Medical Image Processing: Algorithms, Real-Time Computing, and Novel Applications
Prof. Raj Shekhar
|Dr. Raj Shekhar
Image processing of 3D and 4D (space + time) medical data, both in software and in hardware, is the primary area of Dr. Shekhar's research. Image processing is integral to both generation and interpretation of medical images. Image reconstruction algorithms underlie common medical image acquisition techniques such as computed tomography (CT), magnetic resonance (MR) imaging, positron emission tomography (PET), and ultrasound. The interpretation of the generated images relies on filtering, enhancement, segmentation, and registration. For clinical efficiency, it is also imperative that these image processing algorithms execute rapidly, preferably in real time.
A specific research focus of Shekhar's team is image registration, which is the process of aligning two images. It is a means to fuse images of different modalities (e.g., CT, MR imaging, PET, ultrasound) to combine complementary information into a single image. It is also a prerequisite for image-based serial comparison for monitoring disease progression and treatment response. In its simplest form, image registration is rigid, with images aligned in a global and uniform way (e.g., an entire image is rotated and scaled uniformly to match another). When the anatomy changes in shape and size, which is often the case, rigid registration is insufficient. Nonrigid image registration aligns tissues and structures that may have deformed between the two images. Shekhar's research involves developing new nonrigid image registration algorithms.
One of the most general, accurate, reliable, and fully automatic approaches to image registration is based on image similarity. Lengthy execution time, however, is a limitation of this iterative approach. We have developed efficient architectures and implemented them on field-programmable gate arrays to reduce the time of image registration from hours to minutes. The goal of the continuing research is ultimately to reduce the execution time to less than 1 second.
Dr. Shekhar, an ECE affiliate and professor in the Department of Diagnostic Radiology and Nuclear Medicine at the University of Maryland School of Medicine (UMSOM), received a grant from the National Cancer Institute/National Institutes of Health for the development of a low-cost hardware accelerator for 3-D medical image registration.
Medical image registration and its acceleration have been major foci of Dr. Shekhar’s research for many years. This latest small-business technology transfer grant was awarded to the consortium of the University of Maryland, Baltimore and IGI Technologies, led by Dr. Shekhar. The team includes William Plishker, Ph.D., and Eliot Siegel, M.D., Chief of Imaging for the VA Health System Maryland and a professor at UMSOM.
The grant will fund initial work in converting academic research into a commercial product capable of performing registration of typical medical images of any type and of any part of the body in 1 minute or less. This and similar technologies will be applied in the clinical setting to improve the accuracy and efficiency of interventional procedures and provide significant benefits in patient safety and positive outcomes.
A team presenting Dr. Shekhar's technology won the University of Maryland's Business Plan Competition, organized annually by the Maryland Technology Enterprise Institute (Mtech).
Related research projects include reconstruction techniques for low-dose CT, 4D data segmentation, and, working closely with University of Maryland Medical Center physicians, developing high-tech, minimally invasive, image-guided treatments and surgeries.
Further information can be found at http://radweb1.radiology.umm.edu/itl/.
return to spotlight on research