Object dependency of resolution and convergence rate in OSEM with filtering

Mustafovic, S.; Thielemans, Kris; Bloomfield, Peter

doi:10.1109/nssmic.2001.1008689

Cited by 13 publications

(18 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…28,30,31 Although there is no proof of convergence, this algorithm always appears to converge in practice. The image characteristics are somewhat similar to those of penalized likelihood reconstruction.…”

Section: Expectation-maximisation-smooth (Ems)mentioning

confidence: 99%

Iterative Reconstruction Methods

Hutton

Nuyts

Zaidi

2006

Quantitative Analysis in Nuclear Medicine Imaging

View full text Add to dashboard Cite

Section: Expectation-maximisation-smooth (Ems)mentioning

confidence: 99%

Iterative Reconstruction Methods

Hutton

Nuyts

Zaidi

2006

Quantitative Analysis in Nuclear Medicine Imaging

View full text Add to dashboard Cite

“…Under our parameterized inhomogeneous Poisson process model, the covariance between rate function estimates at any two voxels and at any two time points and (instantaneous) or averaged from to (average) can be obtained via (14) ( 15) where (16) (17) Since rate function estimates are related to control vertices deterministically, an estimate of the covariance matrix, is necessary and sufficient for estimating the variances/covariances of instantaneous or average rate estimates at any pair of voxels at any pair of time points. Once is estimated, it can be used in (14) or (15) to obtain the desired expressions.…”

Section: Variance Estimationmentioning

confidence: 99%

“…Nuyts and Fessler [18] compared the resolution properties of images reconstructed using penalized-likelihood versus those of images reconstructed with maximum-likelihood and postsmoothed with the penalized-likelihood impulse response. Mustafovic et al [14], [15] showed that uniform spatial resolution could also be achieved by applying a spatially varying filter between the iterations of the (ordered subsets) expectation-maximization or separable paraboloidal surrogates algorithms [4]. In contrast to the growing number of papers on the analysis of single frame penalized ML reconstruction, there is little previous work on mean, variance and spatiotemporal resolution properties of penalized ML dynamic PET images.…”

mentioning

confidence: 99%

Mean and covariance properties of dynamic PET reconstructions from list-mode data

Asma

Leahy²

2006

IEEE Trans. Med. Imaging

View full text Add to dashboard Cite

Abstract-We derive computationally efficient methods for the estimation of the mean and variance properties of penalized likelihood dynamic positron emission tomography (PET) images. This allows us to predict the accuracy of reconstructed activity estimates and to compare reconstruction algorithms theoretically. We combine a bin-mode approach in which data is modeled as a collection of independent Poisson random variables at each spatiotemporal bin with the space-time separabilities in the imaging equation and penalties to derive rapidly computable analytic mean and variance approximations. We use these approximations to compare bias/variance properties of our dynamic PET image reconstruction algorithm with those of multiframe static PET reconstructions.

show abstract

“…This interference could be avoided by using a heuristic modi cation. For every pixel, we compute the "directional" Fisher information in vertical, horizontal and diagonal directions using (11). If the "directional" Fisher information (11) is maximum along a vertical or horizontal direction, then the weights of the diagonal elements are simply set to zero.…”

Section: B Emission Tomographymentioning

confidence: 99%

“…image analysis based on kinetic modeling or on standardized uptake values [7]), it is desirable to have uniform spatial resolution. Recently, methods have been proposed to impose uniform resolution, by combining the likelihood with a data dependent penalty [8][9][10] or by tuning the characteristics of a lter applied during interations [11]. An alternative method to obtain uniform resolution is to post-smooth the reconstruction obtained after many iterations of a maximum-likelihood (ML) reconstruction algorithm [12], [13].…”

Section: Introductionmentioning

confidence: 99%

Comparison between post-smoothed maximum-likelihood and penalized-likelihood for image reconstruction with uniform spatial resolution

Fessler

2002 IEEE Nuclear Science Symposium Conference Record

View full text Add to dashboard Cite

Regularization is desirable for image reconstruction in emission tomography. One of the most powerful regularization techniques is the penalized-likelihood reconstruction algorithm (or equivalently, maximum-a-posteriori reconstruction), where the sum of the likelihood and a noise suppressing penalty term (or Bayesian prior) is optimized. Usually, this approach yields position dependent resolution and bias. However, for some applications in emission tomography, a shift invariant point spread function would be advantageous. Recently, a new method has been proposed, in which the penalty term is tuned in every pixel in order to impose a uniform local impulse response. In this paper, an alternative way to tune the penalty term is presented. The performance of the new method is compared to that of the post-smoothed maximum-likelihood approach, using the impulse response of the former method as the post-smoothing filter for the latter. For this experiment, the noise properties of the penalizedlikelihood algorithm were not superior to those of post-smoothed maximum-likelihood reconstruction.

show abstract

Object dependency of resolution and convergence rate in OSEM with filtering

Cited by 13 publications

References 8 publications

Iterative Reconstruction Methods

Iterative Reconstruction Methods

Mean and covariance properties of dynamic PET reconstructions from list-mode data

Comparison between post-smoothed maximum-likelihood and penalized-likelihood for image reconstruction with uniform spatial resolution

Contact Info

Product

Resources

About