Optimizing Transducer Acquisition Scheme for Rapid Ultrasound Tomography of Limbs

Tank based ultrasound tomography systems enable the reconstruction of compressional sound speed and density maps in hard and soft tissue such as bone and muscle. Compressional sound speed can serve as a biomarker and help to identify healthy from unhealthy tissue. Tomography systems must balance trade-offs between cost, complexity, acquisition time, and the number of transducers. In order to minimize patient motion artifacts, acquisition times should be short. In order to minimize cost, fewer transducers elements are preferred. A tomography system consisting of two single element transducers that are mechanically rotated took roughly 40 minutes to acquire a single tomographic slice. We present an optimized acquisition framework that minimizes the uncertainty of the reconstructed compressional sound speed map. Using an initial estimate of the sound speed model, we predict the ray-paths for all possible source-receiver pairs. Based on the ray-paths, we determine transducer locations that minimize the uncertainty of estimated the sound speed map. Using the proposed framework we reconstruct sound speed maps of hard and soft tissue from 25% of the measurements used for a uniformly sampled acquisition strategy in which the receiver positions are equally spaced and data is recorded at all possible back aperture locations. We demonstrate how the acquisition scheme changes when minimizing soft tissue sound speed error versus hard tissue sound speed error.