Dictionary Representation for Liver Ablation Visualization

Robert M. Pohlman, M.S.

Working closely with collaborators in the Radiology Department at the University of Wisconsin School of Medicine and Public Health, we have shown that ultrasound based elasticity imaging provides an alternative, non-ionizing, real-time imaging modality that may be used for tumor visualization of liver microwave ablation (MWA) procedures. Using the microwave antennae for introducing quasi-static deformations required for elastography, coined electrode displacement elastography (EDE), ultrasound strain imaging has demonstrated high contrast and contrast-to-noise ratios in phantoms, animal models, and human patients.

Current work in our lab involves improving tumor visualization of EDE by analyzing displacement estimation approaches, developing modulus reconstruction techniques, and incorporating state-of-the-art machine learning and computer vision algorithms. An example of these improvements is shown in the image above with a B-mode image in the top left, displacement estimation in the bottom left, displacement estimation after adaptive filtering using dictionary learning in the bottom right, and top right strain tensor image from adaptively filtered displacement. We hope to transform EDE into a comparable imaging modality as contrast-enhanced computed tomography (CECT), the current clinical gold standard used for validation of MWA efficacy in the clinical setting.

For additional details please visit: https://www.medphysics.wisc.edu/tvarghese/