Normalization approach for long axial field-of-view PET system

For PET system, it is necessary to update normalization factor regularly to keep latest estimates of sensitivity of each line of response (LOR). Conventional component-based normalization method need to scan the rotating rod source and a static cylindrical source which cover the whole axial field-of-view (FOV) of system to generate the all normalization factors. For PET system with a long axial FOV, it is unrealizable to update the normalization factor as normal. Therefore, we firstly proposed a complete set of schemes for long axial FOV system to generate all components of normalization. In our method, the geometry components were generated by scanning a short rod source, which rotating around the transverse center at one edge of axial FOV, an additional scale factor generated from simulation data was used to aliment the short rod data as long rod data. For detector efficiency components, a continually moving cylinder source was used instead of long axial cylinder. This method was performed on uMI Panorama GS. We also evaluated the uniformity of a long axial cylinder image, which has similar result with conventional way. The results show that our method greatly simplify the procedure of normalization calibration for long axial FOV PET system.