A noise equivalent counts approach to transmission imaging and source design

Abstract

Noise equivalent counts are a convenient and effective means to assess PET (positron emission tomography) emission image quality. The authors have extended the method to include the effects of transmission imaging on the statistics of attenuation-corrected PET data. The result of the calculations is a noise figure which describes the SNR (signal-to-noise ratio) performance of the elements of the attenuation-corrected emission sinogram. The noise figure demonstrates the trade-off between emission and transmission imaging performance, and can be used to determine optimal partitioning of imaging time between emission and transmission scans. Also, the technique can be used to compare the efficacy of simultaneous transmission/emission imaging techniques and multiple orbiting rod source geometries. Experimental and simulated results from the GE 2048 PET scanner are used to demonstrate the model. In a sample imaging situation in that system geometry (0.2- mu Ci /cm/sup 3/ activity in a 20-cm flood phantom, 3-mCi orbiting rod source(s)), an optimal single rod source protocol achieves a noise figure of 14.2 (arbitrary units); a dual rod source, 11.4; and simultaneous transmission/emission scanning, 10.7. >