Advancements and Clinical Applications of Machine Learning for Hand Pose Estimation.

Abstract

Hand pose estimation has substantial potential for clinical applications by accurately capturing the kinematics of the hand. Hand pose estimation employs the following two main approaches: vision-based methods, such as Red Green Blue Depth cameras, and sensor-based methods involving wearable devices. Machine learning enables the development of models that can accurately predict hand pose estimation metrics using large, complex data sets. Despite marked progress, challenges remain, including computational requirements, anatomical complexity, and the lack of clinical data sets for model training, particularly for pathologies affecting the hand.