Emission Optical Projection Tomography (eO PT) is a relatively new imag- ing modality that bridges a gap between micro Magnetic Resonance Imag- ing and Confocal Laser Scanning Microscopy. eO PT can be used to image the anatomy and gene expression of intact biological specimens at high resolution and thus provides an alternative to time consuming methods such as serial sectioning. Tomographic image reconstruction for eOPT is currently performed using the Filtered Back Projection algorithm which, while being fast, does not account for the physics of image formation and thus can result in reconstructions of reduced resolution and questionable quantitative consistency. This thesis describes work that was done on eOPT in three areas, including image formation, tomographic reconstruction, and memory savings, the latter of which were required to bring implementation of 3D iterative reconstruction algorithms within reach for the relatively high-resolution eO PT imaging modality. In the area of image formation, measurements were taken to reveal the effects of optical blurring, diffraction and charge-coupled device (CCD) camera noise. Accurate models of each of these phenomena were developed and compared against the measurements. The subject of image reconstruction was first addressed with a modi- fication to the FBP algorithm designed to correct for the quantitative inaccuracies suspected of being introduced by the FBP algorithm when reconstructing specimens consisting of very fine detail. This was done by incorporating the quantitative aspects of the model of image formation into the FBP algorithm. The full model of image formation was incorpo- rated into the iterative Maximum Likelihood Expectation Maximisation (MLEM) algorithm. The third strand of this thesis focuses on various memory saving meth- ods developed to enable the implementation and testing of a variation of MLEM known as the Ordered Subsets Expectation Maximisation (OSEM). , Without such memory saving methods, the implementation of an iterative 3D reconstruction algorithm such as MLEM or OSEM using a full model of image formation would have remained beyond the capacity of modern computers for the foreseeable future, requiring several Terabytes of RAM. Comparisons were made between the quality of and the time required to produce FBP and OSEM reconstructions of the same data sets given the availability of limited computing resources. The feasibility of adopting OSEM reconstructions as an alternative to FBP reconstructions was dis- cussed, based on the use of currently available cutting edge computing hardware.

Although significant progress has been made in imaging devices during the past few decades, the photographs acquired by digital cameras are still far from perfection due to the physical limitations of hardware such as aperture, lens and sensor. This fact brings out the demand for study on image enhancement: a computational technique that aims to improve the interpretability or perception of information in photographs for human viewers. The work in this thesis mainly focuses on three tasks in image enhancement. / Finally, since the radiance of the real world spans several orders of magnitude and its dynamic range dramatically exceeds the capability of the current digital cameras, there often exist some undesirable over- or under-exposed regions in a photograph. The third part of this thesis aims at producing one great looking well-exposed image that is virtually impossible with a single exposure by compositing a stack of photos at different exposures taken with a conventional camera. Particularly, a simple but effective method is presented to describe how to take advantage of the gradient information to accomplish exposure composition in both static and dynamic scenes. Compared to conventional high dynamic range (HDR) imaging work, the proposed approach is quite appealing in practice since it is computationally efficient and easy to use, and frees users from the tedious radiometric calibration and tone mapping steps. / Firstly, since the camera sensor has limited resolution, the acquired images cannot capture the scene very detailedly. Hence, people often resort to a postprocessing technique called super-resolution (SR) to enhance the resolution of the captured images. In the first part of this thesis, two approaches are presented to address the challenging single image SR problem, which is to recover a high-resolution (HR) image from one low-resolution (LR) input. Specifically, a novel learning-based framework is designed specifically for face image SR task from the perspective of DCT domain. In addition, an efficient two-step scheme is developed to super-resolve generic image by exploiting the salient edges of the input LR image. / Secondly, due to the limitation of lens and aperture, some cameras cannot produce pleasant photographs with desired focus setting. For example, portrait photography that requires shallow depth of field (DOF) is not allowed when using the compact point-and-shoot cameras. In the second part of this thesis, a new and complete postprocessing-based focus editing system that is able to handle the tasks of focus map estimation, image refocusing and defocusing, is developed to overcome the optical limitations and create different kinds of novel photos with desired focus setting from an imperfect photo. / Throughout this work, extensive experiments on various real and synthetic image data are conducted to evaluate the performance of the proposed algorithms. / Zhang, Wei. / Adviser: Wai-Kuen Chan. / Source: Dissertation Abstracts International, Volume: 73-03, Section: B, page: . / Thesis (Ph.D.)--Chinese University of Hong Kong, 2010. / Includes bibliographical references (leaves 116-125). / Electronic reproduction. Hong Kong : Chinese University of Hong Kong, [2012] System requirements: Adobe Acrobat Reader. Available via World Wide Web. / Electronic reproduction. [Ann Arbor, MI] : ProQuest Information and Learning, [201-] System requirements: Adobe Acrobat Reader. Available via World Wide Web. / Abstract also in Chinese.

