Existing microstructure parameters of computed tomography (CT) are able to compute architectural properties of the bone from ex-situ and ex-vivo scans while they are highly affected by the issues of noise and low resolution when applied to clinical in-vivo imaging. A set of improvements of the standard workflow for the quantitative computation of micro-structure from clinical in-vivo scans is proposed in this thesis. Robust methods are proposed (1) for the calibration of density values, (2) the binarization into bone and marrow phase, (3) fuzzy skeletonization and (4) the calibration of the CT volumes in particular for the computation of micro-structural parameters. Furthermore, novel algorithms for the computation of rod-volume fraction with 3D rose diagrams and fractal approaches are proposed and the application of local texture operators to diffusion tensor imaging is proposed. Finally an existing computer program for the application in radiology departments, Structural Insight, was improved and largely extended. In particular the methods of the microstructural calibration, the fractal and the texture operators showed significant improvements of accuracy and precision for the prediction of fracture risk and the quantitative assessment of the progress of Alzheimer's disease, in comparison to existing state-of-the art methods. The methods were tested on artificial and in-vitro data and as well on real-world computed tomography and magnetic resonance imaging (MRI) studies. The proposed novel methods improve the computation of bone characteristics from in-vivo CT and MRI in particular if the methods are combined with each other. In consequence, this allows to assess more information from existing data or to conduct studies with less ray exposure and regarding the MRI method in shorter time than nowadays required.

The vestibular system consists of several organs that contribute to ones sense of balance. One set of organs, otoconial organs, have been shown to respond to linear acceleration (1949). Hair bundles (and hair cells), which are the mechano-electric transducers found within otoconial organs, respond to displacement of the overlying otoconial membrane (OM). Structure, position and orientation of the OM within the head may influence the stimulus of hair bundles by changing the deformation characteristics of the OM. Therefore, studying the deformation characteristics of the OM with finite element models presents a unique advantage: the ability to study how different variables may influence the deformation of the OM. Previous OM models have ignored complicated OM geometry in favor of single degree of freedom (De Vries 1951)or distributed parameter models (Grant et al. 1984; Grant and Cotton 1990; Grant et al. 1994). Additionally, OMs have been modeled considering three dimensional geometry (Benser et al. 1993; Kondrachuk 2000; 2001a), however OM layer thicknesses were assumed to be constant. Further, little research has investigated the effect of position and orientation of otoconial organs on the deformation of the OM (Curthoys et al. 1999), due to natural movement of the head. The effect of structure, position and orientation of the utricle of a red ear slider turtle on the stimulation of hair bundles in the OM is investigated here. Using confocal images, a finite element model of the utricle OM is constructed considering its full 3D geometry and varying OM layer thickness. How specific geometric variables, which are missing from other OM models, effect the deformation of the utricle OM is studied. Next, since hair bundles are part of the structure of the OM, their contribution to the deformation of the utricular OM is quantified. Then, using computed tomography of a turtle head and high speed video of turtle feeding strikes, acceleration at the utricle during natural motion is estimated. Finally, the effects of orientation of the utricle in the head on the stimulus of hair bundles within the organ is investigated. In summary, a model and methods are developed through which deformation of the turtle utricle OM through natural movements of the head may be studied. Variables that may contribute to utricle OM deformation are investigated. Utricle OM geometry, hair bundles, position and orientation all play a role in utricle OM deflection and therefore hair bundle stimulus. Their effects are quantified and their roles are discussed in this dissertation. / Ph. D.

