The influence of misregistration between CT and SPECT images on the accuracy of CT-based attenuation correction of cardiac SPECT/CT imaging: Phantom and clinical studies

Integration of single photon emission computed tomography (SPECT) and computed tomography (CT) scanners into SPECT/CT hybrid systems permit detection of coronary artery disease in myocardial perfusion imaging (MPI). Misregistration between CT and emission data can produce some errors in uptake value of SPECT images. The aim of this study was evaluate the influence of attenuation correction (AC) versus non-attenuation correction (NC) images and the effect of misregistration on all segments of SPECT images for quantitative and qualitative analysis. 99 patients (45 males, 54 females) underwent stress/rest myocardial perfusion imaging (MPI) using 99mTc-MIBI were used in this study. We also utilized cardiac insert and lung insert in cylinder phantom. Phantom studies were performed with and without defect. The misregistration of all patient data was measured and variation in misregistration of our population was recorded. The effect of attenuation correction (AC) and non-attenuation correction (NC) images were also evaluated in both phantom and patient data. The CT images were shifted by ±1, ±2, ±3 pixels along X-, Y- and Z-axis (Left/right, dorsal/ventral, cephalic/caudal) for both phantom and patient studies. Differences between misalignment data and misregistration correction images were also measured. Results displayed with 20 segments polar map analysis and illustration in standard orientations for cardiac tomographic images. In the patient population data, 1.5% were perfectly registered, 17% and 73% misaligned under 1 pixel and more than 1 pixel, respectively. AC of SPECT images showed increased uptake value in normal phantom and false positives findings were disappeared versus to NC images. In patient data, statistically significant variation were shown for the most segments before and after AC (P-value<=0.004) and also between AC of SPECT image and misregistration correction images (P-value<=0.048). Along X-axis, in 3 pixel shift in right direction, the percent of relative difference in lateral wall were 11.94% for mid anterolateral. Along Y-axis, the Ventral shift caused -15.9% changes in basal inferolateral and along Z-axis -8.59 % changes in apical anterolateral were also observed in caudal direction when 3 pixel shifts were used. This study showed that CT-based attenuation correction of cardiac images in hybrid SPECT/CT is important to improve image quality. Misalignment in caudal, cephalad, ventral and right direction introduced significant variation even in 1 pixel shift. It is important to apply misregistration correction even in small misalignment routinely in clinical myocardial perfusion imaging.