No description available.

In this thesis, a block-based motion estimation algorithm suitable for Super-Resolution (SR) image reconstruction is introduced. The motion estimation problem is formulated as an energy minimization problem that consists of both a data and regularization term. To handle cases when motion estimation fails, a block-based validity method is introduced, and is shown to outperform all other validity methods in the literature in terms of hybrid de-interlacing. By combining the validity metric into the energy minimization framework, it is shown that 1) the motion vector error is made less sensitive to block size, 2) a more uniform distribution of motion-compensated blocks results, and 3) the overall motion vector error is reduced. The final motion estimation algorithm is shown to outperform several state-of-the-art motion estimation algorithms in terms of both endpoint error and interpolation error, and is one of the fastest algorithms in the Middlebury benchmark. With the new motion estimation algorithm and validity metric, it is shown that artifacts are virtually eliminated from the POCS-based reconstruction of the high-resolution image.

Despite being the most preventable cancer, cervical cancer remains the third leading cause of cancer death worldwide. Over 85% of cervical cancer incidence and mortality occurs in low-resource countries where screening programs for early detection are either inadequate or unavailable. In the developed world, where screening programs are well organized, incidence and mortality rates are greatly reduced. Recent advances in optical imaging have the potential to enable cervical cancer screening at the point-of-care, even in the hands of less experienced providers. High performance optical imaging systems can be constructed at relatively low cost, and image analysis can be automated; thus, these technologies may provide a way to bridge the gap to cervical cancer screening for developing countries. This work focuses on the design, construction, and clinical testing of a novel multimodal optical imaging (combination of wide-field imaging and high-resolution) for early detection of cervical neoplasia. The Multimodal Digital Imager (MDI) acquires in vivo images of cervical tissue in fluorescence, narrow band reflectance, and orthogonal polarized reflectance modes using multiple illumination wavelengths. The High Resolution Microendoscope (HRME) was used to interrogate clinically suspicious areas with subcellular spatial resolution, revealing changes in nuclear to cytoplasmic area ratio. In vivo image data from the wide-field system was combined with image data from a high- resolution microendoscope (HRME) in order to test the effectiveness of the multimodal optical imaging in discriminating between cervical neoplasia and non-neoplastic. Multimodal optical imaging coupled with computer aided diagnostic achieved a sensitivity of 82% and specificity of 85% for discriminating cervical neoplastic from non-neoplastic This work has demonstrated that multimodal optical imaging; combination of wide-field and high-resolution optical imaging of the cervix can assist in the detection of cervical neoplasia and can be implemented effectively in a low-resource setting.

The main objective of this research was to develop a two-stage optical imaging platform to improve detection of cancer in Barrett’s esophagus (BE). BE caused by chronic reflux and patients with BE are at a higher risk for developing esophageal adenocarcinoma (EAC). However, neoplasia in BE is often unidentifiable under standard endoscopy, and studies have shown nearly half of early cancers can go unidentified by this method. Widefield imaging (resolves ~100 microns) allows efficient surveillance of large BE segments. Two widefield imaging techniques were identified to improve contrast between benign and abnormal lesions during an ex vivo 15 patient feasibility study. Cross-polarized imaging (CPI) reduced specular reflection and improved vascular contrast. Vital-dye fluorescence imaging (VFI) using topically-applied proflavine improved visualization of glandular pattern. Moreover, relevant pathologic features visible during VFI were seen in corresponding histology slides as well as high resolution images of the same sites. Based on these results, a cap-based Multispectral Digital Endoscope (MDE) was designed and built. The MDE can image in three different imaging modes: white light imaging, CPI, and VFI. Modifications to a Pentax EPK-i video processor and a Pentax endoscope were made to incorporate these imaging modes into one system. A 21 patient in vivo pilot study with 65 pathologically correlated sites demonstrated the feasibility of using this system in vivo; image criteria were developed to classify neoplasia with a sensitivity and specificity of 100% and 76% respectively. High resolution imaging (resolves ~2-5 micron) may verify the disease presence in suspicious areas identified using widefield techniques. 2-NBDG, a fluorescent metabolic marker, was used as to identify neoplastic biopsies. In a study with 21 patients yielding 38 pathologically correlated biopsies and 158 image sites, 2-NBDG imaging allowed classification of cancerous biopsies with a sensitivity of 96% and specificity of 90%. The unique contributions of these results is the development of a multimodal cap-based endoscopic system to identify suspicious areas in BE, and using a metabolic marker to verify the presence of disease. This application extends beyond esophageal cancer detection and may be explored for cancer detection in other organ sites characterized by columnar epithelium.

