a poor variable for ex vivo margin assessment; and Lymphazurin^TM concentration (denoted as [Lymphazurin^TM]) changed more than 200% in 30 minutes. Although the percent error in [Lymphazurin^TM] was high, all other tissue parameters could be quantified with <3.3% error even when Lymphazurin^TM was 80&#956;M. No significant difference was found between benign and malignant rates of change, and baseline values were not significantly correlated with elapsed time post-excision. Initial values from benign non-cauterized mastectomy (n=13) and cauterized lumpectomy (n=59) sites were compared to assess the effect of cautery. [THb] was the only parameter that was significantly higher in the cauterized lumpectomies (p=0.013) compared to non-cauterized mastectomies. </p><p>The work in this dissertation shows the feasibility of using an optical device for margin assessment and that [&#946;-carotene] and [&#946;-carotene]/<&#956;s'> emerge as important variables for differentiating negative and close/positive margins. These two parameters were likely most important since they were least affected by kinetics, cautery, and the presence of Lymphazurin^TM.</p> / Dissertation

Infrared neural stimulation (INS) uses pulsed infrared light to directly stimulate neural tissue with high spatiotemporal precision and is well documented for peripheral nerve applications; however, prior to this dissertation, INS had not been demonstrated for the central nervous system. This dissertation presents the first successful application of INS in the central nervous system and increases our understanding the effects of pulsed infrared light irradiation on cellular dynamics in the brain. Pulsed infrared light is shown to evoke both excitatory and inhibitory neural activity, and evokes robust optical intrinsic signals indicating multiple cellular mechanisms are activated by INS. Optical imaging of calcium signals evoked by INS identified astrocyte sensitivity to pulsed infrared light confirming that both neurons and astrocytes are stimulated. Application of INS in non-human primate visual cortex demonstrated that pulsed infrared light evokes excitatory neural activity and modulates visually evoked signals, identifying the potential of INS to encode functionally relevant signals into cortex. Overall, these results establish INS as neurostimulation modality for use in the brain, and this dissertation provides the necessary foundation to further develop INS for use in the central nervous system in both research and clinical applications.

Reading is a complex cognitive behavior, which relies on the incorporation of a network of brain regions. White matter is the information transfer pathway between distant brain regions, and thus plays an important role in mediating reading ability. Diffusion Tensor Imaging (DTI) is an MR technique to characterize white matter microstructure by probing the propensity of water molecules¡¯ diffusion in in vivo tissues. This dissertation seeks to investigate the reading network in the brain using diffusion MRI. The first part of the dissertation studies the cortical network with a focus on the putative visual word form area (VWFA), which is reproducibly found to be selectively recruited by visual orthographic conversion. We studied the structural connectivity patterns of the VWF-system in children with typically developing (TD) reading ability and with reading difficulty (RD). We found that the architecture of the VWFA connectivity is fundamentally different between TD and RD groups, with TD showing greater connectivity to linguistic regions than RD, and RD showing greater connectivity to visual regions than TD. The second part of the dissertation studies subcortical-cortical network, with a focus on the thalamus, the way-station of information transfer in white matter. Abnormal thalamo-cortical connectivity was found in the RD group in sensorimotor, orbital frontal and insula cortices. These results suggest that the thalamus plays a key role in reading behavior by mediating the functions of task specific cortical regions. Despite the valuable information DTI can provide, it suffers from fundamental limitations, especially when multiple fiber bundles are present. To address this problem, the third part of the dissertation proposes a new method to study complex white matter structures. It improves the current spherical deconvolution method by relaxing the assumption that all fiber bundles share the same response kernel. The in vivo experiments show that this Multiple Kernel Spherical Deconvolution (MKSD) approach can identify crossing fiber bundles and simultaneously provide an estimate of the diffusion properties intrinsic to each fiber bundle.

BIOMEDICAL INFORMATICS DESIGN AND IMPLEMENTATION OF A COMPUTERIZED INFORMATICS TOOL TO FACILITATE CLINICIAN ACCESS TO A STATES PRESCRIPTION DRUG MONITORING PROGRAM DATABASE STEVEN JOHN WHITE Thesis under the direction of Professor Dario Giuse Within the past decade, prescription drug abuse has emerged as a nationwide epidemic, with opioid-related poisoning deaths more than tripling since 1999. In an effort to bring this public health crisis under control, 43 states, including Tennessee, have enacted prescription drug monitoring programs (PDMPs), computerized databases of DEA-controlled substance prescriptions filled at pharmacies within the given state. Such programs have been found to be effective in curbing prescription opioid abuse by alerting prescribers to aberrant prescription-filling activity. However, they are commonly underutilized and have workflow barriers that impede clinical use. Ideally, PDMP queries could be generated seamlessly from within a medical enterprises electronic health record (EHR) system, using an application-programming interface (API) supplied by the states PDMP vendor. However, the enabling legislative language currently prohibits such access. Therefore, we developed and evaluated a Perl software program activated from within Vanderbilt University Medical Centers EHR patient chart to send the properly coded/formatted user and patient-demographic information packets to the Tennessee PDMP website, without the use of an API. The program parses the returned data file for important prescription information and displays the filtered information to the user. By allowing the query to occur in the background, the users tether time to the computer is decreased from 3 minutes to 10 seconds per query. During the evaluation phase, we used a quasi-experimental intervention design with two alternating 2-week control and intervention periods. Twenty-eight ED attending physicians participated in the study and queried the PDMP at their clinical discretion. During integrated PDMP query tool availability, 5.9 % (169/2844) of emergency department patients were screened compared with 2.2 % (62/2786) during periods when the tool was not available (p<0.001, Pearsons Chi square). Data was not viewed in 20% of integrated tool assisted queries. The EHR-integrated PDMP query tool was well regarded by study physicians as an enhancement to workflow.

