Browsing by Subject "Image analysis"
Now showing 1 - 12 of 12
- Results Per Page
- Sort Options
Item Computational image analysis of mass lesions on dynamic contrast-enhanced breast MRI(2009-12) Wu, Qiu, active 2009; Fussell, Donald S., 1951-; Whitman, Gary J.This dissertation presents results of a medical image analysis project leading towards development of a comprehensive set of methods and tools for computational image analysis of dynamic contrast-enhanced (DCE) breast magnetic resonance image (MRI), with the aim to aid the physician in interpreting DCE breast MRI examinations. Toward this goal, we developed image analysis methods that would be needed in a breast MRI computer aided diagnosis (CADx) system. A novel contribution of this dissertation is the performance evaluation for each of the major algorithm components developed in this dissertation project. This dissertation begins with reviewing breast imaging techniques, including routinely used modalities in current clinical practice and emerging techniques still in development. We discuss at length the principles of DCE breast MRI, a very sensitive breast imaging modality that has been increasingly used in clinical practice. Then we review the diagnostic guidelines for interpreting DCE breast MRI, and explain the needs and challenges that arise in developing computational image analysis system for breast MRI applications. In this dissertation project, both the morphological and kinetic features of the lesion are automatically extracted for diagnostic purpose. In order to extract morphological features from the segmented lesions, the lesion needs to be accurately segmented out from its surrounding tissues. We utilized a probabilistic method to obtain an optimal segmentation map based on several algorithmic segmentation outputs. In evaluating the performance of segmentation algorithms, we compared the algorithmic segmentation results against manually segmented lesions, and further assessed the segmentation impact on subsequent classification stage. In order to extract accurate kinetic information, the motion needs to be compensated across image volumes acquired sequentially. In this dissertation, we comparatively assessed the similarity metric in registering DCE breast MR images. The performance of cross correlation(CC) coefficient, and mutual information (MI) were studied in both rigid and non-rigid registration schemes. Numerical results and statistical properties were reported. The resultant image quality after registration is discussed both qualitatively and quantitatively. In this dissertation we implemented a classification system based upon quantitative morphological and kinetic features in improving the specificity of breast MRI. Morphological and kinetic features of the lesion were extracted automatically, and then the feature selection step was utilized to select the most relevant features to maximize the classifier performance. In our study, the area under the receiver operating curve (AUC) is used as the performance metric of the classifier, and our results are competitive with those of previous studies. The dissertation concludes by summarizing the contribution of this project and suggesting the future directions of quantitative and highly automated approaches to breast MR image analysis.Item Evaluation of load transfer mechanisms between soil and geogrid using transparent soil(2017-12-11) Peng, Xin, Ph. D.; Zornberg, Jorge G.; Bhasin, Amit; Cox, Brady; Liechti, Kenneth; Williamson, EricThe increasing use of different types of geosynthetics in the design of reinforced soil retaining structures and of stiffened roadways require proper evaluation of the mobilization of load-transfer mechanisms between soil particles and geosynthetics. Compared to other planar geosynthetics, geogrids develop a soil-interlocking interaction that leads to the mobilization of different types of resistance from different rib elements. The resistance mobilized by the different rib elements ultimately determine the mechanical responses for both: (1) the ultimate pullout resistance; and (2) the confined geogrid stiffness. Geogrids with different aperture shapes involve rib elements along different orientations. Different rib orientations, as well as different rib dimensions, may lead to different load transfer mechanisms. The contributions of the different mechanisms have been difficult to be quantify and evaluate using conventional experimental techniques. In this study, a new experimental approach involving the use of transparent soil with laser aided imaging, was developed to visualize and quantify load-transfer mechanisms between soil particles and geogrids with different geometric characteristics. Laser beams with up to 350 mW output power and with a wavelength of 638 nm were adopted to allow tracking the transparent soil particles at a plane perpendicular to the soil-geogrid interface. The collimated beam resulted in well-defined individual particles in the selected plane of the soil model. High-definition cameras were used to track the displacement fields of both the confined geogrid specimen