Matching Items (64)
Description
There is a growing interest for improved high-accuracy camera calibration methods due to the increasing demand for 3D visual media in commercial markets. Camera calibration is used widely in the fields of computer vision, robotics and 3D reconstruction. Camera calibration is the first step for extracting 3D data from a 2D image. It plays a crucial role in computer vision and 3D reconstruction due to the fact that the accuracy of the reconstruction and 3D coordinate determination relies on the accuracy of the camera calibration to a great extent. This thesis presents a novel camera calibration method using a circular calibration pattern. The disadvantages and issues with existing state-of-the-art methods are discussed and are overcome in this work. The implemented system consists of techniques of local adaptive segmentation, ellipse fitting, projection and optimization. Simulation results are presented to illustrate the performance of the proposed scheme. These results show that the proposed method reduces the error as compared to the state-of-the-art for high-resolution images, and that the proposed scheme is more robust to blur in the imaged calibration pattern.
ContributorsPrakash, Charan Dudda (Author) / Karam, Lina J (Thesis advisor) / Frakes, David (Committee member) / Papandreou-Suppappola, Antonia (Committee member) / Arizona State University (Publisher)
Created2012
Description
Diabetic retinopathy (DR) is a common cause of blindness occurring due to prolonged presence of diabetes. The risk of developing DR or having the disease progress is increasing over time. Despite advances in diabetes care over the years, DR remains a vision-threatening complication and one of the leading causes of blindness among American adults. Recent studies have shown that diagnosis based on digital retinal imaging has potential benefits over traditional face-to-face evaluation. Yet there is a dearth of computer-based systems that can match the level of performance achieved by ophthalmologists. This thesis takes a fresh perspective in developing a computer-based system aimed at improving diagnosis of DR images. These images are categorized into three classes according to their severity level. The proposed approach explores effective methods to classify new images and retrieve clinically-relevant images from a database with prior diagnosis information associated with them. Retrieval provides a novel way to utilize the vast knowledge in the archives of previously-diagnosed DR images and thereby improve a clinician's performance while classification can safely reduce the burden on DR screening programs and possibly achieve higher detection accuracy than human experts. To solve the three-class retrieval and classification problem, the approach uses a multi-class multiple-instance medical image retrieval framework that makes use of spectrally tuned color correlogram and steerable Gaussian filter response features. The results show better retrieval and classification performances than prior-art methods and are also observed to be of clinical and visual relevance.
ContributorsChandakkar, Parag Shridhar (Author) / Li, Baoxin (Thesis advisor) / Turaga, Pavan (Committee member) / Frakes, David (Committee member) / Arizona State University (Publisher)
Created2012
Description
Image segmentation is of great importance and value in many applications. In computer vision, image segmentation is the tool and process of locating objects and boundaries within images. The segmentation result may provide more meaningful image data. Generally, there are two fundamental image segmentation algorithms: discontinuity and similarity. The idea behind discontinuity is locating the abrupt changes in intensity of images, as are often seen in edges or boundaries. Similarity subdivides an image into regions that fit the pre-defined criteria. The algorithm utilized in this thesis is the second category.
This study addresses the problem of particle image segmentation by measuring the similarity between a sampled region and an adjacent region, based on Bhattacharyya distance and an image feature extraction technique that uses distribution of local binary patterns and pattern contrasts. A boundary smoothing process is developed to improve the accuracy of the segmentation. The novel particle image segmentation algorithm is tested using four different cases of particle image velocimetry (PIV) images. The obtained experimental results of segmentations provide partitioning of the objects within 10 percent error rate. Ground-truth segmentation data, which are manually segmented image from each case, are used to calculate the error rate of the segmentations.
This study addresses the problem of particle image segmentation by measuring the similarity between a sampled region and an adjacent region, based on Bhattacharyya distance and an image feature extraction technique that uses distribution of local binary patterns and pattern contrasts. A boundary smoothing process is developed to improve the accuracy of the segmentation. The novel particle image segmentation algorithm is tested using four different cases of particle image velocimetry (PIV) images. The obtained experimental results of segmentations provide partitioning of the objects within 10 percent error rate. Ground-truth segmentation data, which are manually segmented image from each case, are used to calculate the error rate of the segmentations.
