Performance evaluation and choice criteria of data compression algorithms by extensive test on CSE database

To solve some problems of data storage and transmission related to the systems for automatic interpretation of the electrocardiogram (ECG) various classes of reversible and irreversible compression algorithms have been tested on the database produced by CSE. Errors of the reconstructed signal have been evaluated by means of PRD, RMS and other synthetic parameters together with the visual analysis by a cardiologist. Tracing rebuilt from the compressed signal has been analyzed using the MEANS program of automatic interpretation, and diagnostic answers with the related measurements have been compared with the results obtained on the original signal. Inconsistent issues have been reexamined by a cardiologist to determine the artifacts and the type and amount of information lost because of compression. The results define the limits of the different classes of compression methods in relation to the accuracy expected in the diagnosis