Title :
A bootstrap approach for analyzing the statistical properties of SPECT and PET images
Author :
Buvat, Irene ; Riddell, Cyril
Author_Institution :
U494 INSERM, Paris, France
Abstract :
We describe a non-parametric bootstrap method to estimate the statistical properties of single photon emission computed tomography (SPECT) and positron emission tomography (PET) images, whatever the type of noise in the projections and the reconstruction algorithm. Using analytical simulations and real PET data, this method is shown to accurately predict the statistical distribution, hence the variance, of reconstructed pixel values for both linear and nonlinear reconstruction algorithms.
Keywords :
positron emission tomography; single photon emission computed tomography; statistical analysis; PET; SPECT; bootstrap method; linear reconstruction algorithms; noise; nonlinear reconstruction algorithms; reconstruction algorithm; statistical distribution; variance; Algorithm design and analysis; Analysis of variance; Analytical models; Computational modeling; Image analysis; Positron emission tomography; Predictive models; Reconstruction algorithms; Single photon emission computed tomography; Statistical distributions;
Conference_Titel :
Nuclear Science Symposium Conference Record, 2001 IEEE
Print_ISBN :
0-7803-7324-3
DOI :
10.1109/NSSMIC.2001.1008603
