Expert System of Ischemia Classification Based on Wavelet MLP

Abstract:

This paper proposes an expert system capable of identifying pathological ECG with signs of ischemia. The system design is based on the knowledge of a team of cardiologists who have been commissioned to identify ECG segments that contain information about the target disease, and subsequently validated the results of the system. The expert system comprises four modules, namely, a pre-processing module which is responsible for improving the SNR, a segmentation module, a DSP module which is responsible for applying the wavelet transform to improve the response of the last module, in charge of the classification. We used a database of about 800 ECG obtained in different clinical and extensively annotated by the team of cardiologists. The system achieves a sensitivity of 87.7 % and a specificity of 82.6 % with the set of ECG testing. © Springer-Verlag Berlin Heidelberg 2014.

Año de publicación:

2014

Keywords:

• ECG ischemia
• classification
• DWT
• MLP
• Expert System

Fuente:

scopus