70 likes | 356 Views
ANN Approach to ECG Classification. Joe Krachey 12/10/2001 ECE 539. Problem:. Classification of irregular heartbeats via MLP neural network. Application: Pre-screening of patients for cardiologists Monitoring/diagnosis equipment Data
E N D
ANN Approach to ECG Classification Joe Krachey 12/10/2001 ECE 539
Problem: • Classification of irregular heartbeats via MLP neural network. • Application: • Pre-screening of patients for cardiologists • Monitoring/diagnosis equipment • Data • Data provided by Massachusetts Institute of Technology and Beth Israel Hospital(MIT/BIH) • Feature vectors provided by Surehka Palreddy • Various data found at physionet.org
DATA Input vector of 9 -RR0, RR1, RR2, RR1/RR0 -Last 5 input vectors are threshold related.
Goals • Determine a ‘optimal’ MLP network for ECG classification • Determine if any input vectors are not needed. • Compare results with AR or KNN neural networks