Analysis of ECG signals and its application in prediction of sudden cardiac death
MetadataShow full item record
An ECG signal can provide a lot of information regarding the functioning of a heart by analyzing the different waves and the intervals present in it. The ability of a detection algorithm to retrieve these components is thus a key factor. This thesis aims at constructing a robust algorithm for accurate detection of the QRS complex. The Pan-Tompkins algorithm is used as the base and modified to suit our data. The changes include a searchback mechanism [Searchback (v1.1)] that we developed to overcome the problems due to high noise presence and a technique to make it computationally efficient. The Searchback (v1.1) mechanism's main objective is to reduce the threshold level for detection that was primarily increased by noise. The entire data is divided into large chunks of blocks which is further split into a number of smaller components called segments. We demonstrate that this method can achieve the same performance as the modified algorithm while significantly reducing the computational load. We document the strategies that work well with our data, and show orders of magnitude speedup using the modified Pan-Tompkins detector. The analyzed data is used to define biological markers that can predict sudden cardiac death. We briefly make an attempt to analyze one such marker, the QT-RR ratio. Further, we try to contrast this ratio obtained for swine that died due to cardiac arrest with the ones that survived. Going into depth of finding such markers will be our next step. Keywords: SCD, Pan-Tompkins, searchback, QT-RR.