Evaluation of the Empirical Mode Decomposition Method as a Tool for Preprocessing Ultrasonic Cardiological Data

Abstract

This study investigates the performance of the Empirical Mode Decomposition (EMD) method in application to the ultrasonic cardiological data. Data acquired using self-made phantom are subjected to 1D and 2D EMD algorithms. The 1D EMD yields higher signal to noise ratio improvement than the 2D EMD. In the first case upsampling only along the RF lines is necessary, whereas in the second case upsampling in both directions is required. The first IMF appears best suited for the segmentation and the 1D EMD results in IMF more suitable for segmentation that the 2D EMD.