Separation of factor images for blood flow estimation in positron emission tomography using ensemble independent component analysis

To calculate regional myocardial blood flow, ensemble ICA (independent component analysis) was evaluated. Myocardial blood flow was estimated from dynamic H/sub 2//sup 15/O PET, and perfusion score was computed from myocardial SPECT data in this study. In ensemble ICA, posterior pdf was approximated by a rectified Gaussian distribution to incorporate non-negativity constraint, which is suitable to dynamic images in nuclear medicine. Major cardiac components were separated successfully by the ensemble ICA method. Mean myocardial blood flow was 1.2/spl plusmn/0.40 ml/min/g in rest, 1.85/spl plusmn/1.12 ml/min/g in stress state. Reproducibility of data analysis for blood flow values were highly correlated (r=0.99). The image contrast between left ventricle and myocardium was better than other method. Perfusion reserve was decreased in the stenosis region. Applied ensemble ICA method will be a feasible method to handle dynamic image sequence obtained by the nuclear medicine techniques.