Abstract
27
Objectives: Iterative reconstruction (IR) has largely replaced filtered back projection in myocardial perfusion scintigraphy (MPS). In addition to scatter correction (SC) and attenuation correction (AC), modern IR methods include complex physical modeling and noise control algorithms which may result in artifactual perfusion deficits that are not readily apparent as such. Resolution compensation (RC) delays recovery of high frequency detail and requires a larger number of iterations. We studied the effect of the inter-iteration 2D low pass filter (2DF) on the recovery of the anatomical detail by RC.
Methods: We retrospectively reviewed 125 consecutive normal MPS as determined by reconstruction with IR using SC and AC. Stress-only images reconstructed with IR using SC, AC, RC and inter-iteration 2DF (Astonish, Philips Healthcare), were displayed using the AutoQuant program and interpreted by two expert readers (R1 and R2). Left ventricular polar maps were obtained from the short-axis slices using the Quantitative Perfusion SPECT (QPS). Seventeen segment stress perfusion scores were compared to the original stress scores using a z-Test.
Results: Of the studies processed with RC and 2DF, 22.4% (R1) and 26.4% (R2) were interpreted as normal, 40.8% (R1) and 40.0% (R2) probably normal, 16.0% (R1) and 18.4% (R2) equivocal, 13.6% (R1) and 12.8% (R2) probably abnormal and 7.2% (R1) and 2.4% (R2) definitely abnormal. Qualitative perfusion deficits were seen in one or more of the following segments: apical n=83 (R1) and n=70 (R2), septal n=44 (R1) and n=27 (R2), inferior n=4 (R1) and n=24 (R2), lateral n=1 (R1) and n=3 (R2). The defects in the basal septal segments were most pronounced resulting in the appearance of a foreshortened septum. QPS demonstrated defects in the septum (segments 2, 3, 8 and 9), the apical segments (14, 15 and 17), and inferior wall (segment 4). There was a slight increase of activity in the mid anterior wall (segment 7). The remaining 8 segments did not show a significant difference. The attached figure demonstrates two clinical examples of apparent perfusion deficits introduced when RC with 2DF was included in the IR: Patient A demonstrates deficits in apex, septum and inferior wall; Patient B demonstrates deficits in the apex and inferoseptal wall.
Conclusion: Increased number of iterations is needed for IR with RC to converge due to slower recovery of high frequency detail of the imaged object. Inter-iteration 2DF diminishes the higher frequency content and may result in incomplete recovery of imaged anatomy. Iterative reconstruction with RC and 2DF in MPS can result in artifactual perfusion defects in the septal, apical and inferior segments. We recommend that IR with RC should not be routinely paired with inter-iteration 2DF in MPS. Research Support: None