Detection of ventricular fibrillation by sequential hypothesis testing of binary sequences

A method is presented for the detection of ventricular fibrillation using binary sequences derived from the surface electrocardiogram. The binary sequences are used to obtain threshold crossing interval and Lempel-Ziv complexity measurements which together form the inputs to a neural network classifier. It is shown that the method outperforms the sequential hypothesis testing of either measurement on the MIT, AHA and CU databases.