The accuracy of plenoptic and tomographic particle image velocimetry (PIV) experimental methods is measured by simulating three-dimensional flows and measuring the errors in the estimated versus true velocity fields. Parametric studies investigate the accuracy of these methods by simulating a range of camera numbers, camera angles, calibration errors, and particle densities. The plenoptic simulations combine lightfield imaging techniques with standard tomographic techniques and are shown to produce higher fidelity measurements than either technique alone. The tomographic PIV simulations are centered around testing software developed for processing large quantities of data that were produced during an experimental investigation of the flow field about a 3D printed model of the flying snake Chrysopelea paradisi. A description of this tomographic PIV experiment is given along with basic results and recommendations for future investigation. / Doctor of Philosophy / Two different experimental measurement techniques that can be used to measure three-dimensional fluid flow fields are discussed. The first measurement technique that is investigated in simulations uses cameras with arrays of lenses to simultaneously capture images of a flow field from multiple different angles. A method of combining the data from multiple cameras is discussed and shown to yield more accurate estimates of the three-dimensional flow fields than from a single camera alone. An additional measurement technique that uses a group of standard cameras to measure three-dimensional flow fields is also discussed with respect to software that was developed for processing a large volume dataset. This software was developed for processing data collected during an experimental investigation of the flow field about a 3D printed model of the flying snake Chrysopelea paradisi. A description of this experiment is given along with basic results and recommendations for future investigation.

Computed tomography (CT) is becoming more available as a diagnostic tool in the evaluation of the equine skull and brain. Objectives of this study were: 1) refine a CT protocol for evaluating the equine pituitary gland, 2.) define the CT anatomy of the pituitary region, 3.) determine a set of normal values for the pituitary dimensions (length, width, height, volume and weight), 4.) refine CT techniques for measuring pituitary size. Horses were scanned using 10x10mm, 10x5mm, 4x4mm and 4x2mm slice thickness and interval combinations. The pituitary glands were removed immediately after CT and gross measurements were performed. CT measurements were compared with gross pituitary measurements using analysis of variance (ANOVA) in a randomized block design. Accuracy percentages were also calculated using gross measurements as the known value. Mean dimensions of the histologically normal pituitary glands were: length 21.07mm, width 21.62mm, height 9.78mm and volume 2.66cm³. The weights ranged from 1.7g to 3.4g with a mean of 2.6g. Computed tomographic measurement analysis demonstrated that the 10mm slices were the most accurate way to estimate the length of the gland. The 4mm slices yielded the highest accuracy values for width, height and volume of the pituitary gland. The volume was underestimated by all interval and slice thickness combinations performed by CT. No evidence of an overlap effect was identified for any of the dimensions. Our findings indicated that contrast-enhanced CT is an accurate technique for estimating pituitary linear dimensions. Three-dimensional CT volumetry may not be an accurate method for estimating pituitary volume. / Master of Science

Velocity tomography is a noninvasive technology that can be used to determine rock mass response to ore removal. Velocity tomography is accomplished by propagating seismic waves through a rock mass to measure velocity distribution of the rock mass. Tomograms are created by mapping this velocity distribution. From the velocity distribution relative stress in the rock mass can be inferred, and this velocity distribution can be mapped at specific time intervals. Velocity tomography is an appropriate technology for the study of rockbursts. Rockbursts are events that occur in underground mines as a result of excessive strain energy being stored in a rock mass and sometimes culminating in violent failure of the rock. Rockbursts often involve inundation of broken rock into open areas of the mine. They pose a considerable risk to miners and can hinder production substantially. The rock mass under investigation in this research is the strata surrounding an underground coal mine in the western United States, utilizing longwall mining. The mine has experienced rockbursts. Seismic data were collected over a nineteen day period, from July 20th, 1997 to August 7th, 1997, although only eighteen days were recorded. Instrumentation consistsed of sixteen receivers, mounted on the surface, approximately 1,200 feet above the longwall panel of interest. The system recorded and located microseismic events, and utilized them as seismic sources. The data were analyzed and input into a commercial program that uses an algorithm known as simultaneous iterative reconstruction technique to generate tomograms. Eighteen tomograms were generated, one for each day of the study. The tomograms consistently display a high velocity area along the longwall tailgate that redistributes with face advance. Numerical modeling and mine experience confirm that the longwall tailgate is subject to high stress. Additionally, microseismic events are correlated with the velocity tomograms. Velocity tomography proves to be an effective method for the study of stress redistribution and rockburst phenomena at underground longwall coal mines, because it generates images that are consistent with prior information about the stress state at the mine and with numerical models of the stress in the mine. / Master of Science