and the soil particles within the laser-illuminated plane. Digital Image Correlation (DIC) techniques, as well as other image-processing approaches were used to define the displacement fields based on images captured during the tests. Using this experimental approach, a series of investigations were conducted based on soil-geosynthetic interaction (SGI) tests: (1) experimental evaluation of the confined performance of geogrids with different geometric characteristics; (2) load transfer modeling of the soil-geogrid interaction using both biaxial and triaxial geogrids. The experimental evaluation illustrated the impact of the geometric characteristics of geogrids (e.g. aperture shapes, aperture sizes, and rib dimensions), on the mobilization of load transfer between soil particles and geogrid specimens. Specifically, the digitally-collected data allowed determination of: (a) geogrid displacement profiles, (b) load-displacement relationships of geogrid specimens, (c) soil stiffening along the loading direction, and (d) the development of shear bands. Using the proposed testing configuration, the experimental results allowed the comparison between the response of triaxial geogrids and that of the biaxial geogrid. Load transfer models of soil-geogrid interaction were developed. They involve independent modeling for (1) longitudinal ribs, and (2) transverse or diagonal ribs of the geogrid specimen. The models developed in this study were validated using the results from tests conducted using the selected biaxial and triaxial geogrids. Good agreement was observed between the experimental measurements and model predictions. The load transfer models also allowed the evaluation of the interference between different load transfer mechanisms during the SGI tests conducted using the biaxial geogrid as a part of this study. In addition, the difference between the load transfer mechanisms developed in the biaxial geogrid and those developed in the triaxial geogrids was evaluated and quantified in this study. Overall, the triaxial geogrids tested as a part of this study showed a higher load transfer efficiency than the biaxial geogrid along the loading direction. This was probably attributed to the comparatively uniform unit tension distribution exhibited by the triaxial geogrids in relation to that of the biaxial geogrid used in this study. The governing component of the resistance mobilized at low displacement levels in the test conducted using the selected biaxial geogrid was the resistance developed by the longitudinal ribs. Instead, at high displacement levels and until pullout failure, the relative contributions of the transverse ribs were larger than those of the longitudinal ribs. The most relevant component of the resistance mobilized throughout the entire tests conducted using the triaxial geogrids as a part of this study, both under low and large displacements, was found to be the resistance developed by the diagonal ribs, with their relative contributions particularly high during the early stages of the tests. In the tests conducted using different triaxial geogrids, the relative contributions of the resistance from different rib components were found to depend on the geometric characteristics of the geogrids. Overall, the conclusions addressed from this study provide valuable insight into identifying the impact of the geometric characteristics of geogrids on the load transfer mechanisms between soil particles and geogrids.Item Evaluation of Superpave Fine Aggregate Angularity Specification(2001-05) Chowdhury, Arif; Button, Joe W.; Kohale, Vipin; Jahn, David W.The validity of the Superpave fine aggregate angularity (FAA) requirement is questioned by both the owner agencies and the paving and aggregate industries. The FAA test is based on the assumption that more fractured faces will result in higher void content in the loosely compacted sample; however, this assumption is not always true. Some agencies have found that cubical shaped particles, even with 100 percent fractured faces, may not meet the FAA requirement for high-volume traffic. State agencies are concerned that local materials, previously considered acceptable and which have provided good field performance, cannot meet the Superpave requirements. Researchers evaluated angularity of 23 fine aggregates representing most types of paving aggregates used in the USA using seven different procedures: FAA test, direct shear test, compacted aggregate resistance (CAR) test, three different image analyses, and visual inspection. The three image analyses techniques included Hough Transform at University of Arkansas at Little Rock (UALR), unified image analysis at Washington State University (WSU), and VDG-40 videograder at Virginia Transportation Research Council (VTRC). A small study was performed to evaluate relative rutting resistance of HMA containing fines with different particle shape parameters using the Asphalt Pavement Analyzer (APA). The FAA test method does not consistently identify angular, cubical aggregates as high quality materials. There is a fair correlation between the CAR stability