A core problem in computer vision and graphics applications is recovering a model of the attributes that create an image. This model is comprised of the intrinsic scene properties: the collection of shapes, paints and lights which together create an image. Conventional methods for recovering intrinsic scene properties rely on multiple observations of the same scene in order to over-constrain the problem. By comparison, recovering these properties from a single image is vastly more difficult. To address this challenge, researchers at UC Berkeley have developed algorithms that infer the most likely intrinsic scene properties of a single image. Furthermore, the researchers have developed corresponding software that can render the scene with adjusted properties such as a different viewpoint, paints, lights, and shapes.  (more...)

The improved resolution and amount of detail afforded by emerging electron microscopy techniques, such as serial block-face scanning electron microscopy (SBFSEM) enable researchers to explore previously unaddressed scientific questions. SBFSEM technique can reveal cell boundaries, e.g. sites of synapses, and intracellular components, such as synaptic vesicles and mitochondria. However, segmentation of the images generated by SBSFEM requires a trained expert to use automated algorithms or manually going through each slice to trace contours around the region of interest, thereby making it a time consuming and labor intensive effort. (more...)

Radiation therapy for solid tumors is traditionally delivered with the radiation source revolving around the patient superior/inferior (head-to-toe) axis. This radiation source trajectory is referred to as coplanar geometry.  Although adding beams from non-coplanar trajectories can improve the dosimetry and reduce normal organ doses from radiotherapy, it is not easily achievable due to the difficulties in plan optimization, collision avoidance, and challenges in creating an efficient beam path so a non-coplanar plan can be delivered within the time allowed by the clinical flow. Therefore, most radiation oncology departments have adopted a policy minimizing non­coplanar beam angles that are collision prone. To realize the benefits of using non-coplanar trajectories, significant pre-planning collision modeling is necessary.  Currently available tools to model non-coplanar radiation trajectories do not account for the individual shapes and sizes of patients.  In addition, use of non-coplanar beams requires complex choreography between the couch (patient platform) and the gantry (rotating radiation deliver source).  Thus, then a large number of non-coplanar beams are needed, manual navigation is inefficient and ultimately impractical.  These limitations indicate a need for improved modeling and delivery systems for non-coplanar radiation beams. (more...)

Small animal imaging modalities, such as micro-CT, micro-PET, and micro-MR, are frequently used in preclinical studies. The laboratory mouse is the most widely used small animal model for cancer, immunology, neurodegenerative, and metabolic disease studies. In order to extract anatomical information from mouse images, it is necessary to perform organ segmentation from the 3D images. Human operator-based processing of 3D images is tedious and subject to bias. Therefore, it is desirable to develop a computerized approach to accomplish this task. A promising solution involves registration of a digital mouse atlas to an acquired image. Organ labeling by the atlas can define organ regions in the mouse image. With this strategy, an atlas can give a more accurate, more reliable, and easier estimation of organ region of a preclinical mouse subject. (more...)

Limited depth of field of a conventional imaging system can be a significant problem in many application areas, such as brightfield microscopy and macro photography. If the surface profile of an object is beyond the focal range, in a single shot only part of the surface can be sharp, while the rest areas are blurred. The spatially varying point-spread-function (PSF) is directly related with the depth map (object surface topography), the focal distance, and other camera settings.  To extend the depth of field, a common method is to take a series of multi-focus images by gradually moving the focal plane, so that different parts of the object can be in focus in different images.  Then a post process called focus stacking is carried out to generate an all-in-focus image by detecting and fusing the in-focus regions together from the observed multi-focus images.   Generally such fusion algorithms take a similar scheme: (1) a local sharpness metric is defined and measured in every pixels within the image sequence; (2) at each position an image index is detected according to the highest sharpness measure, and an index map is then formed; (3) an all-in-focus image is generated based on the index map.  However, most local sharpness metrics can be easily affected by image content and noise, producing an inaccurate index map that may affect the output image quality.  For example, discontinuity of the index map frequently happens in the low signal-to-noise-ratio (SNR) regions (e.g. flat areas), and generates artifacts in the output.   Smoothing the sharpness measure can somehow alleviate this problem, but also reduce the detection accuracy in the boundary regions leading to extra blur. (more...)

Magnetoencephalography (MEG) is a functional imaging modality that directly detects neuronal activity with a millisecond temporal resolution. UC San Diego researchers previously developed a multi-core beamformer (MCBF, see SD2010-340) approach that reconstructs common-mode source time-courses and their correlations networks from noisy MEG data, without requiring both a priori information and expensive and impractical computation. However, the performance of MCBF degrades at low correlations and cannot reconstruct individual source time-courses.A detailed description for the related technology SD2010-340 can be found at http://invent.ucsd.edu/technology/cases/2010/SD2010-340.shtml. (more...)

