Accurate estimation of the Fisher information matrix for the PET image reconstruction problem

Quanzheng Li; Evren Asma; Jinyi Qi; James R. Bading; Richard M. Leahy

Accurate estimation of the Fisher information matrix for the PET image reconstruction problem

Quanzheng Li, Evren Asma, Jinyi Qi, James R. Bading, Richard M. Leahy

Biomedical Engineering

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Abstract

The Fisher Information Matrix (FIM) plays a key role in the analysis and applications of statistical image recon-struction methods based on Poisson data models. The elements of the FIM are a function of the reciprocal of the mean values of sinogram elements. Conventional plug-in FIM estimation methods do not work well at low counts, where the FIM estimate is highly sensitive to the reciprocal mean estimates at individual detector pairs. A generalized error look-up table (GELT) method is developed to estimate the reciprocal of the mean of the sinogram data, which achieves a bias of less than 2% for mean sinogram values greater than 0.7. This approach is also extended to randoms precorrected data. Based on these techniques, an accurate FIM estimate is obtained for both Poisson and randoms precorrected data. As an application, the new GELT method is used to improve resolution uniformity and achieve near-uniform reconstructed image resolution in both high and low count situations.

Original language	English (US)
Title of host publication	IEEE Nuclear Science Symposium Conference Record
Editors	S.D. Metzler
Pages	2012-2016
Number of pages	5
Volume	3
State	Published - 2003
Event	2003 IEEE Nuclear Science Symposium Conference Record - Nuclear Science Symposium, Medical Imaging Conference - Portland, OR, United States Duration: Oct 19 2003 → Oct 25 2003

Other

Other	2003 IEEE Nuclear Science Symposium Conference Record - Nuclear Science Symposium, Medical Imaging Conference
Country	United States
City	Portland, OR
Period	10/19/03 → 10/25/03

Keywords

Fisher Information Matrix
Hyperparameters
Uniform Resolution

ASJC Scopus subject areas

Computer Vision and Pattern Recognition
Industrial and Manufacturing Engineering

Access to Document

Fingerprint Dive into the research topics of 'Accurate estimation of the Fisher information matrix for the PET image reconstruction problem'. Together they form a unique fingerprint.

Cite this

Li, Q., Asma, E., Qi, J., Bading, J. R., & Leahy, R. M. (2003). Accurate estimation of the Fisher information matrix for the PET image reconstruction problem. In S. D. Metzler (Ed.), IEEE Nuclear Science Symposium Conference Record (Vol. 3, pp. 2012-2016). [M5-3]

Li, Q, Asma, E, Qi, J, Bading, JR & Leahy, RM 2003, Accurate estimation of the Fisher information matrix for the PET image reconstruction problem. in SD Metzler (ed.), IEEE Nuclear Science Symposium Conference Record. vol. 3, M5-3, pp. 2012-2016, 2003 IEEE Nuclear Science Symposium Conference Record - Nuclear Science Symposium, Medical Imaging Conference, Portland, OR, United States, 10/19/03.

@inproceedings{040cf6e83b6d4101b9ec4f9492300da3,

title = "Accurate estimation of the Fisher information matrix for the PET image reconstruction problem",

abstract = "The Fisher Information Matrix (FIM) plays a key role in the analysis and applications of statistical image recon-struction methods based on Poisson data models. The elements of the FIM are a function of the reciprocal of the mean values of sinogram elements. Conventional plug-in FIM estimation methods do not work well at low counts, where the FIM estimate is highly sensitive to the reciprocal mean estimates at individual detector pairs. A generalized error look-up table (GELT) method is developed to estimate the reciprocal of the mean of the sinogram data, which achieves a bias of less than 2% for mean sinogram values greater than 0.7. This approach is also extended to randoms precorrected data. Based on these techniques, an accurate FIM estimate is obtained for both Poisson and randoms precorrected data. As an application, the new GELT method is used to improve resolution uniformity and achieve near-uniform reconstructed image resolution in both high and low count situations.",

keywords = "Fisher Information Matrix, Hyperparameters, Uniform Resolution",

author = "Quanzheng Li and Evren Asma and Jinyi Qi and Bading, {James R.} and Leahy, {Richard M.}",

year = "2003",

language = "English (US)",

volume = "3",

pages = "2012--2016",

editor = "S.D. Metzler",

booktitle = "IEEE Nuclear Science Symposium Conference Record",

note = "2003 IEEE Nuclear Science Symposium Conference Record - Nuclear Science Symposium, Medical Imaging Conference ; Conference date: 19-10-2003 Through 25-10-2003",

}

TY - GEN

T1 - Accurate estimation of the Fisher information matrix for the PET image reconstruction problem

AU - Li, Quanzheng

AU - Asma, Evren

AU - Qi, Jinyi

AU - Bading, James R.

AU - Leahy, Richard M.

PY - 2003

Y1 - 2003

N2 - The Fisher Information Matrix (FIM) plays a key role in the analysis and applications of statistical image recon-struction methods based on Poisson data models. The elements of the FIM are a function of the reciprocal of the mean values of sinogram elements. Conventional plug-in FIM estimation methods do not work well at low counts, where the FIM estimate is highly sensitive to the reciprocal mean estimates at individual detector pairs. A generalized error look-up table (GELT) method is developed to estimate the reciprocal of the mean of the sinogram data, which achieves a bias of less than 2% for mean sinogram values greater than 0.7. This approach is also extended to randoms precorrected data. Based on these techniques, an accurate FIM estimate is obtained for both Poisson and randoms precorrected data. As an application, the new GELT method is used to improve resolution uniformity and achieve near-uniform reconstructed image resolution in both high and low count situations.

AB - The Fisher Information Matrix (FIM) plays a key role in the analysis and applications of statistical image recon-struction methods based on Poisson data models. The elements of the FIM are a function of the reciprocal of the mean values of sinogram elements. Conventional plug-in FIM estimation methods do not work well at low counts, where the FIM estimate is highly sensitive to the reciprocal mean estimates at individual detector pairs. A generalized error look-up table (GELT) method is developed to estimate the reciprocal of the mean of the sinogram data, which achieves a bias of less than 2% for mean sinogram values greater than 0.7. This approach is also extended to randoms precorrected data. Based on these techniques, an accurate FIM estimate is obtained for both Poisson and randoms precorrected data. As an application, the new GELT method is used to improve resolution uniformity and achieve near-uniform reconstructed image resolution in both high and low count situations.

KW - Fisher Information Matrix

KW - Hyperparameters

KW - Uniform Resolution

UR - http://www.scopus.com/inward/record.url?scp=11844282745&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=11844282745&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:11844282745

VL - 3

SP - 2012

EP - 2016

BT - IEEE Nuclear Science Symposium Conference Record

A2 - Metzler, S.D.

T2 - 2003 IEEE Nuclear Science Symposium Conference Record - Nuclear Science Symposium, Medical Imaging Conference

Y2 - 19 October 2003 through 25 October 2003

ER -