Evaluating the clinical relevance of force-correlated ultrasound

In this paper, we propose a new modality for automated diagnostic assessment of tissues in the context of Duchenne Muscular Dystrophy (DMD). In this force-correlated ultrasound imaging method, we first perform an automated extraction of a multitude of ultrasound images captured across a range of contact forces - a force video, or a force sweep. These images are then coupled with a mechanism to enhance the diagnostic fidelity of the image with regard to DMD. For this, we propose a variance map, which we compute as the pixel-wise standard deviation image for a multiscale stack generated using each image. Using a biomarker quantification scheme of mean gray scale level (GSL) on the enhanced fidelity force-correlated ultrasound images, a k-means clustering is then performed to discriminate the DMD subjects from the control subjects. We present our results on the use of this novel modality in the diagnostic assessment of DMD on data gathered from a clinical study with our system.