Independent component analysis in biomedical applications and accelerometer data logging system

In biomedical applications, the obtained biomedical data must be cleared from all undesired artifacts such as environmental and sensor noises prior to signal processing. If the frequency properties of artifacts are known, various appropriate filtering techniques can be used to remove them. However, if the information about artifact sources is limited or not exists, it is required to utilize various statistical methods to remove the artifacts. Independent Component Analysis (ICA) as the most known Blind Source Separation method can be used for filtering or detecting sources.In this thesis study, ICA, ICA methods and the realization problems have been introduced and ICA has been applied to various biomedical applications. For ICA applications, two different novel data logging systems have been developed: simultaneous multi channel sound data logging system and simultaneous multi point multi dimensional accelerometer data logging system. The sound data logging system was used for general ICA applications and the ICA realization problems were examined. The accelerometer data logging system was used to collect knee vibroarthrographic (VAG) signals and for diagnosing knee problems. The obtained results were presented.Moreover, ICA was applied to identification of artifact sources and artifact removing problems. Heart pulses and sweating artifacts in EEG recordings were identified and removed by ICA. Furthermore, the accelerometer data logging system has been used for filtering jaw motion artifacts from the EEG recordings.