Computer Vision based Ultrasound Scan Analysis and Feature Detection

Designed a robust algorithm to enhance the detection of musculoskeletal features in ultrasound scans of the lower limb.

Posted by Han Runbing on Mon, Sep 19, 2022

Data Collection and Experiments

Soft Tissue Artefacts (STA) are a major block for estimating biomechanics and capturing skin marker-based motion. In our study, we developed a technique to determine the actual motion of underlying bone with the analysis of ultrasound (US) bone image and to correct for soft tissue artifacts.

Our method was optimized by experimental data and evaluated by different approaches. Bone region detection block achieved 0.83 IoU. Bone outline segmentation block resulted in 80% good segmentations.