ANALYSIS OF BIOMEDICAL SIGNALS PROCESSING TECHNIQUES

Abstract

The biomedical signal is a summarizing term for all kinds of signals that can be continu ally measured and monitored from biological beings. Electrocardiogram (ECG) and Elec troencephalogram (EEG) are most important 1-D biomedical signals as they are linked with activities of heart and brain respectively, the most important organs of human body. Magnetic Resonance Imaging (MRI) is the most popular medical imaging tech nique. It has a wide range of applications in medical diagnosis and it is preferred over other methods for medical imaging purpose for the reason that it does not involve any ionizing radiation. Importance of MRI can be understood with the fact that over 50,000 MRI scanners are estimated to be in use worldwide for biomedical imaging purpose. Acquisition of a biomedical signal is not sufficient but it is required to process the acquired signal to get the relevant information “buried” in it. This may be due to the fact that the signal is affected by noise during signal acquisition and thus must be “cleaned” using some signal processing technique or method to minimize effects of noise and to en hance useful information. There are different types of noises or artifacts in biomedical signals. Baseline wander and ocular artifacts are the most important artifacts in case of ECG and EEG respectively. This research is mainly focused on proposing novel methods for removal of base line wander and ocular artifact from ECG and EEG. A new method is proposed for base line wander artifact denoising from ECG using a cascaded combination of Complete En semble Empirical Mode Decomposition (CEEMD) and Morphological functions with adaptive Structure Elements (SEs). The proposed method maintains morphology of ECG during denoising and denoising performance is independent of heart rate in case of stress ECG. ABSTRACT

iii A new method is proposed for ocular artifact removal from EEG using Stationary Wave let Enhanced Independent Component Analysis (ICA) with a novel threshold technique. The proposed method preserves morphological information present in EEG and the novel threshold technique makes denoising more efficient. MR image is the most important 2-D biomedical signal (Biomedical Image) and segmentation is one of the most important steps of MRI denoising and classification. A novel fuzzy energy based level set method is proposed in this research work for segmen tation of MR images. Proposed method deals effectively and simultaneously with intensi ty inhomogeneity and noise problems of medical image by integrating active contour with Fuzzy C-Means (FCM) clustering. Denoising of MR images is further enhanced by using a mean filter based spatial term with proposed FCM based energy function. Performance of proposed methods is tested with various publicly available da tasets and compared with earlier state-of-the-art methods.