A Study on Dictionary Selection in Compressive Sensing for ECG Signals Compression and Classification.

The paper proposes a comparative analysis of the projection matrices and dictionaries used for compressive sensing (CS) of electrocardiographic signals (ECG), highlighting the compromises between the complexity of preprocessing and the accuracy of reconstruction. Starting from the basic notions of CS theory, this paper proposes the construction of dictionaries (constructed directly by cardiac patterns with R-waves, centered or not-centered) specific to the application and the results of their testing. Several types of projection matrices are also analyzed and discussed. The reconstructed signals are analyzed quantitatively and qualitatively by standard distortion measures and by the classification of the reconstructed signals. We used a k-nearest neighbors (KNN) classifier to evaluate the reconstructed models. The KNN module was trained with the models from the mega-dictionary used in the classification block and tested with the models reconstructed with class-specific dictionaries. In addition to the KNN classifier, a neural network was used to test the reconstructed signals. The neural network was a multilayer perceptron (MLP). Moreover, the results are compared with those obtained with other compression methods, and ours proved to be superior.

On the Classification of ECG and EEG Signals with Various Degrees of Dimensionality Reduction.

Classification performances for some classes of electrocardiographic (ECG) and electroencephalographic (EEG) signals processed to dimensionality reduction with different degrees are investigated. Results got with various classification methods are given and discussed. So far we investigated three techniques for reducing dimensionality: Laplacian eigenmaps (LE), locality preserving projections (LPP) and compressed sensing (CS). The first two methods are related to manifold learning while the third addresses signal acquisition and reconstruction from random projections under the supposition of signal sparsity. Our aim is to evaluate the benefits and drawbacks of various methods and to find to what extent they can be considered remarkable. The assessment of the effect of dimensionality decrease was made by considering the classification rates for the processed biosignals in the new spaces. Besides, the classification accuracies of the initial input data were evaluated with respect to the corresponding accuracies in the new spaces using different classifiers.

A computerized data analysis method for electrogastrographic signals.

UNLABELLED: Many recent studies showed that EGG signal can be used to diagnose and monitor patients with gastric motility disorders. MATERIAL AND METHOD: This paper presents the way to record and process the EGG signal, and the results of study on a sample of 10 patients. These patients are initially evaluated by upper endoscopy were are normal without gastric motility dysfunction or some gastric lesions. The EGG signals are obtained with the surface electrodes placed along the projection of the stomach axis on the abdomen, amplified with an electrogastrogram amplifier module and analyzed on a personal computer. The used signal analysis method is based on spectral analysis of EGG signals. RESULTS: This method showed a normal gastric activity for seven patients of the sample and a gastric dysryhthmia for three of them. CONCLUSION: The computerized spectral analysis of electrogastrography signals proves to be a non-invasive, high sensitive, reproducible and cost-effective method for the diagnosis of gastric dysrhythmias and normal gastric electrical rhythm even in sub clinical patients.