Motion Correction in Myocardial SPECT Imaging Using Polynomial Curve Modeling

Background and Patient motion during myocardial perfusion SPECT can produce artifacts in reconstructed images which might affect clinical diagnosis. This paper attempts to present a new approach for the detection and correction of cardiac motion utilizing the data obtained during the imaging process. Our method quantifies motion through polynomial curves modeling onto the 1D vectors corresponding to image frames obtained in summed profiles. To evaluate our method, physical sources and NCAT phantom were intentionally moved 1, 2 and 3 pixels at early, mid and late stages of data acquisition. The full width at half maximum (FWHM) values, the maximum count in the reconstructed image of physical sources, and maximum count in the myocardial wall in 360o view from cardiac perspective were compared with motion-free condition before and after motion correction. Depending on the amount of motion and time of motion occurrence, the FWHM of point source changed from 5.56% to 37.49% compared to the motion-free condition while the relative change of FWHM for these sources changed from 0.8% to 1.84% after correction. The observed changes for maximum count in reconstructed images of point sources were from 1.91% - 21.33% which decreased to 0.13%-0.76% after motion correction. The corresponding changes for extended sources were 1.09%- 21.76% which reduced to 0.06%-0.45% after applying motion correction. In addition, the results showed that the distribution of radioactivity in the cardiac wall for the corrected image are very similar to the real values while the same values in the reconstructed image significantly differed from the original in which motion occurred during imaging. Our method detected and corrected cardiac motion during SPECT imaging through curve fitting of a polynomial onto the 1D vector values corresponding to image frames. Therefore, image misinterpretation can be avoided by correcting motion artifacts.