Evaluation of a multi-camera visual-tracking system for detecting and correcting body motion in cardiac SPECT

Objectives Body motion of ~13 mm is sufficient to cause minor to moderate defects in Cardiac SPECT. We designed a clinical motion-monitoring system using multiple near-infrared cameras from Vicon to stereo-image small retro-reflective markers on strechy bands wrapped about the chest and abdomen. The goal was to evaluate in patients the accuracy with which the motion can be corrected when incorporated in iterative reconstruction with attenuation and detector response compensation.

Methods Twelve patients (8 men, 4 women thus far) undergoing Tl-201 stress-rest cardiac perfusion imaging were ask to move a small amount during a second rest acquisition directly after a first rest with Beacon (Philips Medical, Milpitas, CA) transmission. The visual tracking system (VTS) contineously monitored patients during the entire study. The first rest served as baseline for comparison, while the second was reconstructed with a) no motion correction, b) VTS determined motion correction, and c) the standard motion correction software in our clinic. Indenpendent polar-map quantification using a site-specific normal database was used for evaluation.

Results The VTS corrected slices compared more favorably with the first rest study in 11 of the 12 patients than slices corected using the clinical software. Also, in the majority of cases the diagnosis (as given by polar-map quantifiction) were not significntly altered with VTS determined motion correction while the clinical software was wrong in 50%. Currently, 3DOF compensation is more robust than 6DOF compensation.

Conclusions We were able to show that VTS motion detection and correction is possible in a routine clinical setting.

Research Support NIH R01 EB001457, R01 HL091017, and Philips Medical Systems