The shape of dementia: new measures of morphological complexity in event-related potentials (ERP) and its application to the detection of Alzheimer's disease.

In this paper, we address the problem of quantifying the commonly observed disorganization of the stereotyped wave form of the ERP associated with the P300 component in patients with Alzheimer's disease. To that extent, we propose two new measures of complexity which relate the spectral content of the signal with its temporal waveform: the spectral matching coefficient and the spectral matching entropy. We show by means of experiments that those measures effectively measure complexity and are related to the shape in an intuitive way. Those indexes are compared with commonly used measures of complexity when comparing AD patients against age-matched healthy controls. The results indicate that AD ERP signals are, indeed, more complex in the shape than that of controls, and this result is evidenced mainly by means of our new measures which have a better performance compared to similar ones. Finally, we try to explain this increase in complexity in light of the communication through coherence hypothesis framework, relating commonly found changes in the EEG with our own results.

Unsupervised feature relevance analysis applied to improve ECG heartbeat clustering.

The computer-assisted analysis of biomedical records has become an essential tool in clinical settings. However, current devices provide a growing amount of data that often exceeds the processing capacity of normal computers. As this amount of information rises, new demands for more efficient data extracting methods appear. This paper addresses the task of data mining in physiological records using a feature selection scheme. An unsupervised method based on relevance analysis is described. This scheme uses a least-squares optimization of the input feature matrix in a single iteration. The output of the algorithm is a feature weighting vector. The performance of the method was assessed using a heartbeat clustering test on real ECG records. The quantitative cluster validity measures yielded a correctly classified heartbeat rate of 98.69% (specificity), 85.88% (sensitivity) and 95.04% (general clustering performance), which is even higher than the performance achieved by other similar ECG clustering studies. The number of features was reduced on average from 100 to 18, and the temporal cost was a 43% lower than in previous ECG clustering schemes.

Weighted-PCA for unsupervised classification of cardiac arrhythmias.

A method that improves the feature selection stage for non-supervised analysis of Holter ECG signals is presented. The method corresponds to WPCA approach developed mainly in two stages. First, the weighting of the feature set through a weight vector based on M-inner product as distance measure and a quadratic optimization function. The second one is the linear projection of weighted data using principal components. In the clustering stage, some procedures are considered: estimation of the number of groups, initialization of centroids and grouping by means a soft clustering algorithm. In order to decrease the procedure computational cost, segment analysis, grouping contiguous segments and establishing union and exclusion criteria per each cluster, is carried out. This work is focused to classify cardiac arrhythmias into 5 groups, according to the standard of the AAMI (ANSI/AAMI EC57:1998/ 2003). To validate the method, some recordings from MIT/BIH arrhythmia database are used. By employing the labels of each recording, the performance is assessed with supervised measures (Se = 90.1%, Sp = 98.9% y Cp = 97.4%), enhancing other works in the literature that do not take into account all heartbeat types.

Unsupervised feature selection in cardiac arrhythmias analysis.

The problem of detecting clinical events related to cardiac arrhythmias in long term electrocardiograms is a difficult one due to the large amount of irrelevant information that hides such events. This problem has been addressed in the literature by means of clustering or classification algorithms that create data partitions according to a cost function based on heartbeat features dissimilarity measures. However, studies about the type or number of heartbeat features is lacking. Usually, the feature sets used are relevant but redundant, which degrades algorithm performance. This paper describes a method for automatic selection of heartbeat features. This method is assessed using real signals from the MIT database and common features used in previous works.