Novel Algorithm for High Speed Medical Image Processing
Inventors: Simon Warfield, Ayelet Akselrod-Ballin
Invention Types: Information Technology/Software
Research Areas: Radiology
Keywords: Instrumentation, SoftwareFor More Information Contact: Herrera Gutierrez, Aida
Medical imaging is an integral part of modern diagnostics and continues to make rapid advances. A number of modalities exist today, such as ultrasound, magnetic resonance imaging, computed tomography and positron emission tomography. The amount and quality of data has increased significantly, which makes a number of standard image analysis techniques extremely computationally intensive, time consuming and difficult to perform. Simon Warfield, PhD, director of the Computational Radiology Laboratory, and his group at Children’s have developed a new high speed image processing algorithm that allows for the exact matching of two or more related images, which is a key requirement in many image analysis methods such as in 3D reconstruction of MRI, CT or ultrasound data, or processing of movie sequences.
Specifically, a new efficient search strategy to substantially accelerate feature based registration is introduced. Previous feature based registration algorithms often use truncated search strategies in order to achieve small computation times. This new accelerated search strategy is based on the realization that the search for corresponding features can be dramatically accelerated by utilizing Johnson-Lindenstrauss dimension reduction.
This new algorithm will open new possibilities for many medical imaging applications, in neuroscience, disease modeling, surgical planning, intervention simulation and diagnosis, and may be a key piece in the development of real-time technologies in other fields such as video processing etc.
Order of magnitude calculations for the search strategy proposed here indicate the algorithm proposed is more than a million times faster than previously utilized naïve search strategies, and this advantage in speed is directly translated into an advantage in accuracy as the fast speed enables more comparisons to be made in the same amount of time. The capabilities of the software were demonstrated using large transmission electron microscopy (TEM) images, where the computation time it took to align two images was reduced from several days to a few minutes.