Geometric distortion of area in medical ultrasound images

Abstract

Medical ultrasound scanners are typically calibrated to a speed of sound corresponding to the soft tissue average of 1540 m s-1. In regions of different sound speed, for example, organs and tumours, the B-mode image becomes geometrically distorted from the true tissue cross-section, due to refraction and the misrepresentation of length. A ray model is developed to predict this distortion for a generalized two-dimensional object with atypical speed of sound, and verified against ultrasound images of a test object. We quantify the areal image distortion as a function of the key dependencies, including the speed of sound mismatch, the scanning format, the object size and its elongation. Our findings show that the distortion of area can be significant, even for relatively small speed of sound mismatches. For example, a 5% speed mismatch typically leads to a 10 - 20% distortion in area. These findings have implications for the accuracy of ultrasound-based evaluation of area and volume.