Background: Previously we reported that compared to iterative reconstruction with ordered subset expectation maximum (OSEM), wide beam reconstruction (WBR), which incorporates resolution recovery and controls noise during reconstruction without applying a post-processing filter, allows half-time SPECT acquisition with preserved diagnostic quality. We now postulate that with further Poisson noise treatment, quarter-time acquisition is possible.
Methods: The half-time WBR algorithm was optimized for quarter-time acquisition based upon anthropomorphic cardiac phantom data and a pilot group of 48 patients (pts). Then using the modified algorithm, 209 pts (91 men, 118 women, mean chest circumference = 40 in) were imaged at rest (R) and stress (S) (9/32 mCi (99m)Tc-sestamibi) full-time with OSEM, and again quarter-time with the modified WBR algorithm. The 180 degrees , 64-stop, full-time single-day rest (R) (25 second-per-stop, sps) and 8-frame per cardiac cycle post-stress (S) (20 sps) gated SPECT, and then quarter-time R (6 sps) and post-S (4 sps) gated SPECT were acquired. Blinded observers graded scan quality (1 = poor to 5 = excellent) based on myocardial uniformity, endocardial/epicardial edge definition, and background noise. Perfusion defects were scored using a 17-segment model. Using three commercially available software methods, end-diastolic volume (EDV), end-systolic volume (ESV), and left ventricular ejection fraction (LVEF) were calculated.
Results: For the 209 prospective pts, mean image quality for R full-time OSEM and quarter-time WBR were similar (3.5 +/- 0.9 vs 3.6 +/- 0.7, p NS). For S, quarter-time WBR quality was superior to full-time OSEM (4.3 +/- 0.7 vs 3.9 +/- 0.7) (P = 1.78 x 10(-17)). In 35 pts with chest circumferences >44 inches a longer, 10 sps WBR acquisition improved resting image quality. Of 48 pts with abnormal scans (SSSs > 2 by OSEM) mean summed stress scores, summed rest scores, and summed difference scores were not significantly different with quarter-time WBR vs full-time OSEM (11.2 vs 10.9), (9.1 vs 9.0), (2.0 vs 1.9) (P NS). For the three software methods, there was a good correlation of LVEF, EDV, and ESV determined by WBR vs OSEM (all r > 0.92). ESVs were generally higher with WBR, primarily due to better delineation of the valve plane at end-systole, whereas EDVs were similar. Thus, EFs were significantly lower with WBR.
Conclusions: For perfusion SPECT quarter-time WBR affords image quality, defect characterization, and functional assessment equivalent to full-time OSEM.