ContributorsHan, Dongmin (Author) / Frakes, David (Thesis advisor) / Adrian, Ronald (Committee member) / Turaga, Pavan (Committee member) / Arizona State University (Publisher)
Created2015
Description
Over the past three decades, particle image velocimetry (PIV) has been continuously growing to become an informative and robust experimental tool for fluid mechanics research. Compared to the early stage of PIV development, the dynamic range of PIV has been improved by about an order of magnitude (Adrian, 2005; Westerweel et al., 2013). Further improvement requires a breakthrough innovation, which constitutes the main motivation of this dissertation. N-pulse particle image velocimetry-accelerometry (N-pulse PIVA, where N>=3) is a promising technique to this regard. It employs bursts of N pulses to gain advantages in both spatial and temporal resolution. The performance improvement by N-pulse PIVA is studied using particle tracking (i.e. N-pulse PTVA), and it is shown that an enhancement of at least another order of magnitude is achievable. Furthermore, the capability of N-pulse PIVA to measure unsteady acceleration and force is demonstrated in the context of an oscillating cylinder interacting with surrounding fluid. The cylinder motion, the fluid velocity and acceleration, and the fluid force exerted on the cylinder are successfully measured. On the other hand, a key issue of multi-camera registration for the implementation of N-pulse PIVA is addressed with an accuracy of 0.001 pixel. Subsequently, two applications of N-pulse PTVA to complex flows and turbulence are presented. A novel 8-pulse PTVA analysis was developed and validated to accurately resolve particle unsteady drag in post-shock flows. It is found that the particle drag is substantially elevated from the standard drag due to flow unsteadiness, and a new drag correlation incorporating particle Reynolds number and unsteadiness is desired upon removal of the uncertainty arising from non-uniform particle size. Next, the estimation of turbulence statistics utilizes the ensemble average of 4-pulse PTV data within a small domain of an optimally determined size. The estimation of mean velocity, mean velocity gradient and isotropic dissipation rate are presented and discussed by means of synthetic turbulence, as well as a tomographic measurement of turbulent boundary layer. The results indicate the superior capability of the N-pulse PTV based method to extract high-spatial-resolution high-accuracy turbulence statistics.
ContributorsDing, Liuyang (Author) / Adrian, Ronald J (Thesis advisor) / Frakes, David (Committee member) / Herrmann, Marcus (Committee member) / Huang, Huei-Ping (Committee member) / Peet, Yulia (Committee member) / Arizona State University (Publisher)
Created2018
Description
Motion estimation is a core task in computer vision and many applications utilize optical flow methods as fundamental tools to analyze motion in images and videos. Optical flow is the apparent motion of objects in image sequences that results from relative motion between the objects and the imaging perspective. Today, optical flow fields are utilized to solve problems in various areas such as object detection and tracking, interpolation, visual odometry, etc. In this dissertation, three problems from different areas of computer vision and the solutions that make use of modified optical flow methods are explained.
The contributions of this dissertation are approaches and frameworks that introduce i) a new optical flow-based interpolation method to achieve minimally divergent velocimetry data, ii) a framework that improves the accuracy of change detection algorithms in synthetic aperture radar (SAR) images, and iii) a set of new methods to integrate Proton Magnetic Resonance Spectroscopy (1HMRSI) data into threedimensional (3D) neuronavigation systems for tumor biopsies.
In the first application an optical flow-based approach for the interpolation of minimally divergent velocimetry data is proposed. The velocimetry data of incompressible fluids contain signals that describe the flow velocity. The approach uses the additional flow velocity information to guide the interpolation process towards reduced divergence in the interpolated data.
In the second application a framework that mainly consists of optical flow methods and other image processing and computer vision techniques to improve object extraction from synthetic aperture radar images is proposed. The proposed framework is used for distinguishing between actual motion and detected motion due to misregistration in SAR image sets and it can lead to more accurate and meaningful change detection and improve object extraction from a SAR datasets.
