Automatic Cardiac Self-Gating of Small-Animal PET Data
Name
11307_2015_868_ReferencePDF.pdf
Size
697.76 KB
Format
Adobe PDF
Checksum (MD5)
f83a603ae9fdd221f30032e10ca267df
Author(s)
Herraiz, Joaquin L.
Herranz, Elena
Cal-González, Jacobo
Vaquero, Juan J.
Desco, Manuel
Cussó, Lorena
Udias, Jose M.
Date Issued
June 2015
Journal
Molecular Imaging and Biology
Publisher
Springer US
Citation
Herraiz, Joaquin L., Elena Herranz, Jacobo Cal-González, Juan J. Vaquero, Manuel Desco, Lorena Cussó, and Jose M. Udias. “Automatic Cardiac Self-Gating of Small-Animal PET Data.” Molecular Imaging and Biology 18, no. 1 (June 9, 2015): 109–116.
Version
Author's final manuscript
Abstract
Purpose
The cardiac gating signal (CGS) in positron emission tomography (PET) studies is usually obtained from an electrocardiography (ECG) monitor. In this work, we propose a method to obtain the CGS in small-animal PET using the acquired list-mode data without using any hardware or end-user input.
Procedures
The CGS was obtained from the number of coincidences over time acquired in the lines-of-response connected with the cardiac region. This region is identified in the image as its value changes with frequencies in the range of 3 to 12 Hz. The procedure was tested in a study with 29 Wistar rats and 6 mice injected with 2-deoxy-2-[[superscript 18]F]fluoro-d-glucose, which underwent a 45-min single-bed list-mode PET scan of the heart syncronized with an ECG. The estimated signals and the reconstructed images using eight-gated frames were compared with the ones obtained using the ECG signal from the monitor.
Results
The differences of the PET-based CGS with respect to the ECG relative to the duration of the heartbeats were 5.6 % in rats and 11.0 % in mice. The reconstructed gated images obtained from the proposed method do not differ qualitatively with respect to the ones obtained with the ECG. The quantitative analysis of both set of images were performed measuring the volume of the left ventricle (LV) of the rats in the end-of-systole and end-of-diastole phase. The differences found in these parameters between both methods were below 12.1 % in the ESV and 9.3 % in the EDV with a 95 % confidence interval, which are comparable to the accuracy (7 %) of the method used for segmenting the LV.
Conclusion
The proposed method is able to provide a valid and accurate CGS in small-animal PET list-mode data.
MIT Department
Madrid-MIT M+Vision Consortium
Massachusetts Institute of Technology. Research Laboratory of Electronics
Terms of Use
Article is made available in accordance with the publisher's policy and may be subject to US copyright law. Please refer to the publisher's site for terms of use.
Persistent DSpace Link
DOI of Published Version
https://doi.org/10.1007/s11307-015-0868-y