value and angle of internal friction (AIF) from the direct shear test. No correlation was found between FAA and CAR stability or between FAA and AIF. Fairly good correlations were found between FAA and all three image analysis methods. Some cubical crushed aggregates with FAA values less than 45 gave very high values of CAR stability, AIF, and ‘angularity’ from imaging techniques. Moreover, the three image analysis methods exhibited good correlation among themselves. A statistical analysis of the SHRP-LTPP (Strategic Highway Research Program-Long-Term Pavement Performance) database revealed no significant evidence relationship between FAA and rutting. This lack of relationship is not surprising since many uncontrolled factors contribute to pavement rutting. The APA study revealed that FAA is not sensitive to rut resistance of HMA mixtures. Image analysis methods appear promising for measuring fine aggregate angularity. Until a suitable replacement method(s) for FAA can be identified, the authors recommend that the FAA criteria be lowered from 45 to 43 for 100 percent crushed aggregate. Analysis of the FAA versus rutting data should be examined later as the amount of data in the SHRP-LTTP database is expanded.Item Grain-scale controls on seal integrity in mudrocks : capillary entry pressure and permeability prediction(2020-06-26) Bihani, Abhishek Dilip; Daigle, Hugh; Lake, Larry W; Prodanovic, Masa; Espinoza, David N; Hayman, Nicholas WMudrocks serve as geological traps and seals for carbon sequestration or for hydrocarbon formation, where mudrock capillary seals having high capillary entry pressure prevent the leakage of underlying fluids. However, they can fail if the buoyant pressure of the trapped fluid overcomes the threshold pressure of the seal. Mudrocks are composed primarily of silt-size and clay-size grains in various fractions. Microstructural observations of mudrocks have shown a silt bridging effect, whereby sufficiently abundant silt-size grains will create a stress chain across the rock matrix to preserve large pores and throats. At shallower depths, this effect can create a dual porosity system, consisting of larger pores and throats near the coarser grains, and smaller pores and throats existing only between the finer clay grains. If the preserved larger pores and throats are connected across a mudrock, it may increase the absolute permeability, and reduce the capillary threshold pressure and tortuosity, thereby decreasing its sealing capacity. Using pore-network modeling, artificial bidisperse grain packs (packings of two sizes) were generated, with and without the effect of gravity, to understand the effects of deposition and compaction on the petrophysical properties. It was observed that when the fraction of larger grains reaches about 40 - 60 % of the total volume of the grain pack, the capillary threshold transitions to a lower value and permits fluid percolation across the grain pack. The discrete element modeling (DEM) compaction simulations showed that on increasing large grain concentrations, strong force chains are formed across large-large and small-large grain contacts which decreases coordination numbers and shields larger pores. An image analysis workflow consisting of multiple filtering and user-guided segmentation steps was used to identify pores, silt grains, and clay from scanning electron microscope (SEM) images of sediments from the Kumano Basin offshore Japan. Statistical analysis showed that larger pores are better preserved when surrounded by detrital, silt size grains, and the presence of a higher fraction of silt-size grains led to a higher concentration of larger pores. The distributions of pore characteristics at different depths showed that larger pores are observed in samples with higher silt fractions despite being deeper. Since the images only offer a 2D view of the 3D rock structure, a digital rocks workflow was applied to reconstruct the mudrock pore space. Lattice Boltzmann simulations were run on the reconstructed grain packs to simulate capillary drainage using high-performance computing. The results showed that at all depths, the capillary threshold pressure for the grain packs with a higher silt fraction was lower than those with a lower silt fraction and that capillary threshold pressure also increased with depth. Thus, using a combination of pore-network modeling, DEM simulations, image analysis, and lattice Boltzmann simulations, I found that there is a significant dependence of the grain concentration and texture on the petrophysical properties. Improving our understanding about the influence of grain concentrations, spatial positions, and sizes on the fluid flow behavior is an important step towards a better characterization of mudrock seals and can help improve risk management efforts in anthropogenic waste storage and estimates of the reserve capacity of petroleum reservoirsItem Hydro-mechanical characterization of unsaturated clays using centrifuge technology(2017-12) Quaglia, Gastón; Zornberg, Jorge G.; Caldwell, Todd; El Mohtar, Chadi; Espinoza, Nicolas D; Gilbert, Robert BA