PT - JOURNAL ARTICLE AU - Harshali Bal AU - Mehmet Aykac AU - Maurizio Conti TI - <strong>A novel approach for scatter correction in PET using energy response modelling</strong> DP - 2020 May 01 TA - Journal of Nuclear Medicine PG - 108--108 VI - 61 IP - supplement 1 4099 - http://jnm.snmjournals.org/content/61/supplement_1/108.short 4100 - http://jnm.snmjournals.org/content/61/supplement_1/108.full SO - J Nucl Med2020 May 01; 61 AB - 108Objectives: Scatter correction in PET imaging is performed using the model based approach on most commercial scanners. This method requires an emission image estimate and an attenuation map for scatter shape estimation. In addition, the attenuation map is also used in order to estimate regions used to perform scatter tail fitting to account for multiple scatter and out-of-field of view scatter. However, this limits the scatter shape estimation to the axial extent of the attenuation map and in some cases may result in image artifacts due to slight mismatch between PET and CT. In this work, we present a novel method to estimate scatter using the energy information in the PET list-mode data for the Siemens SiPM PET/CT scanner. The proposed approach offers to improve the accuracy of scatter shape by accounting for multiple scattered events, out-of-field of view scatter and 3D scatter without the dependence of the attenuation map. Methods: The method is based upon the premise that for a given isotope, the energy response of the true coincidences will possess a baseline shape which depends only on the scanner configuration. In this method, first the PET list-mode data was binned into sinograms guided by a certain energy sampling rate. Using this technique, a baseline energy response was obtained by binning the list-mode PET data for an F-18 line source. Subsequently, PET measurement for any object/patient was also rebinned into sinograms and the measured energy response for each element of the sinogram was adjusted to match the baseline response and compute scatter. To do this, each measured energy response was normalized to the baseline response using a high energy window subset of the response. The difference between the normalized measured response and baseline response resulted in the raw scatter estimate. Due to the limited energy resolution of the scanner, the raw scatter estimate contains some true coincidences and lacks scatter from high energy bins. To correct for this, a linear combination of high energy scatter and normalized high energy trues was added to obtain the final scatter estimate. The linear coefficients were estimated using a low count-rate NEMA scatter dataset such that the final scatter estimate was matched to the measured scatter. A Gaussian filter was applied to the final scatter estimate and the resulting 3D scatter estimate was referred to as the energy response modelling (ERM) based scatter. In order to validate this approach, F-18 phantom experimental data and few clinical datasets obtained on the Siemens Biograph Vision were used. PETCT scans with cylinder, NEMA IQ and an extended oval phantom were performed. Clinical datasets included whole-body F-18 FDG scans for patients. Each dataset was processed with scatter correction using the single scatter simulation model based relative scatter correction (SSS-rel) and the ERM based approach. Reconstructions were performed using TOF-OP-OSEM (4iterations, 5 subsets). Image quality evaluation was done by comparing activity concentrations obtained for the two approaches with respect to the measured concentration for phantom data and the relative difference in activity concentration with the two methods for clinical data. Results: Image quality obtained between the two scatter correction approaches were comparable. In the case of the uniform cylinder phantom and the NEMA IQ phantom, activity concentrations obtained with the 2 methods was similar. However, for the large oval phantom, activity concentration with ERM was closer to the measured activity concentration and had more uniform activity distribution compared to the SSS-rel approach. Images obtained from clinical data were visually comparable. Conclusions: A novel energy based scatter correction approach was developed and found to provide image quality comparable to that with the model based approach with potential to provide improved scatter correction in the case of large patients and in case of PET and CT mismatch.