Improving the accuracy and stability of the speckle tracking technique in processing images obtained in echocardiographic examinations

Abstract

Tracking techniques are widely used in the processing of images in ultrasound cardiological examinations. One of the applications of this method consists in the automatic delineation of the internal and external borders of the cardiac chambers for all frames of the source video. However, owing to significant deformations of the image, which are often observed in practice, a loss of tracking can occur. We propose a new technique to improve the accuracy and stability of the classical speckle tracking algorithm taking into account the special features of the video picture of the ultrasound examination of the heart. The method is based on an assessment of the nature of the movement of regions of the heart wall using the obtained images, prediction of the position of the speckle in the next frame, and using this information to calculate the weighted function of mutual correlation used to determine the exact position of a speckle. Testing of the proposed method on the actual clinical results of echocardiographic examinations has demonstrated its high stability and accuracy, which is sufficient for practical application in ultrasound diagnostics.