This repository contains a functional code composed by Matlab and bash codes for Interbeat-interval (IBI), Heart-rate (HR) calculation, and HRV metrics from an artifactual Blood Volume Pulse (BVP) signal, and a corresponding accelerometer. Please refer to our paper from EUSIPCO 2016 Torres, J. M. M., Ghosh, A., Stepanov, E. A., & Riccardi, G. (2016, August). Heal-t: An efficient ppg-based heart-rate and ibi estimation method during physical exercise. In 2016 24th European Signal Processing Conference (EUSIPCO) (pp. 1438-1442). IEEE.
Photoplethysmography (PPG) is a simple, unobtrusive and low-cost technique for measuring blood volume pulse (BVP) used in heart-rate (HR) estimation. However, PPG based heart-rate monitoring devices are often affected by motion artifacts in on-the-go scenarios, and can yield a noisy BVP signal reporting erroneous HR values. Recent studies have proposed spectral decomposition techniques (e.g. M-FOCUSS, Joint-Sparse-Spectrum) to reduce motion artifacts and increase HR estimation accuracy, but at the cost of high computationalload. The singular-value-decomposition and recursive calculations present in these approaches are not feasible for the implementation in real-time continuous-monitoring scenarios. In this paper, we propose an efficient HR estimation method based on a combination of fast-ICA, RLS and BHW filter stages that avoids sparse signal reconstruction, while maintaining a high HR estimation accuracy. The proposed method outperforms the state-of-the-art systems on the publicly available TROIKA data-set.
HEAL-T application can be executed in bash following the command:
sh runHEALT.sh DATA_PATH NAME_OUTPUT_folder METHOD_SELECTOR FS Fp1 Fz1 Fp2 Fz2 inc1 inc2 overlap WIN S_sel-"always 1" fileo.out MATLAB_PATHor - if DATA_PATH will be the same output path*
sh runHEALT.sh DATA_PATH METHOD_SELECTOR FS Fp1 Fz1 Fp2 Fz2 inc1 inc2 overlap WIN S_sel fileo.out MATLAB_PATH** YOU CAN ALSO RUN THIS CODE IN MATLAB directly **
- Please open matlab command window prompt and execute
addpath(genpath(HEAL-T_path))- