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.

Single cell analysis has become increasingly important in understanding disease onset, progression, treatment and prognosis, especially when applied to cancer where cellular responses are highly heterogeneous. Through the advent of single cell computerized tomography (Cell-CT), researchers and clinicians now have the ability to obtain high resolution three-dimensional (3D) reconstructions of single cells. Yet to date, no live-cell compatible version of the technology exists. In this thesis, a microfluidic chip with the ability to rotate live single cells in hydrodynamic microvortices about an axis parallel to the optical focal plane has been demonstrated. The chip utilizes a novel 3D microchamber design arranged beneath a main channel creating flow detachment into the chamber, producing recirculating flow conditions. Single cells are flowed through the main channel, held in the center of the microvortex by an optical trap, and rotated by the forces induced by the recirculating fluid flow. Computational fluid dynamics (CFD) was employed to optimize the geometry of the microchamber. Two methods for the fabrication of the 3D microchamber were devised: anisotropic etching of silicon and backside diffuser photolithography (BDPL). First, the optimization of the silicon etching conditions was demonstrated through design of experiment (DOE). In addition, a non-conventional method of soft-lithography was demonstrated which incorporates the use of two positive molds, one of the main channel and the other of the microchambers, compressed together during replication to produce a single ultra-thin (<200 µm) negative used for device assembly. Second, methods for using thick negative photoresists such as SU-8 with BDPL have been developed which include a new simple and effective method for promoting the adhesion of SU-8 to glass. An assembly method that bonds two individual ultra-thin (<100 µm) replications of the channel and the microfeatures has also been demonstrated. Finally, a pressure driven pumping system with nanoliter per minute flow rate regulation, sub-second response times, and < 3% flow variability has been designed and characterized. The fabrication and assembly of this device is inexpensive and utilizes simple variants of conventional microfluidic fabrication techniques, making it easily accessible to the single cell analysis community.

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.

Recent advances in camera architectures and associated mathematical representations now enable compressive acquisition of images and videos at low data-rates. While most computer vision applications of today are composed of conventional cameras, which collect a large amount redundant data and power hungry embedded systems, which compress the collected data for further processing, compressive cameras offer the advantage of direct acquisition of data in compressed domain and hence readily promise to find applicability in computer vision, particularly in environments hampered by limited communication bandwidths. However, despite the significant progress in theory and methods of compressive sensing, little headway has been made in developing systems for such applications by exploiting the merits of compressive sensing. In such a setting, we consider the problem of activity recognition, which is an important inference problem in many security and surveillance applications. Since all successful activity recognition systems involve detection of human, followed by recognition, a potential fully functioning system motivated by compressive camera would involve the tracking of human, which requires the reconstruction of atleast the initial few frames to detect the human. Once the human is tracked, the recognition part of the system requires only the features to be extracted from the tracked sequences, which can be the reconstructed images or the compressed measurements of such sequences. However, it is desirable in resource constrained environments that these features be extracted from the compressive measurements without reconstruction. Motivated by this, in this thesis, we propose a framework for understanding activities as a non-linear dynamical system, and propose a robust, generalizable feature that can be extracted directly from the compressed measurements without reconstructing the original video frames. The proposed feature is termed recurrence texture and is motivated from recurrence analysis of non-linear dynamical systems. We show that it is possible to obtain discriminative features directly from the compressed stream and show its utility in recognition of activities at very low data rates.

Human operators have difficulty driving cranes quickly, accurately, and safely because of the slow response of heavy crane structures, non-intuitive control interfaces, and payload oscillations. Recently, a novel hand-motion crane control system has been proposed to improve performance by coupling an intuitive control interface with an element that reduces the complex oscillatory behavior of the payload. Hand-motion control allows operators to drive a crane by simply moving a hand-held radio-frequency tag through the desired path. Real-time location sensors are used to track the movements of the tag and the tag position is used in a feedback control loop to drive the crane. An input shaper is added to eliminate dangerous payload oscillations. However, tag position measurements are corrupted by noise. It is important to understand the noise properties so that appropriate filters can be designed to mitigate the effects of noise and improve tracking accuracy. This work discusses implementing filtering techniques to address the issue of noise in the operating environment. Five different filters are used on experimentally-acquired tag trajectories to reduce noise. The filtered trajectories are then used to drive crane simulations. Filter performance is evaluated with respect to the energy usage of the crane trolley, the settling time of the crane payload oscillations, and the safety corridor of the crane trajectory. The effects of filter window lengths on these parameters are also investigated. An adaptive filtering technique, namely the Kalman filter, adapts to the noise characteristics of the workspace to minimize the tag tracking error and performs better than the other filtering techniques examined.

Current treatment methods for cerebral aneurysms are providing life-saving measures for patients suffering from these blood vessel wall protrusions; however, the drawbacks present unfortunate circumstances in the invasive procedure or with efficient occlusion of the aneurysms. With the advancement of medical devices, liquid-to-solid gelling materials that could be delivered endovascularly have gained interest. The development of these systems stems from the need to circumvent surgical methods and the requirement for improved occlusion of aneurysms to prevent recanalization and potential complications. The work presented herein reports on a liquid-to-solid gelling material, which undergoes gelation via dual mechanisms. Using a temperature-responsive polymer, poly(N-isopropylacrylamide) (poly(NIPAAm), the gelling system can transition from a solution at low temperatures to a gel at body temperature (physical gelation). Additionally, by conjugating reactive functional groups onto the polymers, covalent cross-links can be formed via chemical reaction between the two moieties (chemical gelation). The advantage of this gelling system comprises of its water-based properties as well as the ability of the physical and chemical gelation to occur within physiological conditions. By developing the polymer gelling system in a ground-up approach via synthesis, its added benefit is the capability of modifying the properties of the system as needed for particular applications, in this case for embolization of cerebral aneurysms. The studies provided in this doctoral work highlight the synthesis, characterization and testing of these polymer gelling systems for occlusion of aneurysms. Conducted experiments include thermal, mechanical, structural and chemical characterization, as well as analysis of swelling, degradation, kinetics, cytotoxicity, in vitro glass models and in vivo swine study. Data on thermoresponsive poly(NIPAAm) indicated that the phase transition it undertakes comes as a result of the polymer chains associating as temperature is increased. Poly(NIPAAm) was functionalized with thiols and vinyls to provide for added chemical cross-linking. By combining both modes of gelation, physical and chemical, a gel with reduced creep flow and increased strength was developed. Being waterborne, the gels demonstrated excellent biocompatibility and were easily delivered via catheters and injected within aneurysms, without undergoing degradation. The dual gelling polymer systems demonstrated potential in use as embolic agents for cerebral aneurysm embolization.

Magnetic Resonance Imaging (MRI) is limited in speed and resolution by the inherently low Signal to Noise Ratio (SNR) of the underlying signal. Advances in sampling efficiency are required to support future improvements in scan time and resolution. SNR efficiency is improved by sampling data for a larger proportion of total imaging time. This is challenging as these acquisitions are typically subject to artifacts such as blurring and distortions. The current work proposes a set of tools to help with the creation of different types of SNR efficient scans. An SNR efficient pulse sequence providing diffusion imaging data with full brain coverage and minimal distortion is first introduced. The proposed method acquires single-shot, low resolution image slabs which are then combined to reconstruct the full volume. An iterative deblurring algorithm allowing the lengthening of spiral SPoiled GRadient echo (SPGR) acquisition windows in the presence of rapidly varying off-resonance fields is then presented. Finally, an efficient and practical way of collecting 3D reformatted data is proposed. This method constitutes a good tradeoff between 2D and 3D neuroimaging in terms of scan time and data presentation. These schemes increased the SNR efficiency of currently existing methods and constitute key enablers for the development of SNR efficient MRI.

Treatment of cerebral aneurysms using non-invasive methods has existed for decades. Since the advent of modern endovascular techniques, advancements to embolic materials have largely focused on improving platinum coil technology. However, the recent development of Onyx®, a liquid-delivery precipitating polymer system, has opened the door for a new class of embolic materials--liquid-fill systems. These liquid-fill materials have the potential to provide better treatment outcomes than platinum coils. Initial clinical use of Onyx has proven promising, but not without substantial drawbacks, such as co-delivery of angiotoxic compounds and an extremely technical delivery procedure. This work focuses on formulation, characterization and testing of a novel liquid-to-solid gelling polymer system, based on poly(propylene glycol) diacrylate (PPODA) and pentaerythritol tetrakis(3-mercaptopropionate) (QT). The PPODA-QT system bypasses difficulties associated with Onyx embolization, yet still maintains non-invasive liquid delivery--exhibiting the properties of an ideal embolic material for cerebral aneurysm embolization. To allow for material visibility during clinical delivery, an embolic material must be radio-opaque. The PPODA-QT system was formulated with commercially available contrast agents and the gelling kinetics were studied, as a complete understanding of the gelling process is vital for clinical use. These PPODA-QT formulations underwent in vitro characterization of material properties including cytotoxicity, swelling, and degradation behaviors. Formulation and characterization tests led to an optimized PPODA-QT formulation that was used in subsequent in vivo testing. PPODA-QT formulated with the liquid contrast agent ConrayTM was used in the first in vivo studies. These studies employed a swine aneurysm model to assess initial biocompatibility and test different delivery strategies of PPODA-QT. Results showed good biocompatibility and a suitable delivery strategy, providing justification for further in vivo testing. PPODA-QT was then used in a small scale pilot study to gauge long-term effectiveness of the material in a clinically-relevant aneurysm model. Results from the pilot study showed that PPODA-QT has the capability to provide successful, long-term treatment of model aneurysms as well as facilitate aneurysm healing.

