Reducing respiratory motion artifacts in radionuclide imaging through retrospective stacking: A simulation study

Respiratory motion artifacts in radionuclide imaging can substantially increase the apparent volume of malignant lesions, and result in reduced activity and signal-to-noise ratios (SNRs) within the tumor region. We present a corrective algorithm, coined retrospective stacking (RS), that combines retrospective amplitude-based binning of data acquired in small time intervals, with rigid or deformable image registration methods. Retrospective stacking is first applied to numerically simulated radionuclide images of a lesion moving with regular and irregular linear motion, as well as hysteresis characteristic of tumors near the lung. The dependence of RS on spatial and temporal resolution is explored, by comparing cross-section visualizations, activity profiles, and SNRs of retrospectively stacked images with those from a simulated motionless lesion. The simulation results are subsequently validated with a phantom positron emission tomography experiment representing a hot lesion oscillating within a warm background. It is seen that by sufficiently reducing the data acquisition timestep, RS can restore the lesion image to nearly its original shape, intensity and SNR, even under noisy conditions.