Velocity tomography is a noninvasive technique that can image the interior of a rock structure. To apply tomography to rock specimens, a propagation wave, which acts as a probe, is used. The propagation wave propagates from a source until it reaches a sensor on the surface of the rock specimen. Tomograms can then be generated from the velocity distribution within the rock structure. Areas of higher velocity are typically representative of higher stress concentrations, whereas areas of low velocity can be areas of fracturing. The variation of velocity tomography described in this thesis uses acoustic emissions as sources for the propagation wave. Acoustic emission sources provide advantages over mechanical sources, since the acoustic emission source is generated by the rock as a result of deformation and fracturing. Velocity tomography of rock structures in the field has numerous applications and advantages. Velocity tomography can be used to monitor rock structures surrounding tunnels and underground openings such as mines. To monitor the rock structure, velocity tomography is used to determine areas of higher stress concentration that may be precursors to rock failure. However, velocity tomography must first be used in a laboratory environment to determine failure in rock samples before being applied to the field. The research presented includes the unconfined compression strength testing of 19 Berea sandstone samples. These samples were loaded to failure and during the experiment the acoustic emission events within the samples were monitored using a commercial acquisition system manufactured by Engineering Seismology Group (ESG) Canada. Source location software, also produced by ESG, was used for the location of the acoustic emission events. Ray inversions were performed on the data from the experiments to generate tomograms. The tomograms generated display the p-wave velocity distribution imaged within the Berea sandstone samples with the ultimate goal of being able to predict rock failure. Based on the experiments discussed in this thesis it can be inferred that velocity tomography is a useful tool for imaging the inside of the Berea sandstone samples. Precursors of rock failure could not be determined in this early stage of research. However, the tomograms do image the p-wave velocity distribution and do show a gradual progression of the p-wave velocity from the initial velocity model to higher velocities. Results of these 19 experiments do provide reasonable confidence in the method and warrant pursuit of further research to refine and improve this method of monitoring velocity tomography. / Master of Science

Mining induced seismicity can lead to bumps which cause problems at many mines within the United States and around the world. This seismicity, often referred to as bumps or bursts, can result in injuries, fatalities, and expensive capital damage and production interruptions. There are many factors that contribute to mining induced seismicity but there is still no concrete method to forecast future seismic activity around a mine. One of the main precursors to large seismic events is an increase in situ stress. One way to find areas within geological strata that are highly stressed is to measure p-wave propagation velocities. High p-wave propagation velocities are associated with high in-situ stress levels. By using tomography programs a three-dimensional velocity model can be constructed. When seismic activity is present the event arrival times at each geophone, the locations of each geophone, and the three dimensional velocity model are used in conjunction with one another to locate the seismic events. This research compares the locations of seismic events from a deep coal mine longwall panel in the western United States with the associated p-wave propagation velocities from the previous 24 hours. The aim of this comparison is to provide a link between high velocities and seismic activity that could potentially be used to forecast future seismic activity. The comparison is completed both qualitatively through the use of a visual analysis, and quantitatively using various numerical and correlation comparisons on the seismic and velocity data. The qualitative comparison is completed using the event locations from the tomography program SIMULPS. The quantitative comparison is completed twice using two different tomography programs, SIMULPS and TomoDD, which use different methods for locating the seismic events. Before these comparisons were completed the stresses around the longwall panel were first modeled using the boundary element modeling program LAMODEL to study the effects of three backfilled cross panel entries which were located ahead of the mining face. The modeling showed similar vertical stress distributions as a panel without cross panel entries but higher stress magnitudes. The qualitative analysis involved comparing tomograms created with SIMULPS with seismicity plots from the following day. One noticeable feature of these tomograms is the presence of a stressed area directly ahead of the face. This stressed area represents the forward abutment. The results of this qualitative analysis illustrate a correlation between high p-wave velocities and seismic activity 24 hours later for several of the days studied. The other days showed little to no correlation. Additionally, not all high p-wave velocity regions resulted in seismic activity. Due to these inconsistencies visually analyzing velocity plots obtained from the program SIMULPS is not a reliable way to forecast the locations of seismic activity 24 hours later. The result of the quantitative comparisons completed with the programs SIMULPS and TomoDD further highlighted inconsistencies in the correlation between high p-wave velocities and associated seismic activity 24 hours later. TomoDD provided better correlation values than SIMULPS and generally showed that as the level of seismicity increased the p-wave propagation velocities 24 hours prior also increased. Although TomoDD provided good correlations for some of the data pairs studied, the overall inconsistencies prompt the need for further study in this area using TomoDD to find the optimal forecasting time period. / Master of Science

