An Iterative Based Motion Correction for SPECT with Simulated Data

Abstract

Purpose: Single Photon Emission Computed Tomography (SPECT) is one of the important imaging modality in nuclear medicine used for diagnostic imaging of different organs of human body. Patient movement during the acquisition time is a well-known cause of artifacts in reconstructed SPECT data. These motion artifacts may significantly affect the diagnostic accuracy. Therefore, the motion correction of patients in tomography images is very much essential for more accurate diagnosis and hence achieves the quality of the images.

Objective: The aim of this project is to develop and evaluate a data driven approach to motion correction without necessity for motion tracking system.

Methods: It has been proposed a method to iteratively estimate and compensate this motion during the reconstruction. The rigid motion was estimated view-by-view in every iteration and then used to update the system matrix. The initial reconstructed image is motion contaminated; it can be used to generate a first rough motion estimate. This motion is taken into account in a reconstruction process to generate a motion-corrected image at the first iteration. Then the motion-corrected image and the motion estimate are alternately updated to increase the likelihood, the iterations are stopped when the updated motion seems to have converged. The image update is done by applying multiple iterations of the OSEM algorithm. A final iterative reconstruction was performed with the last motion estimate.

Results: The method was evaluated on simulations phantom study. From the simulations phantom data, we produced the several simulated motion induced data. Motion correction algorithms were applied with the simulated motion induced data. The quality of the reconstructed images was improved substantially after the compensation. Mean structural similarity was increased. In the simulation studies, most of the motion blurring in the reconstructed images disappeared after the compensation.

Conclusions: The proposed method effectively eliminated motion-induced artifacts in SPECT scans. Since only measured raw data are needed for the motion estimation and compensation, the proposed method can be applied retrospectively to SPECT scans affected by motion.

Bangladesh J. Nuclear Med. 26(1): 19-24, 2023