Kinect-driven Patient-specific Head, Skull, and Muscle Network Modelling for Facial Palsy Patients

Facial palsy negatively affects both professional and personal life qualities of involved patients. Classical facial rehabilitation strategies can recover mimics into their normal symmetrical movements appearances. However, there is a lack objective, quantitative, in-vivo texture muscle activation bio-feedbacks for personalizing programs diagnosing recovering progresses. Consequently, this study proposed novel patient-specific modelling method generating full patient specific head model from visual sensor then computing the in real-time further clinical decision making. The modeling workflow includes (1) Kinect-to-head, (2) head-to-skull, (3) network definition & generation processes. In Kinect-to-head process, subject-specific data acquired new user neutral mimic were used his/her geometrical with texture. particular, template was deformed to optimally fit high-definition points by Kinect sensor. Moreover, also merged images left, right, center view. head-to-skull generic skull so that its shape statistically fitted model. defined models strains during movements. Muscle insertion attachment as vertex positions on respectively based standard anatomy. Three healthy subjects two patients selected validating method. positions, magnetic resonance imaging (MRI)-based compared Kinect-based models. infrared depth-based smiling [u]-pronouncing appropriate animated Kinect-driven Hausdorff distance metric these comparisons. computed lengths tested validated reported values literature. With current hardware configuration, could be fast generated within 17.16±0.37s framerate 40 fps. best mean error 1.91 mm 3.21 On regions, errors 1.53 2.82 respectively. insertion/attachment point 1.09 2.16 2.02 2.00 regions. All above one-time validation procedure. muscles exhibited shortening elongating pronunciation sound [u] Extracted features (i.e. length strain) are agreement experimental literature data. This animating biomechanical bio-feedbacks. applied muscle-oriented paralysis grading other analysis applications.