In the third application a set of new methods that aim to improve upon the current state-of-the-art in neuronavigation through the use of detailed three-dimensional (3D) 1H-MRSI data are proposed. The result is a progressive form of online MRSI-guided neuronavigation that is demonstrated through phantom validation and clinical application.
The contributions of this dissertation are approaches and frameworks that introduce i) a new optical flow-based interpolation method to achieve minimally divergent velocimetry data, ii) a framework that improves the accuracy of change detection algorithms in synthetic aperture radar (SAR) images, and iii) a set of new methods to integrate Proton Magnetic Resonance Spectroscopy (1HMRSI) data into threedimensional (3D) neuronavigation systems for tumor biopsies.
In the first application an optical flow-based approach for the interpolation of minimally divergent velocimetry data is proposed. The velocimetry data of incompressible fluids contain signals that describe the flow velocity. The approach uses the additional flow velocity information to guide the interpolation process towards reduced divergence in the interpolated data.
In the second application a framework that mainly consists of optical flow methods and other image processing and computer vision techniques to improve object extraction from synthetic aperture radar images is proposed. The proposed framework is used for distinguishing between actual motion and detected motion due to misregistration in SAR image sets and it can lead to more accurate and meaningful change detection and improve object extraction from a SAR datasets.
In the third application a set of new methods that aim to improve upon the current state-of-the-art in neuronavigation through the use of detailed three-dimensional (3D) 1H-MRSI data are proposed. The result is a progressive form of online MRSI-guided neuronavigation that is demonstrated through phantom validation and clinical application.
ContributorsKanberoglu, Berkay (Author) / Frakes, David (Thesis advisor) / Turaga, Pavan (Thesis advisor) / Spanias, Andreas (Committee member) / Berisha, Visar (Committee member) / Arizona State University (Publisher)
Created2018
Description
Introduction: There are 350 to 400 pediatric heart transplants annually according to the Pediatric Heart Transplant Database (Dipchand et al. 2014). Finding appropriate donors can be challenging especially for the pediatric population. The current standard of care is a donor-to-recipient weight ratio. This ratio is not necessarily a parameter directly indicative of the size of a heart, potentially leading to ill-fitting allografts (Tang et al. 2010). In this paper, a regression model is presented - developed by correlating total cardiac volume to non-invasive imaging parameters and patient characteristics – for use in determining ideal allograft fit with respect to total cardiac volume.
Methods: A virtual, 3D library of clinically-defined normal hearts was compiled from reconstructed CT and MR scans. Non-invasive imaging parameters and patient characteristics were collected and subjected to backward elimination linear regression to define a model relating patient parameters to the total cardiac volume. This regression model was then used to retrospectively accept or reject an ‘ideal’ donor graft from the library for 3 patients that had undergone heart transplantation. Oversized and undersized grafts were also transplanted to qualitatively analyze virtual transplantation specificity.
Results: The backward elimination approach of the data for the 20 patients rejected the factors of BMI, BSA, sex and both end-systolic and end-diastolic left ventricular measurements from echocardiography. Height and weight were included in the linear regression model yielding an adjusted R-squared of 82.5%. Height and weight showed statistical significance with p-values of 0.005 and 0.02 respectively. The final equation for the linear regression model was TCV = -169.320+ 2.874h + 3.578w ± 73 (h=height, w=weight, TCV= total cardiac volume).
Discussion: With the current regression model, height and weight significantly correlate to total cardiac volume. This regression model and virtual normal heart library provide for the possibility of virtual transplant and size-matching for transplantation. The study and regression model is, however, limited due to a small sample size. Additionally, the lack of volumetric resolution from the MR datasets is a potentially limiting factor. Despite these limitations the virtual library has the potential to be a critical tool for clinical care that will continue to grow as normal hearts are added to the virtual library.
Methods: A virtual, 3D library of clinically-defined normal hearts was compiled from reconstructed CT and MR scans. Non-invasive imaging parameters and patient characteristics were collected and subjected to backward elimination linear regression to define a model relating patient parameters to the total cardiac volume. This regression model was then used to retrospectively accept or reject an ‘ideal’ donor graft from the library for 3 patients that had undergone heart transplantation. Oversized and undersized grafts were also transplanted to qualitatively analyze virtual transplantation specificity.
