NEMA NU4-2008 image quality performance report for the microPET focus 120 and for various transmission and reconstruction methods

Abstract

This work aimed to evaluate the image quality and accuracy of attenuation and scatter corrections provided with the microPET Focus 120 scanner using the National Electrical Manufacturers Association NU4-2008 image quality phantom.

Methods: Attenuation correction was obtained from transmission measurements using either a (68)Ge or a (57)Co point source. Fully corrected emission images were reconstructed using Fourier rebinning (FORE) and filtered backprojection (FBP). For attenuation data obtained with the (57)Co source, fully corrected emission images were also reconstructed using FORE and 2-dimensional (2D) ordered-subset expectation maximization (OSEM), 3-dimensional (3D) filtered backprojection (3DRP), 3D OSEM, and 3D maximum a posteriori methods. The mean activity, the coefficients of variation (COVs) of the uniform slices, the recovery coefficients (RCs) for hot rods, and the spillover ratio (SOR) for nonemitting water and air compartments were measured.

Results: For (57)Co-based attenuation correction, the mean activity value differed by less than 3% from the true activity. Measuring the attenuation with (68)Ge resulted in lower reconstructed activity and higher COV. On the basis of (57)Co measurements, the SORs for air and water nonemitting compartments were the closest to zero for attenuation correction. The RC measured on emission images corrected for attenuation but not for scatter did not show any significant difference linked to the transmission method. However, higher RCs were noted for transmission measurement with (68)Ge in coincidence with windowing when emission data were corrected for attenuation and scatter. This resulted from a lower mean value in the uniform area. 2D and 3DRP reconstruction methods showed little effect on the mean activity value, whereas iterative 3D methods gave 7% higher values. Higher RCs were found with iterative reconstruction than with FBP and 3DRP. However, the SOR seemed to be optimal with FBP. SORs were higher with iterative methods and decreased with the number of iterations.

Conclusion: For studies of small rodents with the Focus 120, (57)Co transmission seems to be the most suitable method for attenuation correction. FORE and 2D reconstruction methods appear to be a good compromise between overall image quality and reconstruction time: OSEM provides the largest contrasts, but FBP provides superior attenuation and scatter correction.