Microfluidics is the study of fluid flow at very small scales (micro -- one millionth of a meter) and is prevalent in many areas of science and engineering. Typical applications include lab-on-a-chip devices, microfluidic fuel cells, and DNA separation technologies. Many of these microfluidic devices rely on micron-resolution velocimetry measurements to improve microchannel design and characterize existing devices. Methods such as micro particle imaging velocimetry (microPIV) and micro particle tracking velocimetry (microPTV) are mature and established methods for characterization of steady 2D flow fields. Increasingly complex microdevices require techniques that measure unsteady and/or three dimensional velocity fields. This dissertation presents a method for three-dimensional velocimetry of unsteady microflows based on spinning disk confocal microscopy and depth scanning of a microvolume. High-speed 2D unsteady velocity fields are resolved by acquiring images of particle motion using a high-speed CMOS camera and confocal microscope. The confocal microscope spatially filters out of focus light using a rotating disk of pinholes placed in the imaging path, improving the ability of the system to resolve unsteady microPIV measurements by improving the image and correlation signal to noise ratio. For 3D3C measurements, a piezo-actuated objective positioner quickly scans the depth of the microvolume and collects 2D image slices, which are stacked into 3D images. Super resolution microPIV interrogates these 3D images using microPIV as a predictor field for tracking individual particles with microPTV. The 3D3C diagnostic is demonstrated by measuring a pressure driven flow in a three-dimensional expanding microchannel. The experimental velocimetry data acquired at 30 Hz with instantaneous spatial resolution of 4.5 by 4.5 by 4.5 microns agrees well with a computational model of the flow field. The technique allows for isosurface visualization of time resolved 3D3C particle motion and high spatial resolution velocity measurements without requiring a calibration step or reconstruction algorithms. Several applications are investigated, including 3D quantitative fluorescence imaging of isotachophoresis plugs advecting through a microchannel and the dynamics of reaction induced colloidal crystal deposition.

Single cell phenotypic heterogeneity studies reveal more information about the pathogenesis process than conventional bulk methods. Furthermore, investigation of the individual cellular response mechanism during rapid environmental changes can only be achieved at single cell level. By enabling the study of cellular morphology, a single cell three-dimensional (3D) imaging system can be used to diagnose fatal diseases, such as cancer, at an early stage. One proven method, CellCT, accomplishes 3D imaging by rotating a single cell around a fixed axis. However, some existing cell rotating mechanisms require either intricate microfabrication, and some fail to provide a suitable environment for living cells. This thesis develops a microvorterx chamber that allows living cells to be rotated by hydrodynamic alone while facilitating imaging access. In this thesis work, 1) the new chamber design was developed through numerical simulation. Simulations revealed that in order to form a microvortex in the side chamber, the ratio of the chamber opening to the channel width must be smaller than one. After comparing different chamber designs, the trapezoidal side chamber was selected because it demonstrated controllable circulation and met the imaging requirements. Microvortex properties were not sensitive to the chambers with interface angles ranging from 0.32 to 0.64. A similar trend was observed when chamber heights were larger than chamber opening. 2) Micro-particle image velocimetry was used to characterize microvortices and validate simulation results. Agreement between experimentation and simulation confirmed that numerical simulation was an effective method for chamber design. 3) Finally, cell rotation experiments were performed in the trapezoidal side chamber. The experimental results demonstrated cell rotational rates ranging from 12 to 29 rpm for regular cells. With a volumetric flow rate of 0.5 µL/s, an irregular cell rotated at a mean rate of 97 ± 3 rpm. Rotational rates can be changed by altering inlet flow rates.