Results: The backward elimination approach of the data for the 20 patients rejected the factors of BMI, BSA, sex and both end-systolic and end-diastolic left ventricular measurements from echocardiography. Height and weight were included in the linear regression model yielding an adjusted R-squared of 82.5%. Height and weight showed statistical significance with p-values of 0.005 and 0.02 respectively. The final equation for the linear regression model was TCV = -169.320+ 2.874h + 3.578w ± 73 (h=height, w=weight, TCV= total cardiac volume).
Discussion: With the current regression model, height and weight significantly correlate to total cardiac volume. This regression model and virtual normal heart library provide for the possibility of virtual transplant and size-matching for transplantation. The study and regression model is, however, limited due to a small sample size. Additionally, the lack of volumetric resolution from the MR datasets is a potentially limiting factor. Despite these limitations the virtual library has the potential to be a critical tool for clinical care that will continue to grow as normal hearts are added to the virtual library.
ContributorsSajadi, Susan (Co-author) / Lindquist, Jacob (Co-author) / Frakes, David (Thesis director) / Ryan, Justin (Committee member) / Harrington Bioengineering Program (Contributor) / School of International Letters and Cultures (Contributor) / Barrett, The Honors College (Contributor)
Created2016-05
Description
Glioblastoma multiforme (GBM) is a malignant, aggressive and infiltrative cancer of the central nervous system with a median survival of 14.6 months with standard care. Diagnosis of GBM is made using medical imaging such as magnetic resonance imaging (MRI) or computed tomography (CT). Treatment is informed by medical images and includes chemotherapy, radiation therapy, and surgical removal if the tumor is surgically accessible. Treatment seldom results in a significant increase in longevity, partly due to the lack of precise information regarding tumor size and location. This lack of information arises from the physical limitations of MR and CT imaging coupled with the diffusive nature of glioblastoma tumors. GBM tumor cells can migrate far beyond the visible boundaries of the tumor and will result in a recurring tumor if not killed or removed. Since medical images are the only readily available information about the tumor, we aim to improve mathematical models of tumor growth to better estimate the missing information. Particularly, we investigate the effect of random variation in tumor cell behavior (anisotropy) using stochastic parameterizations of an established proliferation-diffusion model of tumor growth. To evaluate the performance of our mathematical model, we use MR images from an animal model consisting of Murine GL261 tumors implanted in immunocompetent mice, which provides consistency in tumor initiation and location, immune response, genetic variation, and treatment. Compared to non-stochastic simulations, stochastic simulations showed improved volume accuracy when proliferation variability was high, but diffusion variability was found to only marginally affect tumor volume estimates. Neither proliferation nor diffusion variability significantly affected the spatial distribution accuracy of the simulations. While certain cases of stochastic parameterizations improved volume accuracy, they failed to significantly improve simulation accuracy overall. Both the non-stochastic and stochastic simulations failed to achieve over 75% spatial distribution accuracy, suggesting that the underlying structure of the model fails to capture one or more biological processes that affect tumor growth. Two biological features that are candidates for further investigation are angiogenesis and anisotropy resulting from differences between white and gray matter. Time-dependent proliferation and diffusion terms could be introduced to model angiogenesis, and diffusion weighed imaging (DTI) could be used to differentiate between white and gray matter, which might allow for improved estimates brain anisotropy.
