Theoretical Study of Penalized-Likelihood Image Reconstruction forRegion of Interest Quantification
Region of interest (ROI) quantification is an important taskin emission tomography (e.g., positron emission tomography and singlephoton emission computed tomography). It is essential for exploringclinical factors such as tumor activity, growth rate, and the efficacy oftherapeutic interventions. Statistical image reconstruction methods basedon the penalized maximum-likelihood (PML) or maximum a posterioriprinciple have been developed for emission tomography to deal with thelow signal-to-noise ratio of the emission data. Similar to the filtercut-off frequency in the filtered backprojection method, theregularization parameter in PML reconstruction controls the resolutionand noise tradeoff and, hence, affects ROI quantification. In this paper,we theoretically analyze the performance of ROI quantification in PMLreconstructions. Building on previous work, we derive simplifiedtheoretical expressions for the bias, variance, and ensemblemean-squared-error (EMSE) of the estimated total activity in an ROI thatis surrounded by a uniform background. When the mean and covariancematrix of the activity inside the ROI are known, the theoreticalexpressions are readily computable and allow for fast evaluation of imagequality for ROI quantification with different regularization parameters.The optimum regularization parameter can then be selected to minimize theEMSE. Computer simulations are conducted for small ROIs with variableuniform uptake. The results show that the theoretical predictions matchthe Monte Carlo results reasonably well.
- Research Organization:
- Lawrence Berkeley National Lab. (LBNL), Berkeley, CA (United States)
- Sponsoring Organization:
- USDOE Director, Office of Science. Office of Biological andEnvironmental Research. Medical Sciences Division; National Institutes ofHealth
- DOE Contract Number:
- DE-AC02-05CH11231; NIHR01 EB00194, R01 EB00363, R01HL71253
- OSTI ID:
- 895524
- Report Number(s):
- LBNL-59547; R&D Project: L0213; BnR: 400412000
- Journal Information:
- IEEE Transactions on Medical Imaging, Vol. 25, Issue 5; Related Information: Journal Publication Date: 1 May 2006
- Country of Publication:
- United States
- Language:
- English
Similar Records
Task-based detectability in CT image reconstruction by filtered backprojection and penalized likelihood estimation
Noise-resolution tradeoffs in x-ray CT imaging: A comparison of penalized alternating minimization and filtered backprojection algorithms