With increasing use of the electrocardiogram (EGG) as a diagnostic tool in cardiology, there exists a requirement for effective ECG compression techniques. This paper describes such a technique based on artificial neural networks (ANNs), and gives detailed results of pregrouping techniques used to improve the performance of an autoassociative compression network. The ANN algorithms used for grouping are a simple competitive learning network, fuzzy min-max clustering and fuzzy ART. The advantages of using principal components analysis (PCA) prior to grouping are discussed, and the results of this approach applied to a large real world data set are presented.
|Title of host publication||IEEE 17th Annual Conference Engineering in Medicine and Biology Society, 1995|
|Number of pages||8|
|Publication status||Published - 1995|
|Name||IEEE Annual Conference Engineering in Medicine and Biology Society|
Sandham, W. A., Thomson, D. C., & Hamilton, D. J. (1995). Electrocardiogram compression using artificial neural networks. In IEEE 17th Annual Conference Engineering in Medicine and Biology Society, 1995 (pp. 18-25). (IEEE Annual Conference Engineering in Medicine and Biology Society ). IEEE. https://doi.org/10.1109/IEMBS.1995.575066