A novel photography method that provides, along with the actual picture taken, several alternative versions of the image. This method produces photo variations such as panoramas, collages, alternative views, composites based on moving subjects, and views using varying capture parameters such as focal length and exposure time. (more...)

While teleimmersion has great potential, available algorithms for real-time stereo reconstruction require several seconds to several minutes to process two images and produce accurate stereo output.  Available FPGA and GPU implementations have inherent drawbacks in ability to reconstruct homogenous regions or regions with repeating patterns. The time-of-flight cameras have low resolution, limited range, high noise, and albedo sensitivity. Therefore,  a practical real-time stereo reconstruction is needed for a system enabling geographically distributed users to interact with each other in a shared virtual space.   University of California investigators have responded to this challenge by developing INTI Multiview, a real-time stereo reconstitution integration for 3D telleimmersion.  With INTI Multiview, each user is presented by their 3D avatar generated in real time. INTI Multiview focuses primarily on integrating multiple stereo reconstructions from different views.  Levels of accuracy comparable to other methods are achieved at a much faster speed on CPU by taking a hybrid approach: performing a local optimization technique (the region matching) and using a global optimization approximation to improve the initial results (anisotropic diffusion).  The investigators have implemented a novel multiscale representation that allows for the highly accurate reconstruction of a scene.  The investigators have successfully tested INTI Multiview in many application areas, such as remote dance choreography, shared geoscientific and archeological applications, and training. This work has further extensions in other applications where real-time stereo data is necessary, e.g. full body motion capture, surveillance and tracking, foreground/background segmentation, autonomous vehicle control.  Markerless 3D reconstruction for human movement analysis (motion capture, visual feedback for gaming, rehabilitation etc.)   (more...)

Tiled multi-projector displays are becoming increasingly more popular for visualization, education, entertainment, training and simulation applications but registering multiple projectors on such a display presents many challenges. Existing solutions are complex, expensive and often require a highly skilled technician to operate. (more...)

Ultrasound image-guidance has become the standard of care for needle-based procedures. The safety and success of image-guided interventions is dependent on anatomic knowledge, visualization, and precise tracking and control of the biopsy needle. Optimal visualization of the structure of interest and biopsy needle requires the use of specialized biopsy needles and high-fidelity ultrasound units, but the vast majority of medical care providers lack the access to high-fidelity ultrasound machines and specialized biopsy needles necessary for accurate images. Instead, the majority of healthcare providers use less expensive, lower resolution hand-carried ultrasound (HCUS) units. Unfortunately, the inability to clearly see the tip of a needle in relation to the object of interest on the lower-fidelity HCUS units limits the utility of such image-guidance. Numerous invasive procedures are performed on a daily basis without the benefit of accurate image-guidance resulting from poor needle tracking. Such procedures include, but are not limited to, arterial cannulation, abscess drainage, mass biopsies, paracentesis, thoracentesis, and nerve blocks. (more...)

UCSD researchers have developed a hybrid method for analyzing diffusion data collected with magnetic resonance imaging (MRI). By using diffusion-weighted gradient pulses in conjunction with a hyperecho pulse sequence, researchers have been able to enhance further the sensitivity of magnetic resonance imaging to diffusion . The increase in sensitivity over existing methods (spin-echo diffusion-weighted imaging [DWI] and stimulated echo DWI) translates into higher quality images with better resolution, or can be used to reduce the time required for a high-quality scan. (more...)

Radiation oncology physicians need to review and approve the set-up of images prior to patient treatment as a second check for the set-up. Secondly, but also important, the radiation oncology department must show that the set-up images have been reviewed and approved by a physician in order to be able to bill for those services. (more...)

BACKGROUND:  Fluorescence microscopy of live cells allows researchers to study molecular and cellular processes in their natural context. However, fine details may be lost due to degradation of image resolution by various factors, including refractive index, mismatches between the optics and the sample media (e.g. oil/water interfaces or air/solid interfaces), or changes in refractive index due to inhomogeneity of the sample (e.g. different cellular compartments). Resolution, therefore, becomes lower as sample depth increases. Software and/or hardware improvements are needed to retrieve the resolution of three dimensional images collected by microscopy.   TECHNOLOGY:  Researchers have improved the resolution of wide field microscopy through the application of adaptive optics, which allows real time correction of aberrations as has previously been used in astronomy and confocal microscopy. Wide field microscopy (i.e. illumination and imaging of the entire field of view) is most efficient at collecting photons of light compared to other methods such as confocal microscopy, and allows fastest acquisition rates and minimal photo-damage for dynamic studies of live samples. The implementation of adaptive optics, which corrects depth-dependent and sample inhomogeneity-induced aberrations, greatly improves the resolution in live cell images. Both hardware and software methods have been designed to improve the resolution of live sample images. Additionally, novel and cost-effective methodology and apparatus have been developed to correct optical aberrations. (more...)