A recent concern is the cause and effect of global climate change. Many institutions give credit for these changes to the increased levels of greenhouse gases in the atmosphere, in particular the increase in the amount of carbon dioxide present. There is a growing interest in carbon capture and storage (CCS) as a means to reduce the global impact of CO₂ on the climate as a greenhouse gas. Carbon capture is the process of removing CO₂ from the atmosphere as well as preventing it from entering the atmosphere by means of exhaust. The captured carbon is stored underground in reservoirs. These reservoirs have the storage space to handle the volume of CO₂ injected as well as a caprock layer preventing the injection fluid from returning to the surface. Additionally, CO₂ can be used for enhanced oil recovery (EOR). To monitor the injection sites used for the CO₂ storage or EOR process, the integrity of the caprock as well as the surrounding rock formations are the locations of interest. Knowing when a joint or a fracture is going to slip is necessary to prevent major failures within geologic strata. It is necessary to prevent these slips from occurring to retain the integrity of the caprock, which is keeping the fluid within the reservoirs. Passive acoustic emissions monitoring was used to determine how effectively failure locations could be located in three unique tests. Coupled with double difference tomography, the failure of a Berea Sandstone sample and Marcellus Shale sample were calculated to determine how well the stress redistribution within the sample could be mapped using the recorded data. For the main indenter tests two samples were tested, a piece of Berea Sandstone and a piece of Marcellus Shale. The secondary test was a transform shear test using sandstone, and the third test for caprock upheaval test attempted to recreate the failure of caprock due to injection pressure. For all tests, the samples were monitored using acoustic emissions software until failure or it was deduced that the test would not produce failure. The secondary tests did not progress through the data analysis as far as the indentation tests, however valuable information was gathered from these tests. The shear test demonstrated the effectiveness of the passive acoustic emissions monitoring system to record shear failure. This test provides confidence in this technology to record and located events that are not occurring in compression. The caprock upheaval tests were not successful in causing failure in the caprock, however during the testing the passive acoustic emissions monitoring system was able record and locate events that occurred within the sample around the boundary on the reservoir. At the reservoir boundaries there was evidence of fluid flowing through the reservoir, and the events align with these locations. This positive result shows that the monitoring system is able to locate events induced by fluid injection. The results of these tests provide confidence in the passive acoustic emissions monitoring system to record accurate data for the caprock integrity monitoring. The tomograms created from the recorded data accurately imaged the areas of interest within the rock samples. From these results, passive acoustic emissions monitoring systems coupled with double difference tomography has proven capable of monitoring homogeneous samples within a laboratory environment. With further testing, this technology could possibly be a viable option for monitoring carbon sequestration sites. / Master of Science

