Share Email Print

Proceedings Paper

3D deformable image processing and integration for neuroanatomical analysis in FDOPA-PET studies
Author(s): Hong-Dun Lin; Kang-Ping Lin; Ren-Shyan Liu
Format Member Price Non-Member Price
PDF $14.40 $18.00

Paper Abstract

Functional FDOPA-PET imaging, which displays aspects of tissue function but a poorly delineate anatomy can be used to investigate physiological activities of tissues for further quantitative analysis, is routinely used for neuroanatomic analysis of brain disorders. To exactly distinguish and define abnormal tissue regions from high noise FDOPA-PET image is an important issue of improving the diagnosis accuracy. In this study, a quantitative analysis based feature extraction and a deformable image processing methods are provided to construct a 3D brain FDOPA-PET image model. The novel feature extraction method based on physiological analysis of kinetic parameters, which generated from kinetic modeling analysis of dynamic FDOPA-PET studies, segments desired brain tissue regions, including striatum, gray and white matters. The proposed 3D multi-resolution optical flow estimation method (OFEM) integrates various normal FDOPA-PET studies and constructs the 3D brain image model. The errors of region difference and tissue physiological curves between the segmented result and the VOI in the striatum tissue perform less than 3% in average, respectively, and the constructed 3D FDOPA-PET model is also applied as a standard template for clinical use. With respect to the experiments, 25 Parkinson’s disease studies are examined to perform the accuracy of proposed method. The results show the constructed FDOPA-PET model is effectively used to investigate and define the abnormal regions in brain. In summary, the developed feature extraction technique can exactly segment important tissue in FDOPA-PET images, and the constructed 3D image model can clearly define brain structure and improve clinical diagnosis accuracy simultaneously.

Paper Details

Date Published:
PDF: 12 pages
Proc. SPIE 6143, Medical Imaging 2006: Physiology, Function, and Structure from Medical Images, 61431O; doi: 10.1117/12.652809
Show Author Affiliations
Hong-Dun Lin, Chung-Yuan Univ. (Taiwan)
BMEC, Industrial Technology Research Institute (Taiwan)
Kang-Ping Lin, Chung-Yuan Univ. (Taiwan)
Ren-Shyan Liu, Taipei Veterans General Hospital (Taiwan)

Published in SPIE Proceedings Vol. 6143:
Medical Imaging 2006: Physiology, Function, and Structure from Medical Images
Armando Manduca; Amir A. Amini, Editor(s)

© SPIE. Terms of Use
Back to Top