Robust and automatic diagnosis of the intraventricular mechanical dyssynchrony for the left ventricle in cardiac magnetic resonance images
- Zhenzhou Wang
- International Journal of Computer Assisted…
BACKGROUND Although there have been recent advances in echocardiography, many studies remain suboptimal due to poor image quality and unclear blood-myocardium border. We developed a novel image processing technique, cardiac variability imaging (CVI), based on the variance of pixel intensity values during passage of ultrasound microbubble contrast into the left ventricle chamber, with the aim of enhancing endocardial border delineation and image quality. METHODS AND RESULTS CVI analysis was performed on simulated data to test and verify the mechanism of image enhancement. Then CVI analysis was applied to echocardiographic images obtained in two different clinical studies, and still images were interpreted by expert reviewers. In the first study (N = 15), using contrast agent EchoGen, the number of observable wall segments in end-diastolic images, for example, was significantly increased by CVI (4.93) as compared to precontrast (3.28) and contrast images (3.36), P < 0.001 for both comparisons to CVI. In the second study (N = 8), using contrast agent Optison, interobserver variability of manually traced end-diastolic volumes was significantly decreased using CVI (22.3 ml) as compared to precontrast (63.4) and contrast images (49.0), P < 0.01 for both comparisons to CVI. CONCLUSION CVI can substantially enhance endocardial border delineation and improve echocardiographic image quality and image interpretation.