Evaluation of attenuation correction methodology in the Allegro™ PET system

This paper describes the evaluation of an attenuation correction methodology implemented in the Philips Allegro™ PET device. The system uses a ¹³⁷Cs point source (single photon emitter at 662 keV) to perform post-injection transmission scans for clinical whole body FDG investigations. The aims of our study were (i) to assess the use of a single scaling factor in the conversion of the attenuation coefficients from the acquired energy of 662 keV to the emission energy of 511 keV, throughout a range of different density materials; (ii) to assess the effects of scatter in the accuracy of the measured attenuation maps and (iii) to evaluate the accuracy of post-injection transmission scanning. Firstly, measurements of attenuation coefficients were carried out for different density materials at 662 keV and 511 keV using a narrow beam geometry laboratory set-up. The same materials were subsequently scanned using the transmission set-up of the PET system. In addition, emission and transmission datasets of the NEMA NU2 phantom including the three inserts (water, air and teflon) were obtained using the Allegro PET system. Transmission scans of the NEMA phantom were carried out in the presence and absence of activity in order to assess the accuracy of the post-injection transmission acquisitions. Finally, the transmission maps of patients studies were analysed. The attenuation coefficients measured with the laboratory benchmark are in good agreement with tabulated data available in the literature. A linear relationship between the attenuation coefficients at 662 keV and 511 keV, through the range of material densities examined was observed for both the benchmark and the scanner transmission measurements. No differences were observed between the attenuation coefficients in the NEMA phantom in the presence and absence of emission activity. Finally, both phantom and patient studies reveal a bias in areas of low densities such as in water and lung, most probably as a result of using a single scaling factor for all attenuation coefficients in order to correct for the effects of scattered events in the measured transmission maps.