A study of the effect of potential angular under-sampling on myocardium perfusion imaging with non-rotational dedicated cardiac SPECT geometry

Objectives Non-rotational dedicated cardiac SPECT geometry is gaining more industrial attraction mainly due to the benefit of simultaneous data acquisition from different angles and no rotation of the gantry. A draw back of this geometry is that the region of complete angular sampling is small and the patient heart or part of it may be under sampled angularly (AUS), e.g., for large patients. This work is to study the effect of AUS on myocardium perfusion imaging.

Methods We acquired a complete data set (parallel-beam data, 180°, 64 projection views, RAO 45° to LPO 45°) of an anthropomorphic phantom with a cardiac insert. The insert had two defects and was otherwise uniform. We generated two series of AUS data sets by removing different number of views from the end or beginning of the complete data set. The generated AUS data had angular range of 169°, 160°, 149°, 141°, and 129°, corresponding to AUS of 11°, 20°, 31°, 39°, and 51°, respectively. The first series had the same start angle RAO 45° and the second the same end angle LPO 45°. We used an MLEM algorithm with 48 iterations for image reconstruction. Reconstructed images were compared to that from the complete data set.

Results There was no visual difference between the 169° and 180° images. With start angle RAO 45°, AUS of 20° or more led to reduced definition of the basal-inferior defect; AUS of 30° or more led to reduced definition of the mid-anterior defect and introduced an apical artifact. With end angle LPO 45°, AUS of 20° led to a slight apical decrease, the decrease became an obvious apical artifact when AUS was 39° or more. AUS of 59° led to reduced definition of the basal-inferior defect.

Conclusions AUS can occur due to large patient size or poor patient positioning when using non-rotational dedicated cardiac SPECT geometry. When the AUS is 20° or more, it can decrease defect definition and introduce artifacts in myocardium perfusion SPECT.

Research Support This work was fully supported by Digirad Corporation, Poway, CA, USA