BIOE7902 Event Detection

BIOE7902 – Assignment 2 – Event Detection 
 Comparison of Three QRS Detection Algorithms Over a Public Database 
 
Tasks: 
1) Implement the 3 algorithms from the paper in Matlab: 
a. Pan & Tompkins algorithm 
b. Hamilton & Tompkins algorithm 
c. Phasor transform algorithm 
2) Run these algorithms on all datasets from the MIT-BIH database: 
https://archive.physionet.org/physiobank/database/mitdb/ 
3) Reproduce Table 2 from the paper. 
4) Reproduce Figure 2 from the paper. 
5) Implement an algorithm of your choice that is even better than the 3 
implemented and compared in the paper E.g. 
https://ieeexplore.ieee.org/abstract/document/898536 
6) why did the Phasor transform algorithm not perform as well in this paper as in 
the original publication? 
 
Submission: 
Submit one Matlab file (either .m or .mlx) which reproduces Table 2 and Figure 2 
from the paper starting from the raw files. Name this file main_analysis.m or 
main_analysis.mlx. You should create helper functions as separate m-files and 
upload all these files as one zip file. Do NOT submit the datafiles from the MITBIH
 database - your code should expect that these files are one level up from 
your code in your file hierarchy. Make sure to comment your code. 
 
Hints: 
1) The code that was originally used for the paper is not available anymore under 
the address listed in the paper but can be found here: 
https://github.com/milegroup/Recg 
a. This code is written in R, so it will not be directly usable in your Matlab 
scripts, but could ofer hints about implementation details 
 
 2) One way to load these files in Matlab is to use: 
https://archive.physionet.org/physiotools/matlab/wfdb-app-matlab/ 
 
This is how you install the toolbox: 
 
[old_path]=which('rdsamp'); if(~isempty(old_path)) rmpath(old_path(1:end-8)); end 
 
wfdb_url='https://physionet.org/physiotools/matlab/wfdb-app-matlab/wfdb-apptoolbox-0-10-0.zip';

 [filestr,status] = urlwrite(wfdb_url,'wfdb-app-toolbox-0-10-0.zip'); 
 
unzip('wfdb-app-toolbox-0-10-0.zip'); 
 
cd mcode 
 
addpath(pwd) 
 
savepath 
 
Here is an example how to load a dataset and the R-peak annotations (you need these 
to compute if your algorithm does the right thing to compute sensitivity and Pos. 
Predictivity): 
 
[sig, Fs, tm] = rdsamp('234', 1); 
 
[ann,anntype,subtype,chan,num,comments]=rdann('234', 'atr', [], [], [], []); 
 
[RR,tms]=ann2rr('234','atr',[],[],[]) 
 
figure(1);clf 
 
plot(tm, sig); 
 
hold on; 
 
plot(tms/Fs, 1,'gv','MarkerFaceColor','k');