number of experimental techniques and analytical methods, with emphasis in centrifuge testing, were implemented in this research to characterize the hydro-mechanical behavior of unsaturated clays. In particular, experimental procedures and back-analysis methods were developed to determine the Soil Water Retention Surface (SWRS) and the unsaturated hydraulic conductivity (k-function) of clays, both of low and high plasticity. New experimental devices and non-intrusive sensors were developed to expressly incorporate four key variables that control unsaturated flow processes: moisture content, suction, volumetric changes, and hydraulic conductivity. The new equipment and sensors are capable of monitoring all relevant variables without interfering with the flow process or the volumetric changes. Specifically, two nonintrusive sensor systems were implemented to upgrade the capabilities of the centrifuge permeameter for unsaturated soils (CPUS): a water content sensor, the GTDR, based on Time Domain Reflectometry, and an image-analysis tool to quantify soil deformation inflight. When combined, both systems combined allowed continuously monitoring the volumetric and water content changes in-flight. These tools allowed generating experimental data for the SWRS and hydraulic conductivity of clays, using steady-state and transient procedures. In addition, a new workbench equipment, the ATX Cell, was developed to facilitate the characterization of the hydro-mechanical behavior of unsaturated clays by simultaneously monitoring the void ratio, matric suction, and water content. It allows testing clays under different imposed stresses, a feature that was particularly relevant when testing on expansive clays. The hydraulic properties of an unsaturated low plasticity clay (RMA soil) were evaluated using centrifuge (N.g) testing with the implementation of the new sensors, as well as other standard (1.g) experimental techniques. RMA’s SWRS was determined over a wide range of void ratio values with samples compacted at optimum water content. Although volumetric changes measured in the ATX Cell were negligible, the level of deformation in RMA soil samples measured during centrifuge testing was found to depend on the initial conditions and stresses imposed during testing A comparison between destructive and non-destructive measurement revealed that destructive measurements result in changes that affect the measured volumetric water content and degree of saturation profiles. Non-intrusive sensors were found to solve this problem improving the definition of the SWRS. RMA k-function was evaluated using both, steady-state and transient methods. Gardner’s outflow method was applied to the transient measurements obtained from the ATX Cell, and Hydrus 1-D code was implemented to determine the k-function using transient information from centrifuge tests. Hydrus-1D overestimate the saturated hydraulic conductivity, however, the k-function was found to show good agreement with the hydraulic conductivity values obtained using steady-state (imposed flow) approach The hydro-mechanical behavior of an unsaturated high plasticity clay, Eagle Ford was characterized implementing all the sensors, devices and testing approaches previously developed in this research. Special focus was placed on the hydraulic properties, as well as on the swelling characteristics of this clay implementing centrifuge technology. The inclusion of the void ratio as an additional variable was found to be particularly relevant when interpreting the experimental data and defining the SWRS and k-function of the clay. The ATX Cell was used to evaluate the hydro-mechanical characteristics of Eagle Ford clay by monitoring the void ratio and water changes through different suction stages under different values of imposed stresses. The volume changes and changes in water content were found to depend on the initial void ratio and the stresses imposed. A coupled behavior was identified between the swelling and the inflow rates. In addition, a series of filter paper and chilled mirror hygrometer test were carried out. The experimental data generated with standard (1.g) methods incorporating the void ratio as an additional variable allowed defining the SWRS. The results from centrifuge tests in Eagle Ford clay showed that volume changes occur as a result of the coupled effect of the imposed stresses (due to the gravitational forces) and water content changes (wetting and drying phases). Image analysis was found to successfully detect the swelling and contraction of the soil in-flight. The volumetric water content measured with the GTDR’s was found to match the values measured using semi-destructive techniques. The use of the GTDR and image analysis techniques allowed obtaining experimental data to define the SWRS of Eagle Ford clay using centrifuge testing. A comparison between two centrifuge (hydrostatic) tests performed at 100g and 200g respectively showed that the void was reduced as a result of the higher imposed stresses, and in consequence a larger portion of the sample remained saturated despite of the higher