Data-driven respiratory gating for ventilation/perfusion lung scan

BACKGROUND: Ventilation/perfusion lung scan is subject to blur due to respiratory motion whether with planar acquisition or single photon emission computed tomography (SPECT). We propose a data- driven gating method for extracting different respiratory phases from lung scan list-mode or dynamic data. METHODS: The algorithm derives a surrogate respiratory signal from an automatically detected diaphragmatic region of interest. The time activity curve generated is then filtered using a Savitzky-Golay filter. We tested this method on an oscillating phantom in order to evaluate motion blur decrease and on one lung SPECT. RESULTS: Our algorithm reduced motion blur on phantom acquisition: mean full width at half maximum 8.1 pixels on non-gated acquisition versus 5.3 pixels on gated acquisition and 4.1 pixels on reference image. Automated detection of the diaphragmatic region and time-activity curves generation were successful on patient acquisition. CONCLUSIONS: This algorithm is compatible with a clinical use considering its runtime. Further studies will be needed in order to validate this method.