Automatic cardiac ventricular boundary detection
Echocardiography is a common diagnostic imaging modality for patients with heart diseases. One essential goal in the analysis of echocardiographic images is to identify the locations of the endocardial boundary. This is necessary in order to visualize the structure of the patient's heart, and...
Saved in:
Main Author: | |
---|---|
Other Authors: | |
Format: | Theses and Dissertations |
Published: |
2008
|
Subjects: | |
Online Access: | https://hdl.handle.net/10356/3474 |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Institution: | Nanyang Technological University |
Summary: | Echocardiography is a common diagnostic imaging modality for patients with heart diseases. One essential goal in the analysis of echocardiographic images is to identify the locations of the endocardial boundary. This is necessary in order to visualize the structure of the patient's heart, and to derive quantitative parameters from the images. However, manual identification of endocardial boundary is time-consuming, inconvenient, and dependant on the competence of the clinician.
In this thesis, three new algorithms are developed for boundary detection in two dimensional (2D) echocardiographic images. For long-axis echocardiographic images, a pre-segmentation algorithm is developed to locate the left ventricular (LV) region without user intervention. The final segmentation to find the actual LV boundary in long-axis echocardiographic images is carried out using a Markovian level set method. A new external force for snakes is proposed as dynamic directional gradient vector flow (DDGVF) for LV boundary detection in short-axis echocardiographic images. The techniques developed in this thesis have the potential to be integrated into an accurate and automatic system that could be used in routine clinical practice. |
---|