suction imposed. Despite the higher hydraulic gradient, the flow velocity was smaller in the second test due to a possible reduction in the hydraulic conductivity. While steady-state techniques were successful implemented for low plasticity clays, this approach was time-consuming for high plasticity clays. The interpretation of transient information was found to become useful at determining the hydraulic conductivity. Swelling tests performed in the centrifuge showed that volumetric water content increased rapidly during primary swelling stage, while secondary swelling occurred under an approximately constant degree of saturation. The image analysis tools allowed defining volumetric changes along the sample, providing additional information to generate the swell-stress curve. The representation of the swelling test in the -e or Sr-e revealed that for samples compacted at different water contents but tested at the same vertical stress final equilibrium state after fully wetting was similar. Although this representation of the swelling test data eliminates the variable time it could be used to provide a method to evaluate the expansion in a soil profile under partial wetting conditions if the swelling is calculated using the void ratio values for a selected initial and final water content. A comparison in the [θ, e] plane of the results obtained from an ATX Cell test and a swelling tests performed at the same vertical stress in samples compacted at different water contents revealed that the final equilibrium achieved by both samples after fully wetting are similar.Item Investigation of the petrophysical properties of unconventional rocks using multiscale network modeling(2015-08) Mehmani, Ayaz; Prodanović, Maša; Bryant, Steven; Dicarlo, David; Milliken, Kitty; Sepehrnoori, Kamy; Torres-Verdin, CarlosUnconventional reservoirs, specifically carbonates, tight gas sandstone and shale gas formations, provide significant potential for the growing world energy demand. However, the positive prospects of these reserves are hampered by considerable uncertainty in estimating their production. Reliable petrophysical models of these media can help reduce the uncertainty in their development. Pore-network models are cost-efficient representations of a porous medium’s pore structure that allow prediction of its macroscopic properties. In this effort, the topology and fluid physics of pores from various scales are integrated into a single-entity three-dimensional (3D) unstructured pore-network model. We start with the simplest shale matrix gas flow model that incorporates pores from nanometer and micrometer scales, but has a connectivity resembling conventional rocks. We quantify the apparent permeability of these networks with relevant, pore size-dependent physical models applied to both scales and compare the results with the continuum no slip boundary condition assumption. The discrepancy between the two can run over several orders of magnitude and grows with the fraction of nanopores and the width of the overall pore size distribution. We next attempt to create more realistic network models, closer to the true topology of the studied unconventional rocks. Workflows for integration of nanometer and micrometer pore structures are then developed for deterministic, geologically informed, process-based and image-based approaches in various unconventional scenarios. We perform a systematic forward analysis of the applicability of tracer breakthrough profiles (TBPs) in revealing the pore structure of tight gas sandstone and carbonate formations. The following features are modeled via 3D multiscale networks: microporosity within dissolved grains and pore-filling clay, cementation in the absence and presence of microporosity (each classified into uniform, pore-preferred, and throat-preferred modes), layering, and vug and microcrack inclusion. A priori knowledge of the extent and location of each process is assumed known. In general, significant qualitative perturbation of the TBPs is observed for uniform and throat-preferred cementation. Layering parallel to the fluid flow direction has a considerable impact on TBPs whereas layering perpendicular to flow does not. Microcrack orientation has a minor effect in perturbing TBPs. In most scenarios TBPs show negligible qualitative sensitivity to the fraction of micropores present. The exception is the case when macropores and pore-filling micropores have equivalent flux contributions. A quantitative parameterization of sensitivity is not conducted; an example of such is measuring the perturbations in pore-volumes associated with the breakthrough profile peaks, has not been conducted. Similar to tracer breakthrough profiles in the context of characterizing heterogeneous porous media in core scale, nitrogen sorption hysteresis is investigated for characterizing pore structure of mudrocks. Three network types are introduced to represent their multiscale pore topology; specifically: regular (Type 1), series (Type 2) and parallel (Type 3). We conclude that, in appropriate size ranges, sorption hysteresis can distinguish the three