Classify ECG Signals Using Long Short-Term Memory Networks

% This script parses the data from the PhysioNet 2017 Challenge and saves
% the data into PhysionetData.mat for quick and easy future use.
% This function is in support of ClassifyECGSignalsUsingLSTMNetworksExample. It may
% change or be removed in a future release.
% Copyright 2019 The MathWorks, Inc.
% Download and unzip the data, training2017.zip, from the PhysioNet website
unzip('https://archive.physionet.org/challenge/2017/training2017.zip')
% Navigate to the directory
cd training2017
% File with filenames and labels
ref = 'REFERENCE.csv';
% Create a table that contains the filenames and corresponding label data
tbl = readtable(ref,'ReadVariableNames',false);
tbl.Properties.VariableNames = {'Filename','Label'};
% Delete 'Other Rhythm' and 'Noisy Recording' signals
toDelete = strcmp(tbl.Label,'O') | strcmp(tbl.Label,'~');
tbl(toDelete,:) = [];
% Load each file in the table and store the corresponding signal data
H = height(tbl);
for ii = 1:H
    fileData = load([tbl.Filename{ii},'.mat']);
    tbl.Signal{ii} = fileData.val;
end
% Leave the training2017 directory
cd ..
% Format the data properly for LSTM training
% Signals: Cell array of predictors
% Labels: Categorical array of responses
Signals = tbl.Signal;
Labels = categorical(tbl.Label);
% Save the variables to a MAT-file
save PhysionetData.mat Signals Labels

2 个评论
显示无隐藏无

Kristofer Soler 2020-3-1

Cheers ! works perfectly !

debojit sharma 2022-3-24

I was trying to use the above program for my own data (communication signal) but I am getting error in the line "tb1.Signal{ii} = fileData.val;" as Reference to non-existent field 'val'. Someone please help

请先登录，再进行评论。