SHB:Type 1(EXP): High Accuracy Motion Analysis using Commodity Depth Camera for Clinical Lower Extremity Assessment

Grants and Contracts Details

Description

Physical therapy has been widely adopted to treat a wide range of medical conditions, from disabilities, injuries, to crippling diseases such as arthritis, lower back pain, or cerebral palsy. A central issue in physical therapy is the assessment of a patient's kinematic performance. With technology advancement, physical therapists can now utilize commercial motion capture (mocap) system to quantify the 3D mechanics common to many conditions they treat. The 3D kinematic feedback leads to more accurate diagnosis and faster progression. Unfortunately the current state of the art for technology to measure 3D motion is expensive, cumbersome, and not widely available to physical therapists. Thus a strong need remains for the development of a portable, low cost, mocap system to give physical therapists precise 3D measurements of movement dysfunction. We propose to use the depth information from a single depth camera (SDC) to capture high-accuracy motion data comparable to that from a commercial mocap system. The use of depth information removes the position ambiguity in a 2D video and makes it possible to capture object's motion from a single viewpoint, eliminating the need for a camera array and its associated acquisition and maintenance cost. A single depth image contains thousands of points, compared to the tens of points in a marker-based system, the large number of points provides the robustness and redundancy for tracking, leading to possible solutions for precise estimation of joint centers and bone movement, or even ameliorating the effect of clothing and tissue deformation. While depth-camera-based pose tracking solutions exist, such as the Kinect system from Microsoft, medical assessment demands almost an order-high accuracy than that for digital entertainment. Based on prior medical studies, our goal is to achieve a positional error of less than 10mm and angular error less than 1 degree. That is four to 10 times better than the current state of the art in SDC mocap.
StatusFinished
Effective start/end date10/1/129/30/16

Funding

  • National Science Foundation: $577,303.00

Fingerprint

Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.