Comparisons of clinical performance with standard measures of PET cameras

The NEMA and IEC performance tests of PET cameras provide standard measurements of their intrinsic physical characteristics and the accuracy of corrections applied to them. However the clinical performance of PET cameras are more clearly defined by using patient data with clinical acquisition and processing parameters. Thus, a series of simple patient measurements are proposed to supplement the standard performance measurements. NEMA and clinical measurements are reported here for 3 PET cameras: the UGM 240 H, UGM CPET and HEAD PENN PET cameras. Totals and trues countrates and scatter fractions are compared to 6 liter (20 cm long) and 22 liter (70 cm) cylinder data, while tumor detectability data is used to illustrate the effects of camera resolution and sensitivity. Typical patient images with small tumors are compared to proposed NEMA image quality tests. The accuracy of corrections to the data are assessed from normal organ SUVs and from whole brain blood flow and metabolism measurements. Patient countrates are well reproduced by the 6 liter (brain) and 22 liter (body) cylinder data. However patient activities are typically 40-70% of the activities for peak trues countrates. The patient scatter fractions subtracted are 20% higher than NEMA measurements and patient volumes in the FOV are more than twice that of the 20 cm diameter cylinders. The sensitivity for typical patient studies is indicative of study noise per unit activity in the FOV. Superimposing measured hot sphere contrasts (for 4, 7 and 10:1 data) on a plot of tumor SUV against diameter the authors see the effects of partial volume and tumor metabolism on tumor detectability. These examples illustrate the value of patient measurements in defining clinical PET camera performance.