A Review of Notable Studies on Using Empirical Mode Decomposition for Biomedical Signal and Image Processing

Abstract

The data-driven empirical mode decomposition (EMD) method is designed to analyze the non-stationary signals like biomedical signals originating from nonlinear biological systems. EMD analysis produces a local complete separation of the input signal in fast and slow oscillations along with the time-frequency localization. EMD extracts the amplitude and frequency modulated (AM–FM) functions, i.e. the intrinsic mode functions (IMFs), that have been widely used for biomedical signal de-noising, de-trending, feature extraction, compression, and identification. To overcome the problems of EMD, like mode mixing, new generations of EMD have been proposed and applied for biomedical signal analysis. Besides, the bidimensional EMD (BEMD) was introduced and improved for image processing. BEMD and its modified versions have been widely used for medical image de-noising, de-speckling, segmentation, registration, fusion, compression, and classification. In this paper, a review of notable studies in the biomedical signal and image processing based on EMD or BEMD method and their modified versions were considered. The studies on using EMD and its modified versions for mono-dimensional and bidimensional(image) signal processing showed the capabilities of the improved EMD and BEMD methods on biomedical signal and image processing.