Método para detecção automática de inversão de eletrodos em eletrocardiogramas

Abstract

Electrocardiogram (ECG) are widely used in clinical practice because of its inherent simplicity and high medical relevance for diagnosing pathologies. The occurrence of electrode inversion in an electrocardiogram (ECG) examination can lead to errors or impossibility of diagnosis. This problem has a greater impact in the Telemedicine scenario, where the distance between the acquisition phase and the information analysis is a critical factor, in this context, an examination presenting nonsignificant information leads to financial, operational and social losses related to the need to reexamine the patient. This work introduces a method for automatic detection of electrodes inversion in ECGs, based on the quantification of the asymmetric of the ECG signal probability distribution. The algorithms performance was evaluated through statistical analysis using concepts of sensitivity, specificity and errors of false positive and false negative. The method was tested using a database provided by the Telemedicine Network of Minas Gerais containing 6967 ECGs, previously classified as discarded by interference, discarded by electrodes inversion, healthy and pathological. The proposed algorithm presents a high detection rate (95%) while minimizing the number of false positives (<10%), this values are competitive when compared with classical methods. In addition, the algorithm has low computational cost, which allows its practical application in embedded hardware.