Comparison between the 1D deconvolution and the ETM scatter correction techniques in PET
- CEA-Service Hospitalier F. Joliot, Orsay (France); and others
Scatter corrections usually degrade the Signal-to-Noise Ratio (SNR) while they improve image quantification. Dual energy corrections provide scatter corrected images with a poor SNR due to the use of two sinograms having low statistics. We have evaluated the SNR on 20 cm uniform cylinder images, acquired on an ECAT 953B/31 with septa in the field-of-view, corrected for scatter using the 1D deconvolution method and an energy based correction developed at Orsay. The latter, referenced as the Estimation of True Method, uses a High Energy Window (HEW) with 550 and 850 keV settings to estimate the true component registered in the Classical Energy Window (CEW) with 250 and 850 keV settings. A sinogram of scattered events is formed from this noisy estimate of the trues. It is filtered and then subtracted from the CEW sinogram to provide a scatter free sinogram. Nine Regions of Interest (ROI) of 18mm diameter have been drawn on a 110 mm diameter circle and reported on 11 direct slices (96 million events each in the CEW and 8 million in the HEW). The SNR has been defined as the ratio of the mean over the standard deviation of all ROI values. With the 1D deconvolution the SNR is 38.0, close to that obtained without scatter correction (39.1) It is lower with the ETM depending on the filter used: with a rectangular window of 9 bins by 15 angles it is 29;.8 (26.9 with a 5 by 5 window) while with a 2D Gaussian filter (7 bins by 13 angles variances) it is 30.8. This value is higher than 22.1 measured on the HEW image. The ETM with adequate filtering allows scatter correction with a SNR acceptable compared with that measured with the 1D deconvolution. Yet the ETM has a clear advantage over the 1D deconvolution in case of asymetrical source distributions in non homogeneous media and in case of off-plane scattering as has been tested on various phantom measurements.
- OSTI ID:
- 197922
- Report Number(s):
- CONF-940605-; ISSN 0161-5505; TRN: 95:007029-0057
- Journal Information:
- Journal of Nuclear Medicine, Vol. 35, Issue Suppl.5; Conference: 41. annual meeting of the Society of Nuclear Medicine, Orlando, FL (United States), 5-8 Jun 1994; Other Information: PBD: May 1994
- Country of Publication:
- United States
- Language:
- English
Similar Records
SU-E-QI-03: Compartment Modeling of Dynamic Brain PET - The Effect of Scatter and Random Corrections On Parameter Errors
An investigation of filter choice for filtered back-projection reconstruction in PET