High resolution imaging of wide fields has been a prerogative of space telescopes for decades. Multi-conjugate adaptive optics (MCAO) is a key technology for the future of ground-based astronomy, especially as we approach the era of ELTs, where the large apertures will provide diffraction limits that will significantly surpass even the James Webb Space Telescope. NFIRAOS will be the first light MCAO system for the Thirty Meter Telescope and to support its development I have worked on HeNOS, its test bench integrated in Victoria at NRC Herzberg. I have aligned the optics, tested the electronic hardware, calibrated the subsystems (cameras, deformable mirrors, light sources, etc.) and characterized the system parameters. Development and support for future MCAO instruments also involves data analysis, a critical process in delivering the expected performance of any scientific instrument. To develop a strategy for optimal stellar photometry with MCAO, I have observed the Galactic globular cluster NGC 1851 with GeMS, the MCAO system on the 8-meter Gemini South telescope. From near-infrared images of this target in two bands, I have found the optimal parameters to employ in the profile-fitting photometry and calibration. As testimony to the precision of the results, I have obtained the deepest near-infrared photometry of a crowded field from the ground and used it to determine the age of the cluster with a method recently proposed that exploits the bend in the lower main sequence. The precise color-magnitude diagram also allows us to clearly observe the double subgiant branch for the first time from the ground, caused by the multiple stellar populations in the cluster. As the only facility MCAO system, GeMS is an important instrument that serves to illuminate the challenges of obtaining accurate photometry using such a system. By coupling the knowledge acquired from an instrument already on-sky with experiments in the lab on a prototype of a future system, I have addressed new challenges in photometry and astrometry, like the promising technique of point spread function reconstruction. This thesis informs the development of appropriate data processing techniques and observing strategies to ensure the ELTs deliver their full scientific promise over extended fields of view. / Graduate

Super-resolution 3D imaging was achieved using newly synthesized photoactivatable quantum dot (PAQ dot) probes. Quantum dots were modified with a novel quencher system to make them photoactivatable. The unique properties of these PAQ dots enable single-fluorophore localization in three dimensions using a confocal microscopy optical sectioning method. Myosin and tropomyosin of rabbit myofibrilar bundles were specifically labeled with the newly synthesized PAQ dot. A sufficient number of single quantum dots were photoactivated, localized and reduced to their centroid and then reconstructed to a super-resolution image. The acquired super-resolution image shows a lateral and an axial sub-diffraction resolution and demonstrates ultrafine striations with widths less than 70 nm that are not evident by conventional confocal microscopy. The striations appear to be related to nebulin thin filament binding protein. This newly developed imaging system is cutting edge for its high resolution and localization as well its simplicity and convenience.

Olivine is the most abundant mineral in Earth's upper mantle and is one of the major minerals discovered in extraterrestrial objects. Its physical properties govern the dynamics of the upper mantle. The most dynamic regions of the upper mantle are sites where melting and melt segregation occurs. These regions are also the most variable in terms of their oxygen fugacity. We therefore conducted piston cylinder experiments to determine the intergranular melt distribution, and explore a range of oxygen buffers. We annealed olivine aggregates in metallic and graphite capsules to determine the oxygen fugacities set by the capsule materials. These experiments show that oxygen fugacities are below their corresponding metal-oxide buffers. The oxygen fugacity in nickel80-iron20 and graphite capsules most closely represents the intrinsic oxygen fugacity of Fo90 olivine, while iron capsules are too reducing perhaps explaining the formation of "dusty" olivine in chondrites. We annealed olivine-basalt aggregates in order to determine the melt distribution. The results show that the length of olivine grain boundaries wetted by melt (grain boundary wetness) increases with increasing melt content to values well above those predicted by a simplified model which is commonly applied to this system. At fixed melt content the grain boundary wetness increases with increasing grain size. These observations emphasize that the dihedral angle of the simplified system is not adequate to characterize the melt distribution in partially molten rocks. Our observations indicate that at upper mantle grain sizes the shear viscosity of partially molten rocks is one order of magnitude lower than predicted by the simplified model. Naturally partially molten rocks exist in the form of olivine-rich troctolites or plagioclase dunites, but the conditions for their formation are not entirely clear. We therefore conducted step-cooling experiments that indicate that slow cooling of samples with a steady-state microstructure reproduces the interstitial geometry observed in natural samples. The grain boundary wetness determined from the interstitial phases is somewhat reduced during slow cooling relative to samples quenched from high temperature. The microstructural similarity of experimental and natural samples suggests that mush zones identified beneath mid-ocean ridges may have lower melt contents than previously envisioned. / 2016-11-18T00:00:00Z

京都大学 / 0048 / 新制・課程博士 / 博士(理学) / 甲第17972号 / 理博第3916号 / 新制||理||1565(附属図書館) / 80816 / 京都大学大学院理学研究科物理学・宇宙物理学専攻 / (主査)教授 高橋 義朗, 教授 田中 耕一郎, 教授 石田 憲二 / 学位規則第4条第1項該当 / Doctor of Science / Kyoto University / DGAM