skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: Precision and accuracy of regional radioactivity quantitation using the maximum likelihood EM reconstruction algorithm

Abstract

The imaging characteristics of maximum likelihood (ML) reconstruction using the EM algorithm for emission tomography have been extensively evaluated. There has been less study of the precision and accuracy of ML estimates of regional radioactivity concentration. The authors developed a realistic brain slice simulation by segmenting a normal subject's MRI scan into gray matter, white matter, and CSF and produced PET sinogram data with a model that included detector resolution and efficiencies, attenuation, scatter, and randoms. Noisy realizations at different count levels were created, and ML and filtered backprojection (FBP) reconstructions were performed. The bias and variability of ROI values were determined. In addition, the effects of ML pixel size, image smoothing and region size reduction were assessed. ML estimates at 1,000 iterations (0.6 sec per iteration on a parallel computer) for 1-cm[sup 2] gray matter ROIs showed negative biases of 6% [+-] 2% which can be reduced to 0% [+-] 3% by removing the outer 1-mm rim of each ROI. FBP applied to the full-size ROIs had 15% [+-] 4% negative bias with 50% less noise than ML. Shrinking the FBP regions provided partial bias compensation with noise increases to levels similar to ML. Smoothing of ML images producedmore » biases comparable to FBP with slightly less noise. Because of its heavy computational requirements, the ML algorithm will be most useful for applications in which achieving minimum bias is important.« less

Authors:
; ; ; ;  [1]
  1. (National Institutes of Health, Bethesda, MD (United States))
Publication Date:
OSTI Identifier:
6926009
Alternate Identifier(s):
OSTI ID: 6926009
Resource Type:
Journal Article
Resource Relation:
Journal Name: IEEE Transactions on Medical Imaging (Institute of Electrical and Electronics Engineers); (United States); Journal Volume: 13:3
Country of Publication:
United States
Language:
English
Subject:
62 RADIOLOGY AND NUCLEAR MEDICINE; BRAIN; POSITRON COMPUTED TOMOGRAPHY; ACCURACY; ALGORITHMS; IMAGE PROCESSING; BODY; CENTRAL NERVOUS SYSTEM; COMPUTERIZED TOMOGRAPHY; DIAGNOSTIC TECHNIQUES; EMISSION COMPUTED TOMOGRAPHY; MATHEMATICAL LOGIC; NERVOUS SYSTEM; ORGANS; PROCESSING; TOMOGRAPHY 550601* -- Medicine-- Unsealed Radionuclides in Diagnostics

