Reliability of old and new ventricular fibrillation detection algorithms for automated external defibrillators

• Published in: Biomedical engineering online Vol. 4; no. 1; p. 60
• Main Authors: Amann, Anton, Tratnig, Robert, Unterkofler, Karl
• Format: Journal Article
• Language: English
• Published: England BioMed Central 27.10.2005; BMC
• Subjects: Algorithms; Artificial Intelligence; automated external defibrillator (AED); Computer Systems; Defibrillators; Diagnosis, Computer-Assisted - methods; ECG analysis; Electrocardiography - instrumentation; Electrocardiography - methods; Equipment Design; Equipment Failure Analysis; Heart Rate; Humans; Reproducibility of Results; Retrospective Studies; Sensitivity and Specificity; sinus rhythm (SR); Therapy, Computer-Assisted - methods; ventricular fibrillation (VF); Ventricular Fibrillation - diagnosis; ventricular fibrillation detection
• ISSN: 1475-925X; 1475-925X
• DOI: 10.1186/1475-925X-4-60

A pivotal component in automated external defibrillators (AEDs) is the detection of ventricular fibrillation by means of appropriate detection algorithms. In scientific literature there exists a wide variety of methods and ideas for handling this task. These algorithms should have a high detection quality, be easily implementable, and work in real time in an AED. Testing of these algorithms should be done by using a large amount of annotated data under equal conditions. For our investigation we simulated a continuous analysis by selecting the data in steps of one second without any preselection. We used the complete BIH-MIT arrhythmia database, the CU database, and the files 7001-8210 of the AHA database. All algorithms were tested under equal conditions. For 5 well-known standard and 5 new ventricular fibrillation detection algorithms we calculated the sensitivity, specificity, and the area under their receiver operating characteristic. In addition, two QRS detection algorithms were included. These results are based on approximately 330,000 decisions (per algorithm). Our values for sensitivity and specificity differ from earlier investigations since we used no preselection. The best algorithm is a new one, presented here for the first time.