Three-dimensional gesture comparison using curvature analysis of position and orientation

Area of Science:

• Medical Robotics
• Human-Computer Interaction
• Surgical Skill Assessment

Background:

• Comparing human gestures requires robust methods to account for variations in position and orientation.
• Surgical skill analysis, particularly in procedures like obstetrics, benefits from objective quantitative assessment tools.

Purpose of the Study:

• To present a new invariant analysis method for comparing human gestures, focusing on surgical tool movements.
• To evaluate the proposed method's effectiveness in differentiating between expert and novice surgical techniques.

Main Methods:

• Digitizing gesture position and orientation using active 3D sensors.
• Computing invariant curvature of 3D position and orientation via Frenet-Serret frames and quaternion space.
• Comparing gesture curvature to a 6-D template using Dynamic Time Warping (DTW).

Main Results:

• The developed algorithm provides an invariant comparison of gestures, unaffected by sensor placement.
• Experimental results demonstrate the method's efficiency in analyzing forceps blade placements in obstetrics.
• The analysis successfully distinguished between senior and novice medical doctor performances.

Conclusions:

• The proposed curvature analysis method offers an effective and invariant approach for human gesture comparison.
• This technique has significant potential for objective surgical skill assessment and training.
• The method's invariance to sensor parameters enhances its applicability in real-world scenarios.