ContributorsAnderies, Barrett James (Author) / Kostelich, Eric (Thesis director) / Kuang, Yang (Committee member) / Stepien, Tracy (Committee member) / Harrington Bioengineering Program (Contributor) / School of Mathematical and Statistical Sciences (Contributor) / Barrett, The Honors College (Contributor)
Created2016-05
Description
The Honors Thesis involved the use of vertically-aligned, piezoelectric nanowire sensor arrays configured by Dr. Henry A. Sodano and Dr. Aneesh Koka from the University of Florida, in order to acquire acceleration data. Originally, the project was focused on interfacing and calibrating the barium titanate (BaTio3) sensors to measure wall shear stress, a fluid dynamic characteristic. In order to gain an understanding of these novel piezoelectric sensors, the experiments performed by Sodano and Koka were to be investigated, replicated, and results reproduced. After initial trial phases, signals failed to be consistently measured from the sensors and the project's emphasis was re-defined. The outlined goals were 1) to re-design the initial system used for signal acquisition, 2) test the improved signal acquisition system, 3) successfully measure output signals from the BaTiO3 nanowire sensors, and 4) determine the cause for inconsistent signal measurements from the piezoelectric nanawire sensors. Following a detailed review of the previous experimental procedures and the initial signal acquisition system, an improved acquisition system was designed and its expected behavior was tested and verified. Despite the introduction of the improved acquisition system, voltage outputs were unable to be measured as a function of shaker table acceleration. It was impossible to verify the effect of the improved signal acquisition system on the measured BaTiO3 nanowire sensor output. Based on an analysis of data collected using a commercial 3-axis acceleromoeter, it is hypothesized that the BaTiO3 nanowire sensors were broken after the first experimental trial due to an excessively applied force from an external source (i.e. shaker table, improper handling during experimentation, and/or improper handling during transportation).
ContributorsThomas, Jonah (Author) / Frakes, David (Thesis director) / LaBelle, Jeffrey (Contributor) / Barrett, The Honors College (Contributor)
Created2014-05
Description
In this study, the specific goal was to evaluate the effectiveness of utilizing a novel virtual reality software package with a haptic device to practice spine surgery. This spine surgery simulator was commissioned by Barrow Neurological Institute (BNI) and is as yet untested. To test the simulator, an experiment was run in which resident neurosurgeons at Barrow Neurological Institute were asked to perform two “virtual surgeries” with the spine surgical simulator, provide observations on the simulator, and then complete a questionnaire evaluating different aspects of the simulator. The mean questionnaire score across all the neurosurgical residents was found to be 65.5 % ± 9.4 % of the maximum score which suggests that certain aspects of the virtual spine surgical simulator were deemed to be effective by the resident neurosurgeons but that improvements need to be made for the simulator to be fully ready as a teaching and planning tool. As of right now, the simulator is more suited as a training tool instead of a planning tool. Improvements that should be implemented include changing the hardware placement of the haptic device and the computer, minimizing aberrant tactile feedback, and adding anatomical and planning detail to the software to provide a more accurate reflection of spine surgery. It was also suggested that future experiments that evaluate an improved simulator should ensure that participants are trained adequately and have enough time to complete surgical operations to get a fair assessment of the tool.
ContributorsIyer, Sudarshan Rajan (Author) / Frakes, David (Thesis director) / Crawford, Neil (Committee member) / Barrett, The Honors College (Contributor) / Harrington Bioengineering Program (Contributor)
Created2015-05
Description
A specific type of Congenital Heart Defect (CHD) known as Coarctation (narrowing) of the Aorta (CoA) prevails in 10% of all CHD patients resulting in life-threatening conditions. Treatments involve limited medical therapy (i.e PGE1 therapy), but in majority of CoA cases, planned surgical treatments are very common. The surgical approach is dictated by the severity of the coarctation, by which the method of treatments is divided between minimally invasive and extensive invasive procedures. Modern diagnostic procedures allude to many disadvantages making it difficult for clinical practices to properly deliver an optimal form of care. Computational Fluid Dynamics (CFD) technique addresses these issues by providing new forms of diagnostic measures that is non-invasive, inexpensive, and more accurate compared to other evaluative devices. To explore further using the CFD based alternative diagnostic measure, this project aims to validate CFD techniques through in vitro studies that capture the fluid flow in anatomically accurate aortic structures. These studies combine particle image velocimetry and catheterization experimental techniques in order to provide a significant knowledge towards validation of fluid flow simulations.
ContributorsPathangey, Girish (Co-author) / Matheny, Chris (Co-author) / Frakes, David (Thesis director) / Pophal, Stephen (Committee member) / Barrett, The Honors College (Contributor) / Harrington Bioengineering Program (Contributor)
Created2015-05