types whereas permeability hysteresis can only separate parallel from series and regular. Furthermore, the simulations show that sorption hysteresis is sensitive to compaction/cementation (closing of throats) in all network types, whereas permeability hysteresis is sensitive to the diagenesis in parallel networks only. A quantitative parameterization of the sensitivity, such as measuring the area enclosed by the hysteresis curve, was not conducted. Molecular diffusion is an important mechanism for hydrocarbon transport within matrix as well as between matrix pores and hydraulic fractures in unconventional shale production. The diffusion coefficient is also an essential parameter in two-dimensional (2D) nuclear magnetic resonance (NMR) map interpretations. However, molecular diffusion in the micro- and nanometer scale pore networks of unconventional shale rocks remain poorly understood. We attempt to link the restricted diffusion coefficient to pore-scale characteristics of shale gas media. A random walk algorithm with discrete time steps is implemented to investigate the effects of pore-throat ratio (the ratio of pore-body radius to pore-throat radius), length ratio (the ratio of throat length to pore radius), pore shape and topology. It is concluded that, at an equal surface-to-volume ratio, diffusion coefficient increases in pores with higher angularity. The effects of pore-throat radius ratio and length ratio are explicitly modeled in 3D structured regular lattices. Results indicate that, up to pore-throat radius ratios of 5, restricted diffusion is considerable in lattices with zero length throats. Furthermore, restricted diffusion decreases with the increase in length ratio. To reduce computational costs, a statistical method is developed to render simulating the effects of connectivity and pore size distribution on 3D unstructured multiscale networks feasible. Finally, we perform a preliminary assessment of the fidelity of the multiscale process-based and image-based approaches in a case study conducted on the Wilcox tight gas sandstone. A novel workflow that combines the multiscale process-based network model with petrographic analysis is developed. This methodology utilizes petrographic information (grain size distribution and sorting, cement type and thickness, microporosity types and fractions, burial sequence) to enable the prediction of the flow properties of the medium in several burial stages throughout the paragenesis of the Wilcox formation. Given the 100 -1000 times scale difference between micropores and macropores and the resultant computational costs, an upscaling scheme is proposed for the microporous clusters in the process-based algorithm. The upscaling presently does not work for the image-based modeling because of the irregularity of microporous regions. We observe discrepancy between the simulated and experimental mercury injection capillary pressure curve and use it to recommend future improvements to the workflow. In this case study, micropores are crucial in contributing to the flow path; therefore, their surface chemistry as well as physical features such as surface roughness must be quantified and taken into account to make reliable predictions of the rock flow properties.Item Investigations of porous media using nuclear magnetic resonance secular relaxation measurements and micro-CT image analysis(2015-08) Johnson, Andrew Charles; Daigle, Hugh; Torres-Verdin, CarlosNuclear magnetic resonance (NMR) has been used as a common and powerful tool for petrophysical investigation of fluid-bearing porous media. A common application in this field is the extraction of pore size distributions, which are important descriptors of pore system morphologies. The common technique is to correlate mercury-injection porosimetry (MICP) measurements with NMR T₁ or T₂ distributions to obtain NMR-derived pore sizes. The limitations of MICP include pore-throat sensitivity and percolation effects, which compromise interpretation of results. Micro-CT image analysis has no such limitations, and measurements of pore size are characterized by pore body voxel counts. Presented are image analysis and NMR-correlated results for samples of Berea sandstone and Silurian dolomite. These results are compared to MICP-correlated results and the discrepancies interpreted as pore throat-to-body aspect ratios. Nuclear magnetic resonance pore size distributions are valid for single-phase fluids in the fast-diffusion NMR relaxation regime. When the effects of proton diffusion through internal magnetic field gradients become prominent, however, this relationship becomes entangled. Simultaneous measurement of longitudinal (T₁) and transverse (T₂) relaxation times using combined inversion recovery-CPMG pulse sequences allows for interpretation of a computed NMR attribute called secular relaxation (T₂sec). This quantity is defined as the difference in transverse and longitudinal relaxation rates (1/T₂-1/T₁) and can reveal important pore system properties. Presented are results that extract internal magnetic field gradient strengths based on changes in T₂sec as