Latest advance in the scatter calibration and combining the scatter calibration with a practical scatter and random approximation technique for dynamic brain imaging in high resolution PET

We describe the latest advance of a novel scatter calibration technique and a combined method (a novel scatter calibration + a practical scatter and random approximation) which further accelerates the image reconstruction task and also improves the quantitative accuracy for dynamic brain imaging in high resolution PET. The basic idea of the combined method is to apply the scatter calibration to the frames during the early stage of the scan which typically contain a high random fraction and a low number of counts and then apply the practical approximation to the later stage of the scan. The scatter calibration technique is based on using the scatter estimate obtained from a reference frame, which is free from the bias due to high random fractions and/or low numbers of counts, to calibrate the scatter in each dynamic frame. The latest advance in the scatter calibration incorporating the pulse pile-up information is presented with time activity curve (TAC) and emission image validations using human brain studies with short frame durations. The practical approximation is based on using an averaged scatter/random estimate and scaling the average estimate according to the number of trues/random counts to obtain the individual scatter/random estimate for each dynamic frame. The computation time of scatter and random estimates was compared between the conventional (frame-based) and the combined method. The accuracy obtained from the combined method was examined by comparing the binding potential (BP) values between the conventional method, the combined method, and the scatter calibration method (gold-standard) using human studies acquired on the high resolution research tomograph (HRRT). A ∼70% time gain in the scatter and random estimations with an addition of more accurate BP values were achieved by applying the combined method as compared to the conventional frame-based method.