The Biochemical Library (BCL) is a protein structure prediction algorithm developed in the Meiler Lab at Vanderbilt University based on the placement of secondary structure elements. This algorithm can use experimental data such as nuclear magnetic resonance (NMR), Cryo-electron microscopy (CryoEM), and electron paramagnetic resonance (EPR), to assist in protein structure prediction but does not have the ability to use SAXS data. The first phase of my project was to add this capability to the BCL and create a SAXS compatibility score. GPU acceleration was used to parallelize the computations. The second phase of the project was to compute SAXS scores for protein models without side chain and loop region coordinate information, but preserve atom type information. Finally, the BCL::SAXS score was added to the minimization process in BCL::FOLD. The SAXS score can be used to filter erroneous initial protein models from further refinement, thus saving time and computation resources.

A fast approach to reconstruct transmit RF field (|B1+|) maps in high field parallel Magnetic Resonance Imaging is presented in this thesis. First the concepts of parallel imaging reconstruction and the Bloch-Siegert (BS) |B1+| mapping method are introduced. Then we present the idea to accelerate BS field mapping based on the fact that the |B1+|-to-phase encoding pulse for each transmit coil and off-resonance frequency applies a unique phase shift to the same underlying image. This enables joint reconstruction of all images in a BS acquisition from an augmented set of virtual receive coils. Simulations are shown to produce accurate |B1+| maps at high acceleration factors for both Cartesian and Non-Cartesian sampling. Phantom and in vivo experiments confirm the simulation results and endorse the proposed approach. It is also shown that reconstruction accuracy can be improved using disjoint sampling patterns between acquisitions. Noise performance analysis in simulation demonstrates that disjoint sampling produces less noise amplification when compared to same sampling across all the coils. The approach is general in that it can be applied using any auto-calibrated multi coil image reconstruction method.

Clinically compatible ultra-short echo time (UTE) imaging sequences for quantitative T2-based bound and pore water imaging in bone were implemented and validated on a 3T human scanner and a 4.7T small bore system. Pore water images were generated by selectively saturating bound water signals with a T2-selective double adiabatic inversion pulse. Bound water images were generated by nulling the pore water signals with a T2-selective adiabatic inversion recovery preparation. In both cases, the magnetization preparation was integrated into a 3D UTE acquisition, with 16 radial spokes acquired per preparation. Images were acquired from human cadaveric femoral mid-shafts, from which isolated bone samples were subsequently extracted for non-imaging analysis using T2 spectroscopic measurements. The results showed a strong correlation between imaging-derived concentrations of bound and pore water and those determined from the isolated bone samples.

Diffusion tensor imaging (DTI) provides a unique approach to probing the microstructure of biological tissues noninvasively and DTI-based tractography is an irreplaceable tool to measure anatomical connectivity in human brain in vivo. However, due to the limitations of DTI techniques and tractography algorithms, tracked pathways might not be completely accurate. Thus, quantifying the agreement between DTI tractography and histological measurements of true fiber pathways is critical for progress in the field. A series of validation studies of DTI tractography is presented in this thesis, including (1) assessment of the relationship between DTI tractography-derived corticocortical connectivity and histological 'ground truth' on a regional and voxelwise basis; (2) localizing the divergence between DTI tractography and histology, followed by qualitative analysis of the reasons for those discrepancies. The work presented here is based on a non-human primate animal model, which has comparable parameters to magnetic resonance imaging (MRI) human data, and thus provides an important guide to interpreting the results of DTI-based tractography measures in the human brain.

Objective: Supervised machine learning methods have shown good performance in text classification tasks in the biomedical domain, but they often require large annotated corpora, which are costly to develop. Our goal is to assess whether active learning strategies can be integrated with supervised machine learning methods, thus reducing the annotation cost while keeping or improving the quality of classification models for biomedical text. Methods: We have applied active learning to two biomedical natural language processing (NLP) tasks: 1) the assertion classification task in the 2010 i2b2/VA Clinical NLP Challenge, which was to determine the assertion status of clinical concepts; and 2) a supervised word sense disambiguation (WSD) task that was to disambiguate 197 ambiguous words and abbreviations in MEDLINE abstracts. We developed Support Vector Machines (SVMs) based classifiers for both tasks. We then implemented several existing and newly developed active learning algorithms to integrate with SVM classifiers and evaluated their performance on both tasks. Results: In assertion classification task, our results showed that to achieve the same classification performance, active learning strategies required much fewer samples than the random sampling method. For example, to achieve an AUC of 0.79, the random sampling method used 32 samples, while our best active learning algorithm required only 12 samples, a reduction of 62.5% in manual annotation effort. In the WSD task, our results also demonstrated that active learners significantly outperformed the passive learner, showing better performance for 177 out of 197 (89.8%) ambiguous terms. Further analysis showed that to achieve an average accuracy of 90%, the passive learner needed 38 samples, while the active learners needed only 24 annotated samples, a 37% reduction of annotation effort. Moreover, we also analyzed cases where active learning algorithms did not achieve superior performance and summarized three causes: (1) poor model in early learning stage; (2) easy WSD cases; and (3) difficult WSD cases, which provide useful insight for future improvements. Conclusion: Both studies demonstrated that integrating active learning strategies with supervised learning methods could effectively reduce annotation cost and improve the classification models in biomedical text processing.