a function of the NMR experimental parameter τ. Further results consider a two-dimensional χ² analysis to attempt to invert for mean pore size and the difference in transverse and longitudinal surface relaxivities. The benefit of these types of analyses is to provide a simple methodology for inferring the average strengths of internal magnetic field gradients and pore sizes from NMR measurements without the need for independent measurements of pore size distributions, such as from mercury injection porosimetry. In addition, secular relaxation analysis removes the effects of bulk fluid relaxation.Item Methods for analysis in digital images of sedimentary rocks(2020-05) Tang, David Guo; Spikes, Kyle; Sen, Mrinal K; Fomel, Sergey B; Daigle, Hugh C; Heidari, ZoyaThis dissertation focuses on the use of digital images to analyze sedimentary rocks. While digital images can provide important insight into a rock, the analysis of such images can be extremely time consuming. To address this problem, I explore various methods for the automation of this process. The utility of these segmented images are then demonstrated across different disciplines in geoscience. In the first study, I explore the use of traditional image processing methods to analyze a thin section for sedimentology purposes. A marker-based watershed approach incorporating both a distance and gradient map is used to identify individual grains. The algorithm is then used to measure the spatial distribution of grain size within a Permian Basin reservoir. The second study utilizes a neural network for identifying six different mineral types in a scanning electron microscope image of a shale. After assigning physical properties to each pixel, a digital rock physics experiment using finite elements is used to simulate a laboratory experiment for the estimation of elastic properties. The final study focuses on the automation of point counting in a thin section for petrography. A convolutional neural network utilizing both plane- and cross-polarized light images is used to identify the percentages of each grain type across an image. The results are comparable with a manual point count. As the amount of data collected becomes increasingly large, the automation of image analysis using the methods proposed here will allow for future users to feasibly work with such data sets.Item Multiscale visualization of chemical enhanced oil recovery(2021-12-09) Mejia, Lucas; Balhoff, Matthew T.; Song, Wen; Mohanty, Kishore K; DiCarlo, David A; Sultan, AbdullahChemical enhanced oil recovery (EOR) involves injecting chemicals such as surfactants, polymers, and alkalis into depleted oil reservoirs to increase oil recovery. Experimental tools such as corefloods and micromodels provide critical insights for the mechanistic understanding and screening of chemicals for EOR. Coreflood experiments are especially valuable for screening. However, imaging cores at high resolution for mechanistic understanding is challenging. Micromodels, synthetic optically accessible porous media that resemble rocks, address some of these challenges by allowing facile high-resolution imaging of displacements at the pore scale. However, they lack many important features of cores. In this work, we develop a novel micromodel, referred to herein as the Coreflood on a Chip, that permits visualization at the pore and core scales. Then, we investigate various forms of chemical EOR in the Coreflood on a Chip including viscous waterflooding, surfactant flooding, and alkali-surfactant-polymer flooding (ASP). Viscous waterflooding experiments were performed by injecting viscous glycerol solution or polymer solution into oil-saturated micromodels with irreducible water. We analyzed the experiments using fractional flow theory and pore-scale lattice Boltzmann simulations and found that irreducible water causes viscous fingering even at very favorable viscosity ratios. Additionally, we conducted corefloods to corroborate our findings translated to displacements in real rocks. Next, we performed surfactant and ASP floods by injecting aqueous chemical solutions with fluorescein into micromodels with oil and brine. This way, we could differentiate the injected aqueous phase from the resident aqueous phase. Our results prove that surfactants are present in and ahead of oil banks. Moreover, our results show that while saturations are well described by a fractional flow solution with two curves, aqueous phase concentrations are not properly described by said model because injected aqueous phase is present ahead of oil banks. Finally, we present a novel microfluidic device that can be utilized to conduct phase behavior salinity scans. We show the prototype microfluidic device is viable for performing salinity assays of crude oil-surfactant-water systems and segregation of microemulsion is visible in the device. In this work we demonstrate that evaluation of chemical EOR across scales permits identifying causal relationships between pore-scale processes and previously unobserved core scale behavior.Item Rapid Test to Establish Grading of Unbound Aggregate