Close menu

SURE

Sunderland Repository records the research produced by the University of Sunderland including practice-based research and theses.

A Hybrid Active Contour Segmentation Method for Myocardial D-SPECT Images

Huang, Chenxi, Shan, Xiaoying, Lan, Yisha, Liu, Lu, Cai, Haidong, Che, Wenliang, Hao, Yongtao, Cheng, Yongqiang and Peng, Yonghong (2018) A Hybrid Active Contour Segmentation Method for Myocardial D-SPECT Images. IEEE Access, 6. pp. 39334-39343. ISSN 2169-3536

Item Type: Article

Abstract

The ischaemic heart disease has become one of the leading causes of mortality worldwide. Dynamic single-photon emission computed tomography (D-SPECT) is an advanced routine diagnostic tool commonly used to validate the myocardial function in patients suffering from various heart diseases. Accurate automatic localization and segmentation of myocardial regions is helpful in creating a 3-D myocardial model and assisting clinicians to perform assessments of myocardial function. Thus, image segmentation is a key technology in preclinical cardiac studies. Intensity inhomogeneity is one of the common challenges in image segmentation and is caused by image artifacts and instrument inaccuracy. In this paper, a novel region-based active contour model that can segment the myocardial D-SPECT image accurately is presented. First, a local region-based fitting image is defined based on the information related to the intensity. Second, a likelihood fitting image energy function is built in a local region around each point in a given vector-valued image. Next, the level set method is used to present a global energy function with respect to the neighborhood center. The proposed approach guarantees precision and computational efficiency by combining the region-scalable fitting energy model and local image fitting energy model, and it can solve the issue of high sensitivity to initialization for myocardial D-SPECT segmentation.

Full text not available from this repository.

More Information

Depositing User: Yonghong Peng

Identifiers

Item ID: 10174
Identification Number: https://doi.org/10.1109/ACCESS.2018.2855060
ISSN: 2169-3536
URI: http://sure.sunderland.ac.uk/id/eprint/10174

Users with ORCIDS

Catalogue record

Date Deposited: 20 Nov 2018 15:49
Last Modified: 18 Dec 2019 16:07

Contributors

Author: Chenxi Huang
Author: Xiaoying Shan
Author: Yisha Lan
Author: Lu Liu
Author: Haidong Cai
Author: Wenliang Che
Author: Yongtao Hao
Author: Yongqiang Cheng
Author: Yonghong Peng

University Divisions

Faculty of Technology
Faculty of Technology > FOT Executive

Subjects

Computing > Data Science
Computing > Artificial Intelligence

Actions (login required)

View Item View Item