Improving the Temporal Resolution of Tomographic Images Using a PI-Line Based Backprojection Filtration Algorithm

Abstract

The fan-beam backprojection filtration (BPF) algorithm is unique in that the temporal resolution of a given pixel within a region-of-interest (ROI) image is related directly to the location of the PI-line segment used to reconstruct this pixel. In particular, a pixel within a PI-line segment crossing the edge of the field-of-view (FOV) exhibits a better temporal resolution than a pixel within a PI-line segment crossing the center of the FOV. This unique property of the BPF algorithm can be exploited by acquiring projection data with the object shifted towards the edge of the FOV and then reconstructing an ROI covering the object with the BPF algorithm. This study demonstrates that this technique, termed off-center object positioning, can be used to suppress motion artifacts. A dynamic ball phantom undergoing contractile motion during 80% of its temporal cycle was given six cycle rates ranging from 60 to 120 cycles per minute (cpm). At each cycle rate, projection data were acquired at five different elevations. These elevations, which represented the distance from the bottom of the phantom to isocenter, ranged from 0 to 15 cm. Mean absolute difference MAD, root mean square error RMS, and correlation CORR metrics were applied over the same ROI reconstructed from motion-contaminated and motion-free data. Our results show that quantitative metrics for BPF reconstructions improved with increasing object elevation over a wide range of cycle rates