Effectiveness of three alternative strategies in reducing the degrading impact ofrespiratory motion on the detection of small pulmonary nodules in SPECT imaging
Abstract
The objective of this investigation is to determine the effectiveness of three methods in reducing the impact of respiratory motion on the detection of small solitary pulmonary nodules in single photon emission computed tomographic imaging. Based on our previous work, 32 NCAT normal torso phantoms were generated over the entire respiratory cycle to create source and attenuation maps for the normal background distribution of Tc-99m NeoTect. Similarly, 32 NCAT sphere phantoms of 1.0 cm diameter were generated over the entire respiratory cycle for each of 150 uniquely located points within the lungs. The SIMIND Monte Carlo program was used to produce SPECT projection data for the normal background and separately for each of 150 lesions for all 32 frames and for a timeaveraged case. Normal and tumor SPECT projection sets each containing one lesion were combined with a clinically realistic noise level and counts as follows: 1) end-expiration with all counts (frame 1), 2) time-averaged with all counts (frame av), and 3) each of the 32 frames containing 1/32 of the total counts. Projection data of the 32 frames was combined to form ten different bins. The first eight bins each consisted of 4 frames. The 9th bin, centered about end-expiration, consisted of 8 frames (quarter binning). The 10th bin, also centered about endexpiration, consisted of 16 frames (half binning). Each of the twelve sets of combined projection data was reconstructed with RBI-EM with RC. Based on known motion for each of the 150 different lesions, the reconstructed volumes of the first eight bins were shifted to superimpose the reconstructions over each of the 150 lesions. Comparison of lesion detection by human-observer LROC studies reveals that quarter binning results in the lowest rate of detection and that the reconstruct and shift method results in the greatest rate of detection. Additionally, the rate of detection by superimposing the reconstructions over the lesion is even superior to frame 1. ©2008 IEEE.