Usefulness of a new anthropomorphic phantom simulating the chest and abdomen regions in PET tests

Abstract

Objective

To investigate the clinical utility of a new anthropomorphic phantom that reproduces the chest and abdomen better than the conventional National Electrical Manufacturers Association (NEMA) body phantom, count rates and image quality of PET images obtained from patients were evaluated.

Methods

Anthropomorphic phantoms were used to include radioactivity in the lung, liver, kidney, and background regions. Two NEMA body phantoms were used for chest and abdominal assessments. The cross calibration factor (CCF) cylinder phantom was also used to reproduce the distribution of radioactivity outside the field of view, simulating the patient brain. Four types of phantoms were used in the PET imaging experiment, and for each phantom, the prompt coincidence count rates, random coincidence count rates, true + scatter coincidence count rates, and single photon count rates were measured. Then, these count rates were compared with count rates from actual clinical data. PET image quality assessment was done using the parameters, noise equivalent count patient (NEC_patient), noise equivalent count density (NEC_density), and liver signal-to-noise ratio (SNR).

Results

Random coincidence count rates showed that the data obtained from each phantom were in good agreement with the clinical data. True + scatter coincidence count rates had better agreement with clinical data when measured for anthropomorphic phantoms than for the NEMA body phantoms. Furthermore, when the CCF Cylinder phantom simulating the brain was placed outside the imaging field of view, the results were closer to the clinical data. PET image quality was 1.4% higher for NEC_patient obtained from anthropomorphic phantoms compared to the mean obtained from clinical data. NEC_density was 15.0% lower than the mean value obtained from clinical data. Liver SNR was 14.8% higher in PET images reconstructed using the 3D-ordered subsets expectation maximization (OSEM) method. It was 10.0% lower in PET images reconstructed with the image reconstruction method Q.Clear (GE Healthcare) using the Bayesian penalized likelihood (BPL) method.

Conclusion

The new anthropomorphic phantom was more consistent with the count rates obtained from clinical data than the conventional NEMA body phantoms were and it was able to better simulate the distribution of radioactivity concentrations in the patients by reproducing the distribution of radioactivity concentrations outside the field of view.