A neural based comparative analysis for feature extraction from ECG signals

Automated ECG analysis and classification are nowadays a fundamental tool formonitoring patient heart activity properly. The most important features used inliterature are the raw data of a time window, the temporal attributes and thefrequency information from the eigenvector techniques. This  paper comparesthese approaches from a topological point of view, by using linear and non-linearprojections and a neural network for assessing the corresponding classi?cationquality.The non-linearity of the feature data manifold carries most of the QRS-complex information. Indeed, it yields high rates of classi?cation with thesmallest number of features. This is most evident if temporal features are used:non-linear dimensionality reduction techniques allow a very large datacompression at the expense of a slight loss of accuracy. It can be an advantagein applications where the computing time is a critical factor. If, instead, theclassification is performed online, the raw data technique is the best one.


ERC Sector:

  • PE7_7 Signal processing
  • PE7_11 Components and systems for applications
  • PE6_11 Machine learning, statistical data processing and applications using signal processing (e.g. speech, image, video)


  • Clustering methods
  • Multilayer perceptrons
  • Electrocardiogram
  • Feature extraction

Gruppi di ricerca

Persone di riferimento