BACKGROUND: In single photon-emission computed tomographic imaging of the chest, nonuniform attenuation correction requires use of a patient-specific attenuation map. The aim of this study was to determine whether an estimate of the regions of the lungs and nonpulmonary tissues of the chest could be obtained by segmenting the photopeak and Compton scatter window images in a phantom and in patients to estimate patient-specific attenuation maps. METHODS AND RESULTS: The photopeak and scatter window slices from 16 consecutive 99mTc-labeled sestamibi perfusion studies were segmented interactively. In these studies, visually reasonable regions could be obtained by estimating a "cold" lung region from scatter window data with additional anatomic information of the myocardium region, the backbone and sternum locations, the liver, and the rib cage from the photopeak window data. In an anthropomorphic torso phantom study and a patient study, comparison was made between the attenuation maps based on segmentation of the emission images and transmission imaging with a slant-hole collimator. It was determined that good agreement in the estimation of the body regions can be achieved with segmentation of the emission images in both the phantom and patient data. Attenuation correction using the maximum-likelihood expectation maximization method was performed on the phantom and the patient data. In both studies, attenuation correction with the segmented attenuation map improved uniformity of the inferior wall region in comparison with the other walls. CONCLUSIONS: The estimation of patient-specific attenuation maps by segmenting the scatter and photopeak window slices of 99mTc-labeled sestamibi studies may be a way of reducing the loss of specificity due to attenuation artifacts. The potential limitations on the accuracy of correction inherent in the method due to the estimation of the regions and assignment of the attenuation coefficients need to be determined further, and the method needs to be further automated before it can be considered for routine clinical use.
BACKGROUND: In single photon-emission computed tomographic imaging of the chest, nonuniform attenuation correction requires use of a patient-specific attenuation map. The aim of this study was to determine whether an estimate of the regions of the lungs and nonpulmonary tissues of the chest could be obtained by segmenting the photopeak and Compton scatter window images in a phantom and in patients to estimate patient-specific attenuation maps. METHODS AND RESULTS: The photopeak and scatter window slices from 16 consecutive 99mTc-labeled sestamibi perfusion studies were segmented interactively. In these studies, visually reasonable regions could be obtained by estimating a "cold" lung region from scatter window data with additional anatomic information of the myocardium region, the backbone and sternum locations, the liver, and the rib cage from the photopeak window data. In an anthropomorphic torso phantom study and a patient study, comparison was made between the attenuation maps based on segmentation of the emission images and transmission imaging with a slant-hole collimator. It was determined that good agreement in the estimation of the body regions can be achieved with segmentation of the emission images in both the phantom and patient data. Attenuation correction using the maximum-likelihood expectation maximization method was performed on the phantom and the patient data. In both studies, attenuation correction with the segmented attenuation map improved uniformity of the inferior wall region in comparison with the other walls. CONCLUSIONS: The estimation of patient-specific attenuation maps by segmenting the scatter and photopeak window slices of 99mTc-labeled sestamibi studies may be a way of reducing the loss of specificity due to attenuation artifacts. The potential limitations on the accuracy of correction inherent in the method due to the estimation of the regions and assignment of the attenuation coefficients need to be determined further, and the method needs to be further automated before it can be considered for routine clinical use.
Authors: Ashequr Rahman; Yansong Zhu; Eric Clarkson; Matthew A Kupinski; Eric C Frey; Abhinav K Jha Journal: Inverse Probl Date: 2020-08-20 Impact factor: 2.407