Due to their scientific significance and practical applications, turbulent flows and flames have been under extensive and intensive research for a long time. Turbulent flows and flames of interests to practice inherently have three-dimensional (3D) spatial structures, and therefore diagnostic techniques that can instantaneously resolve their 3D spatial features have long been desired and probably are needed to ultimately answer some of the open research questions. The goal of this dissertation thus is to investigate such diagnostics and demonstrate their capability and limitations in a range of turbulent flows/flames. To accomplish this goal, this dissertation developed and evaluated the following three diagnostic methods: tomographic chemiluminescence (TC), volumetric laser induced fluorescence (VLIF), and super-resolution planar laser induced fluorescence (SR-PLIF). First, 3D flame topography of well-controlled laboratory flames was measured with TC method and validated by a simultaneous 2D Mie scattering measurement. The results showed that the flame topography obtained from TC and the Mie scattering agreed qualitatively, but quantitative difference on the order of millimeter was observed between these two methods. Such difference was caused by the limitations of the TC method. The first limitation involves TC's reliance on chemiluminescence of nascent radicals (mainly CH*) in reacting flows, causing ambiguity in the definition of flame front and limiting its applications to certain types of reactive flow only. The second limitation involves TC's inability to study an isolated region of interest because the chemiluminescence is emitted everywhere in the flame. Based on the above understanding of the TC technique, the second part of this dissertation studied a VLIF method to overcome the above limitations of the TC technique. Compared with the TC technique, the VLIF method can be used in either reacting or non-reacting flow and on any particular region of interest. In the VLIF technique, the fluorescence signal was generated by exciting a target species with a laser slab of certain thickness. The signal was recorded by cameras from different perspectives, and then a VLIF tomographic algorithm was applied to resolve the spatial distribution of the concentration of the target species. An innovative 3D VLIF algorithm was proposed and validated by well-designed experiment. This model enables analysis of VLIF performance in terms of signal level, size of the field of view in 3D, and accuracy. However, due to the limited number of views and the tomographic reconstruction itself, the spatial resolution of VLIF methods is limited. Hence, the third part of this dissertation investigated a SR-PLIF method to provide a strategy to improve the spatial resolution in two spatial directions, and also to extend the measurement range of scanning 3D imaging strategies. The SR-PLIF method used planar images captured simultaneously from two (or more) orientations to reconstruct a final image with resolution enhanced or blurring removed. Both the development of SR algorithm, and the experimental demonstration of the SR-PLIF method were reported. / Ph. D. / Optical diagnostics have become indispensable tools for the study of the turbulent flows and flames. Due to the inherently 3D structure of turbulent flows and flames, diagnostic techniques which can provide 3D measurements have been long desired. Therefore, this dissertation reports the development of three optics diagnostic methods that can provide such measurement capability, with a detailed discussion of their capabilities and limitations. The methods studied are tomographic chemiluminescence (TC), volumetric laser-induced fluorescence (VLIF), and super-resolution planar laser induced fluorescence (SR-PLIF). For the TC technique, the emission of light from combustion radicals (CH* and OH*) was recorded by multiple cameras placed at different orientations. A numerical algorithm was then applied to reconstruct the 3D flame structure. For the VLIF technique, a laser slab was used to excite a specific chemical species in the flame, which were captured from different perspectives to reconstruct the flow or flame structure in 3D. For the SR-PLIF technique, a series of planar images were recorded from multiple orientations to reconstruct a target image with higher resolution or to extend the measurement volume of scanning 3D diagnostics. It is expected that the results obtained in this dissertation lay the groundwork for further development and expanded application of 3D diagnostics for the study of turbulent flows and combustion processes.

Electron tomography is an invaluable technique with the capability of carrying out thorough 3D structural, chemical and morphological characterization of materials at nanometer scale. Tilting range, increment and reconstruction algorithms are three of the main factors affecting the quality of tomograms. An anisotropic degradation can be observed with restricted tilting range and increment. Therefore, this study was carried out to investigate the accuracy of the reconstruction results of MgO (cube-shape) generated by FBP, SART and SIRT tomographic algorithms under various reconstruction conditions, i.e. tilting range and increment. Examining the experimental data with known morphology permits quantitative determination of the accuracy of the reconstruction results by measuring the distortion of the cube in all directions. Moreover, distortion measurements in all directions reveal the relationship between level of distortion and the alpha tilt angle. / Master of Science