Please use this identifier to cite or link to this item:
Full metadata record
|dc.guide||Saxena, S. C.||-|
|dc.description.abstract||ECG is a graphic record of the electrical potentials produced by the electro-mechanical activity of the heart. ECG signal and its parameters have to be detected from various time varying or nonstationary physiological signals in a random noise background. The need for ECG data compression arises in order to extract relevant information.from a plethora of cardiographic data. ECG data compression entails the reduction in the number of sample points required to generate an ECG wave without losing its information content. Data compression is required for various reasons which include the economic utilization of storage space for ECG data bases and the transmission of digitized ECGs to an ECG processing and recording establishment via telephone etc. There are various techniques of data compression which are applied in different clinical situations. In this study the "Template method" for data compression has been used. It is a logical method and involves making a template of an ECG wave which then forms the basis of comparison with subsequent data waves ; record being stored only when the deviation is greater than a certain limit say ten percent. The "Moving Sign Slope Method" which is a straight line approximation of an ECG wave has also been discussed. Data compression here is achieved by approximating the ECG wave by straight lines within a corridor which is defined by the maximum allowable limit. The results show that the Template method yields the greatest advantage in a clinical situation where the ECG wave is not deviating significantly from its initial characteristics, whereas the Moving Sign Slope Method, being a statistical method of data compression results in large data compression with good t'ave reconstruction fidelity only when the allowable error limit is one percent.||en_US|
|dc.title||DATA COMPRESSION OF ELECTROCARDIOGRAM||en_US|
|Appears in Collections:||MASTERS' DISSERTATIONS (Electrical Engg)|
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.