Title :
Realtime markerless rigid body head motion tracking using the Microsoft Kinect
Author :
Noonan, Philip J. ; Howard, John ; Cootes, Timothy F. ; Hallett, William A. ; Hinz, Rainer
Author_Institution :
Wolfson Mol. Imaging Centre, Univ. of Manchester, Manchester, UK
fDate :
Oct. 27 2012-Nov. 3 2012
Abstract :
Accurate and reliable head motion tracking is essential to enable robust motion correction of brain Positron Emission Tomography (PET). Systems such as the Polaris Position Sensor offer high spatial and temporal resolution, <; 1mm at 20Hz, tracking of a tool which is attached to the head. It is difficult to rigidly fix the tracking tool to the patient without discomfort which results in unreliable data due to tool slippage. An open-source, markerless, tracking system using the consumer grade Microsoft Kinect is presented which provides realtime motion tracking of the head at 30Hz. Realtime tracking is achieved using a general purpose graphics processing unit (GPU). The tracking algorithm is insensitive to changes in facial expressions and partial occlusions. Phantom and human data is presented comparing the tracking offered by both the Kinect based and Polaris system.
Keywords :
brain; graphics processing units; infrared imaging; medical image processing; positron emission tomography; target tracking; Microsoft Kinect; Polaris Position Sensor; brain PET; facial expressions; general purpose GPU; graphics processing unit; open source markerless tracking system; partial occlusions; positron emission tomography; realtime markerless rigid body head motion tracking; robust motion correction; spatial resolution; temporal resolution; tool slippage; tracking algorithm;
Conference_Titel :
Nuclear Science Symposium and Medical Imaging Conference (NSS/MIC), 2012 IEEE
Conference_Location :
Anaheim, CA
Print_ISBN :
978-1-4673-2028-3
DOI :
10.1109/NSSMIC.2012.6551510