Share Email Print

Proceedings Paper

Semi-automated measurements of heart-to-mediastinum ratio on 123I-MIBG myocardial scintigrams by using image fusion method with chest X-ray images
Author(s): Ryosuke Kawai; Takeshi Hara; Tetsuro Katafuchi; Tadahiko Ishihara ; Xiangrong Zhou; Chisako Muramatsu; Yoshiteru Abe; Hiroshi Fujita
Format Member Price Non-Member Price
PDF $17.00 $21.00

Paper Abstract

MIBG (iodine-123-meta-iodobenzylguanidine) is a radioactive medicine that is used to help diagnose not only myocardial diseases but also Parkinson’s diseases (PD) and dementia with Lewy Bodies (DLB). The difficulty of the segmentation around the myocardium often reduces the consistency of measurement results. One of the most common measurement methods is the ratio of the uptake values of the heart to mediastinum (H/M). This ratio will be a stable independent of the operators when the uptake value in the myocardium region is clearly higher than that in background, however, it will be unreliable indices when the myocardium region is unclear because of the low uptake values. This study aims to develop a new measurement method by using the image fusion of three modalities of MIBG scintigrams, 201-Tl scintigrams, and chest radiograms, to increase the reliability of the H/M measurement results. Our automated method consists of the following steps: (1) construct left ventricular (LV) map from 201-Tl myocardium image database, (2) determine heart region in chest radiograms, (3) determine mediastinum region in chest radiograms, (4) perform image fusion of chest radiograms and MIBG scintigrams, and 5) perform H/M measurements on MIBG scintigrams by using the locations of heart and mediastinum determined on the chest radiograms. We collected 165 cases with 201-Tl scintigrams and chest radiograms to construct the LV map. Another 65 cases with MIBG scintigrams and chest radiograms were also collected for the measurements. Four radiological technologists (RTs) manually measured the H/M in the MIBG images. We compared the four RTs’ results with our computer outputs by using Pearson’s correlation, the Bland-Altman method, and the equivalency test method. As a result, the correlations of the H/M between four the RTs and the computer were 0.85 to 0.88. We confirmed systematic errors between the four RTs and the computer as well as among the four RTs. The variation range of the H/M among the four RTs was obtained as 0.22 based on the equivalency test method. The computer outputs were existed within this range. We concluded that our image fusion method could measure equivalent values between the system and the RTs.

Paper Details

Date Published: 20 March 2015
PDF: 6 pages
Proc. SPIE 9414, Medical Imaging 2015: Computer-Aided Diagnosis, 941433 (20 March 2015); doi: 10.1117/12.2081630
Show Author Affiliations
Ryosuke Kawai, Gifu Univ. School of Medicine (Japan)
Takeshi Hara, Gifu Univ. School of Medicine (Japan)
Tetsuro Katafuchi, Gifu Univ. Univ. of Medical Science (Japan)
Tadahiko Ishihara , Gifu Univ. Hospital (Japan)
Xiangrong Zhou, Gifu Univ. School of Medicine (Japan)
Chisako Muramatsu, Gifu Univ. School of Medicine (Japan)
Yoshiteru Abe, Yaizu City Hospital (Japan)
Hiroshi Fujita, Gifu Univ. School of Medicine (Japan)

Published in SPIE Proceedings Vol. 9414:
Medical Imaging 2015: Computer-Aided Diagnosis
Lubomir M. Hadjiiski; Georgia D. Tourassi, Editor(s)

© SPIE. Terms of Use
Back to Top
Sign in to read the full article
Create a free SPIE account to get access to
premium articles and original research
Forgot your username?