A virtual trainer by natural user interface for cognitive rehabilitation in dementia

The aim of this work is the design and the development of an ICT platform integrating advanced Natural User Interface technologies for multi-domain Cognitive Rehabilitation without the direct physician involvement to the rehabilitation session. The platform is made up of a set-top-box connected to a TV monitor, a Microsoft Kinect RGB-D sensor and a (optional) WWS Smartex e-shirt for clinical signs monitoring. Customized algorithms for calibration, people segmentation, body skeletonization and hands tracking through the RGB-D sensor have been implemented in order to infer knowledge about the reaction of the end-user to the Graphical User Interface designed for specific cognitive domains. For proper interaction, gestures of Alzheimer Disease's patients are acquired by Microsoft Kinect in the nominal functioning range, allowing 100% hands detection rate, useful for an error free human-machine interaction. © 2014 Springer International Publishing.