Tissue Characterization of Equine Tendons With Clinical B-Scan Images Using a Shock Filter Thinning Algorithm

Abstract—The fiber bundle density (FBD) calculated from ultra-sound B-scan images of the equine superficial digital flexor tendon (SDFT) can serve as an objective measurement to characterize the three metacarpal sites of normal SDFTs, and also to discriminate a healthy SDFT from an injured one. In this paper, we propose a shock filter algorithm for the thinning of hyper-echoic structures observed in B-scan images of the SDFT. This algorithm is further enhanced by applying closing morphological operations on filtered images to facilitate extraction and quantification of fiber bundle fas-cicles. The mean FBD values were calculated from a clinical B-scan image dataset of eight normal and five injured SDFTs. The FBD values measured at three different tendon sites in normal cases show a highest density on the proximal site (five cases out of eight) and a lowest value on the distal part (seven cases out of eight). The mean FBD values measured on the entire tendon from the whole B-scan image dataset show a significant difference between normal and injured SDFTs: 51 for the normal SDFTs and 39 for the injured ones . This difference likely indicates dis-ruption of some fiber fascicle bundles where lesions occurred. To conclude, the potential of this imaging technique is shown to be effi-cient for anatomical structural SDFT characterizations, and opens the way to clinically identifying the integrity of SDFTs. Index Terms—B-scan images, equine tendon, fiber bundle density, mathematical morphological operations, shock filter, thinning algorithm, two-dimensional/three-dimensional (2D/3D) anatomical structures, ultrasound imaging. I