Pre-Processing Edge Detection Image Enhancement and Spatial Object Velocity Estimation for Echocardiograph Diagnostic Test

Echocardiograph imaging is a primary modality in the diagnosis of heart disease. Compared to other imaging techniques, such as X-Ray, MRI, and PET, echocardiograph imaging owes its great popularity to the fact that it is a safe and non-invasive procedure for visualizing the heart and vasculature. The echocardiograph image however is corrupted by speckle noise and low contrast, which make feature detection and tracking difficult. This thesis focuses on two important issues for the clinical applications of medical echocardiograph images: speckle suppression and motion estimation. The thesis first presents visualization enhancement method to clarify the heart structure and the movement of the valves. This method is designed to extract the contours of heart boundaries from a sequence of echocardiograph images, where it started with pre-processing to reduce noise and get better image quality. These pre-processing operations involved the use of median filtering, morphological opening and contrast adjustment. Thereafter, Sobel edge detection was applied and the resulted image combined with image after opening stage. This method validated visually by medical students using real echocardiograph images. Performance improvement of this method evaluated as it provides very significant speckle suppression and edge enhancement for the purposes of visualization and automatic structure detection. Second issue in this thesis is improving the detection of wall motion abnormality by quantitative analysis. The analysis of the left ventricular wall motion in routine is mostly based on visual interpretation of echocardiograph image. The interpretation of these images is widely dependent on operator training and is subject to large variability. To reduce this inter and intraobservers variability, the utilization of optical flow technique presented to estimate the left ventricular wall motion and create a cardiac motion profile based on the anatomical structure of the left ventricular wall in cross sectional view. This profile presenting the anatomical structure provides an additional means for functional imaging and eliminates the need to build a large dataset containing specific parameters of the patient to obtain an accurate diagnosis. In addition, the segmentation into three parts corresponding to the three major coronary arteries was meaningful for cardiac surgery. As a result, this method achieves an estimation of regional myocardial function with percentage of validations 71.4%, which is encouraging. In conclusion, estimation of regional myocardial deformation from intracardiac echocardiography by depending on anatomical knowledge is feasible.This work could be an important aid to improve and support diagnostic accuracy and the prognostic method for left ventricular diseases.

PDF FSKTM_2010_6_F.pdf Download (160kB)

Actions (login required)

View Item