% ecgBag.tar is a collection of ECG processing algorithms for use % with Matlab mostly by Gari Clifford and available under the GNU % public license (see bottom of this file). % % % They are not intended as optimal algorithms, but they're not bad..... % Please don't blame me if anything goes wrong - but feel free to % email me and ask questions, which I'll answer if I have time % Gari Clifford (gari@ieee.org) % % calc_lfhf.m - calculates the LF/HF ratio, LF and HF metrics % lomb.m - calculates the Lomb-Scargle periodogram % (PSD for unevenly sampled signals) % rpeakdetect.m - finds R peaks in ECG - see rrextract % rrextract.m - runs over rpeakdetect to extract RR intervals % clean_hrv4.m - removes abnormally timed beats % clean_RR_RS.m - removes abnormally timed & shaped beats % example.m - an example file to run all this % maxjiggle.m - finds peaks if the fiducial point is slightly off % interp_RR.m - interpolates an unevenly sampled signal by your favourite % (ad hoc) method. NOT recommended - it just introduces % errors into a PSD calculation - use the Lomb-Scargle % method, it's so much better. % filterECG256Hz.m - ECG, 256Hz, Least Sq. FIR LP & HP filters % (cascaded) 70dB 0.05-40Hz 1dB ripple % filterECG128Hz.m - ECG, 128Hz, Least Sq. FIR LP & HP filters % (cascaded) 70dB 0.05-40Hz 1dB ripple % lfhf_sliding_win.m - calculates a time series of LF/HF values by % sliding a window across the data % sliding a window across the data % % generate_rr_with_FM.m - Creates an artificial RR interval % time series % % down_sample_realistically.m - takes an evenly sampled time series % and extracts a plausible (irregularly % sampled) RR interval time series % ECGfilterTimeAlignedSVD.m - uses a QRS detector and SVD to filter ECG % select_train.m - makes a matrix of fiducual point centered % P-QRS-T complexes. % recon_ECG.m - unwraps a matrix of fiducual point centered % P-QRS-T complexes into a 1-D signal % svdFilter.m - filters data using SVD - variant of ECGfilterTimeAlignedSVD.m % wienerFilter.m - Wiener filtering for the ECG % plotWiener.m - plots the spectral response of wienerFilter.m % wavelet_decompose.m - A wavelet filter for ECG analysis by Nick Hughes % wabp.m - James Sun's version of Wei Zong's wabp.c (on www.physionet.org) % (i.e. - a QRS detector for blood pressure signals) % % snr.m - calcualtes the signal to noise ration in dB % % conv_norm_annot.c ... this C-prog parses the output of the WFDB % rdann program to extract the Normal to Normal intervals and puts % it in a readable format for Matlab - read the header for details % conv_sleep_annot.c ... as above, but works on sleep staging... % - again, read the header for details, and type: % type readme % % This software is offered freely and without warranty under the % GNU public license. Please don't blame me if anything goes wrong :) % % gari@ieee.org % for the polysomnographic DB, you would run rdann as % rdann -r slp02a -a st % and it spits out the data to stdout .. . so best to do - % rdann -r slp02a -a st > fileout.dat % (note that you must have the data on a current path to do this). % % put simply: % rdann -r slp01a -a st > slp01a.ascii % conv_sleep_annot slp01a.ascii slp01a.tRnKart > log1 % rdann -r slp01a -a ecg > slp01a.tmp % conv_norm_annot slp01a.tmp slp01a.bts > log2 % % slp01a.tRnKart should be a 3 column file of [time_stamp sleep_stage other_label] % slp01a.bts should be a 2 column file of [time_stamp beat_label] % % The time stamps are in seconds since the beginning of the file % (so check your sampling frequency is the same as the #define in the C file). % % The sleep stage are: % -1 W subject is awake % 0 R REM sleep % 1 1 sleep stage 1 % 2 2 sleep stage 2 % 2 2 sleep stage 3 % 4 4 sleep stage 4 % % The other labels are % -0.1 No label in 3rd column % -2 H Hypopnea % -0.2 HA Hypopnea with arousal % -3 OA Obstructive apnea % -0.3 X Obstructive apnea with arousal % -4 CA Central apnea % -0.4 CAA Central apnea with arousal % -5 L Leg movements % -0.5 LA Leg movements with arousal % -0.6 A Unspecified arousal % -10 MT Movement time % % The beat labels are % 1 == normal % 2 -> 19 == abnormal/ectopics % 2 = L % 3 = R % 4 = B % 5 = A % 6 = a % 7 = J % 8 = S % 9 = V % 10 = r % 11 = F % 12 = e % 13 = j % 14 = n % 15 = E % 16 = f % 17 = Q % 30 and above == artefact % 20 -- unspecified (\?) % 25 -- paced (P) % 0 -- error - classification not recognized % % This program is free software; you can redistribute it and/or modify % it under the terms of the GNU General Public License as published by % the Free Software Foundation; either version 2 of the License, or % (at your option) any later version. % % This program is distributed in the hope that it will be useful, % but WITHOUT ANY WARRANTY; without even the implied warranty of % MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the % GNU General Public License for more details. % % You should have received a copy of the GNU General Public License % along with this program; if not, write to the Free Software % Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA % % Most of these m-files and their dependents should be freely available from % Physionet - http://www.physionet.org/ - in the near future... but please % report any bugs to gari@ieee.org % Change Log % ---------- % Robert Tratnig at Graz Uni added the following comments and changes: % % clean_hrv_4.m: % I have added a line that says "stp=0;". % I call this line "38". % According to this indication I added lines 38,39,40,43,44,45,47,50,51,82. % Additionally I have altered line 40 (I added a condition: "(stp==0) & ". % % The same scheme I have applied to clean_RR_RS.m % Here I call the line that says "stp=0;" line "51". % In that context I added lines 51,52,53,56,57,58,60,61,62,102. % Also here I have altered a line: 54 (I added a condition: "(stp==0) & ". % % I have to do this because of the following reason: % When analyzing the file cu06.dat of the Creighton University Databank (CU_DB) % in the MIT-data collection, the parameter transfered to this function % clean_hrv4.m and clean_RR_RS.m (temp_hrv) was just a 3 rows long array of 2 % columns. Additionally no row satisfied the condition of the while loop, so I % had an error called "Index exceeds matrix dimensions". % % Many thanks for that Rob! % % % % ToDo % ---- % % - vectorise lomb.m ... currently very slow. this code was emailed to % me sometime in the past by a colleague who obtained it freely from the % internet. I have made minor modifications but have so far failed to have % time (or sufficient reason) to vectorise it (lomb.c does the job!) % % - Add optimisation routine to make either end of RR tachogram window % have same mean and gradient within a tolerance as per Schreiber's % surrogate window matching.