Citation Formats

Carson, R.E., Yan, Y., Chodkowski, B., Yap, T.K., and Daube-Witherspoon, M.E. Precision and accuracy of regional radioactivity quantitation using the maximum likelihood EM reconstruction algorithm. United States: N. p., 1994. Web. doi:10.1109/42.310884.
Carson, R.E., Yan, Y., Chodkowski, B., Yap, T.K., & Daube-Witherspoon, M.E. Precision and accuracy of regional radioactivity quantitation using the maximum likelihood EM reconstruction algorithm. United States. doi:10.1109/42.310884.
Carson, R.E., Yan, Y., Chodkowski, B., Yap, T.K., and Daube-Witherspoon, M.E. Thu . "Precision and accuracy of regional radioactivity quantitation using the maximum likelihood EM reconstruction algorithm". United States. doi:10.1109/42.310884.
@article{osti_6926009,
title = {Precision and accuracy of regional radioactivity quantitation using the maximum likelihood EM reconstruction algorithm},
author = {Carson, R.E. and Yan, Y. and Chodkowski, B. and Yap, T.K. and Daube-Witherspoon, M.E.},
abstractNote = {The imaging characteristics of maximum likelihood (ML) reconstruction using the EM algorithm for emission tomography have been extensively evaluated. There has been less study of the precision and accuracy of ML estimates of regional radioactivity concentration. The authors developed a realistic brain slice simulation by segmenting a normal subject's MRI scan into gray matter, white matter, and CSF and produced PET sinogram data with a model that included detector resolution and efficiencies, attenuation, scatter, and randoms. Noisy realizations at different count levels were created, and ML and filtered backprojection (FBP) reconstructions were performed. The bias and variability of ROI values were determined. In addition, the effects of ML pixel size, image smoothing and region size reduction were assessed. ML estimates at 1,000 iterations (0.6 sec per iteration on a parallel computer) for 1-cm[sup 2] gray matter ROIs showed negative biases of 6% [+-] 2% which can be reduced to 0% [+-] 3% by removing the outer 1-mm rim of each ROI. FBP applied to the full-size ROIs had 15% [+-] 4% negative bias with 50% less noise than ML. Shrinking the FBP regions provided partial bias compensation with noise increases to levels similar to ML. Smoothing of ML images produced biases comparable to FBP with slightly less noise. Because of its heavy computational requirements, the ML algorithm will be most useful for applications in which achieving minimum bias is important.},
doi = {10.1109/42.310884},
journal = {IEEE Transactions on Medical Imaging (Institute of Electrical and Electronics Engineers); (United States)},
number = ,
volume = 13:3,
place = {United States},
year = {Thu Sep 01 00:00:00 EDT 1994},
month = {Thu Sep 01 00:00:00 EDT 1994}
}
  • The expectation-maximization (EM) algorithm for computing maximum-likelihood estimates of transmission images in positron-emission tomography (PET) is extended to include measurement error, accidental coincidences and Compton scatter. A method for accomplishing the maximization step using one step of Newton's method is proposed. The algorithm is regularized with the method of sieves. Evaluations using both Monte Carlo simulations and phantom studies on the Siemens 953B scanner suggest that the algorithm yields unbiased images with significantly lower variances than filtered-backprojection when the images are reconstructed to the intrinsic resolution. Large features in the images converge in under 200 iterations while the smallest featuresmore » required up to 2,000 iterations. All but the smallest features in typical transmission scans converge in approximately 250 iterations. The initial implementation of the algorithm requires 50 sec per iteration o a DECStation 5000.« less
  • Algorithms that calculate maximum likelihood (ML) and maximum a posteriori solutions using expectation-maximization have been successfully applied to SPECT and PET. These algorithms are appealing because of their solid theoretical basis and their guaranteed convergence. A major drawback is the slow convergence, which results in long processing times. This paper presents two new heuristic acceleration methods for maximum likelihood reconstruction of ECT images. The first method incorporates a frequency-dependent amplification in the calculations, to compensate for the low pass filtering of the back projection operation. In the second method, an amplification factor is incorporated that suppresses the effect of attenuationmore » on the updating factors. Both methods are compared to the one-dimensional line search method proposed by Lewitt. All three methods accelerate the ML algorithm. On the test images, Lewitt's method produced the strongest acceleration of the three individual methods. However, the combination of the frequency amplification with the line search method results in a new algorithm with still better performance. Under certain conditions, an effective frequency amplification can be already achieved by skipping some of the calculations required for ML.« less
  • Purpose: To investigate the performance of a new penalized-likelihood PET image reconstruction algorithm using the 1{sub 1}-norm total-variation (TV) sum of the 1st through 4th-order gradients as the penalty. Simulated and brain patient data sets were analyzed. Methods: This work represents an extension of the preconditioned alternating projection algorithm (PAPA) for emission-computed tomography. In this new generalized algorithm (GPAPA), the penalty term is expanded to allow multiple components, in this case the sum of the 1st to 4th order gradients, to reduce artificial piece-wise constant regions (“staircase” artifacts typical for TV) seen in PAPA images penalized with only the 1stmore » order gradient. Simulated data were used to test for “staircase” artifacts and to optimize the penalty hyper-parameter in the root-mean-squared error (RMSE) sense. Patient FDG brain scans were acquired on a GE D690 PET/CT (370 MBq at 1-hour post-injection for 10 minutes) in time-of-flight mode and in all cases were reconstructed using resolution recovery projectors. GPAPA images were compared PAPA and RMSE-optimally filtered OSEM (fully converged) in simulations and to clinical OSEM reconstructions (3 iterations, 32 subsets) with 2.6 mm XYGaussian and standard 3-point axial smoothing post-filters. Results: The results from the simulated data show a significant reduction in the 'staircase' artifact for GPAPA compared to PAPA and lower RMSE (up to 35%) compared to optimally filtered OSEM. A simple power-law relationship between the RMSE-optimal hyper-parameters and the noise equivalent counts (NEC) per voxel is revealed. Qualitatively, the patient images appear much sharper and with less noise than standard clinical images. The convergence rate is similar to OSEM. Conclusions: GPAPA reconstructions using the 1{sub 1}-norm total-variation sum of the 1st through 4th-order gradients as the penalty show great promise for the improvement of image quality over that currently achieved with clinical OSEM reconstructions.« less
  • The authors have implemented the EM reconstruction algorithm for volume acquisition from current generation retracted-septa PET scanners. Although the software was designed for a GE Advance scanner, it is easily adaptable to other 3D scanners. The reconstruction software was written for an Intel iPSC/860 parallel computer with 128 compute nodes. Running on 32 processors, the algorithm requires approximately 55 minutes per iteration to reconstruct a 128 x 128 x 35 image. No projection data compression schemes or other approximations were used in the implementation. Extensive use of EM system matrix (C{sub ij}) symmetries (including the 8-fold in-plane symmetries, 2-fold axialmore » symmetries, and axial parallel line redundancies) reduces the storage cost by a factor of 188. The parallel algorithm operates on distributed projection data which are decomposed by base-symmetry angles. Symmetry operators copy and index the C{sub ij} chord to the form required for the particular symmetry. The use of asynchronous reads, lookup tables, and optimized image indexing improves computational performance.« less
  • Recently, the authors proposed an extension of the expectation maximization (EM) algorithm that was able to handle regularization terms in a natural way. Although very general, convergence proofs were not valid for many possibly useful regularizations. They present here a simple convergence result that is valid assuming only continuous differentiability of the penalty term and can be also extended to other methods for penalized likelihood estimation in tomography.