In this assignment you will investigate various ways of analysing ECG data taken from on an online database.
- Using data from ftp://ftp.ieee.org/uploads/press/rangayyan, files eeg3.dat, eeg4.dat, or eeg5.dat, (sampled at 200 Hz) implement the Pan-Tompkins method (with a simple, fixed threshold) to compute the duration of individual beats.
- Using data from the PhysioBank data base, specifically the MIT-BIH Noise Stress Test Database, download data using the ATM to test how reliable your Pan-Tompkins method is in the presence of noise. The physioBank records have accurate beat information for these noisy signals. Does your Pan-Tompkins method yield the same average beat rate? At what noise level does it break down? The PhysioBank ATM will produce
*.mat files which you can load directly into Matlab. You also need to get the sample rate from the corresponding
*.info file. You will need to download hour-long data sets because the noise alternates at 2 minute intervals.
- The program ECG_analysis_AP_gen.m produces a simulated biological signal (action potentials) corrupted with 60 Hz and thermal noise. The simulated noisy signal has two amplitudes of action potentials which you need to measure as accurately as possible. Write a Least Mean Squares adaptive filter to remove the 60 Hz, combined with a 4 pole butterworth to remove thermal noise, while not distorting the amplitude of the action potentials by more than 5%. Plot the adpative filter response verus time to show that it is converging. Plot the histogram of the action potential amplitude and compare the means to the uncorrupted signal.
Your written lab report should include the sections mentioned in the policy page, and :
- The questions and requirements of the three assignment sections.
June 2009 . Copyright Cornell university