Simultaneous activity and crystal efficiencies reconstruction for continuous motion bed acquisition

Abstract

PET scanner quality check control is performed on a daily basis. In Siemens scanners, for example, data are acquired from a uniform cylinder and the estimation of a crystal efficiencies (CE) normalization component is carried out. Besides producing a normalization array, the CE analysis provides information about whether or not additional calibration should be performed. In previous work we demonstrated that the CE can be simultaneously reconstructed with activity from TOF patient data. Therefore, scanner performance can be monitored as patient scans occur, eliminating the need for the frequent quality check scan. We extended this work to include Continuous Bed Motion (CBM) acquisition. In CBM data formation, counts from various detector pairs are combined, taking into account the bed motion. While it is beneficial for activity reconstruction, axial CE structure is practically lost in the corresponding computed normalization array due to averaging over motion. We suggest the use of additional data, produced similarly with step by step (stationary) acquisition. In this complimentary data set, activity is integrated over motion; however, the CE structure is preserved. The same rebinner can be used to generate both data sets. The use of two complimentary data sets makes CBM a practical option for a patient based scanner performance check. The algorithm can be interpreted as activity ML reconstruction with a nested loop of CE estimation, which uses the same data compressed differently. Patient data from a Siemens mCT Flow scanner were used to validate the algorithm's performance. The proposed method produced crystal efficiencies maps similar to those of the daily scans and therefore is suitable for a patient-based Quality Control (QC) procedure.