Products: Automation of Aggregate Characterization Using Laser Profiling and Digital Image Analysis(2002-06) Haas, Carl T. (Carl Thomas); Rauch, Alan F.; Kim, Hyoungkwan; Browne, CraigThis final project report describes the development of a laser scanning device for measuring the gradation and other morphological characteristics of unbound construction aggregates. The device is called the LASS, for "Laser-based Aggregate Scanning System." The method of characterizing aggregate particles from three-dimensional (3D) laser profiling required the development of particle segmentation algorithms, particle measurement algorithms, and generalized particle descriptions. With the LASS, true 3D data of aggregate particles are obtained by laser profiling. This data is first transformed into digital images. Second, segmentation and particle measurement algorithms separate the particles and process each particle data individually with the aid of various digital image technologies. Finally, in order to provide a generalized, quantitative, and representative way to characterize aggregate particles, 3D particle descriptors were developed using the multi-resolution analysis feature of wavelet transforms. Verification tests show that this approach can characterize various aggregate properties in a fast, accurate, and reliable way. When implemented, this ability to automatically analyze multiple characteristics of an aggregate sample will lead to reduced labor costs in the laboratory, but more importantly, to better control of the quality of aggregate products. ICAT Project 503 was undertaken to study rapid, automated methods of determining the grain size distribution of unbound aggregate products. Automatic measurement of particle size and shape properties has the potential to overcome problems with manual measurements such as subjectivity, labor intensity, and slow speed. Testing machines that rely on 2D digital image analysis were evaluated as described earlier in Report ICAR 503-2.Item Rapid Test to Establish Grading of Unbound Aggregate Products: Evaluation of Potential Aggregate Grading Technologies(2000-02) Rauch, Alan F.; Haas, Carl T. (Carl Thomas); Kim, Hyoungkwan; Browne, CraigA research study is underway to develop automated methods for rapidly grading aggregates on the production line in a typical aggregate separation or mixing facility. This interim report serves to document preliminary work on this project and presents: (1) A discussion of typical plant layouts, to identify potential sampling locations. The best sampling locations appear to be just after final screening, where sorted material is sent to stockpiles, and just before mixing the final product, where aggregates are fed from either stockpiles or charge bins. (2) A thorough examination of six potential technologies that could be used to rapidly determine particle size. After a critical review and a formal decision analysis, both digital image analysis and laser profiling appear to be equally promising and worthy of additional study. (3) A discussion of our current thinking on how to configure scanning equipment of this kind. By considering innovative methods for presenting aggregate particles to the scanning sensor, the opportunity exists for advancing this technology. An outline of project work planned for the immediate future. Continuing our study of both digital image analysis and laser profiling, we plan to: 1. Conduct a limited, independent evaluation of three commercial particle-sizing machines, which all use digital image analysis. 2. Perform preliminary tests using a laser profiler. 3. Construct a laboratory test bed that can be used to test and evaluate various scanning sensors.Item Two technologies for single-molecule proteomics, three technologies for image analysis(2020-02-05) Boulgakov, Alexander Alexeyvich; Marcotte, Edward M.Proteins are central players in biology. Being able to detect and quantify proteins in various circumstances such as in biochemically fractionated cellular lysates has proven to be highly informative about their characteristics, functions, and relationships with other proteins and cellular components in general. We currently lack high-throughput technologies for quantifying proteins that would resolve complex mixtures at single-molecule resolution across the large dynamic ranges found in cells. Here I present progress towards the two single-molecule proteomics technologies my colleagues and I have been developing: fluorosequencing and reverse translation. We computationally explore the feasibility and informational power of each technique, motivating further work. We demonstrate a working proof-of-concept for fluorosequencing, and significant progress towards a proof-of-concept for reverse translation. In addition to proteomics, I have contributed three computational technologies that can broadly be grouped as image analysis: quantitation of fluorescent nerve agent probes by chromaticity changes; recovery of molecular positions by DNA sequencing of immobilized, barcoded oligonucleotides; and automated, quantitative co-localization of protein puncta in cells.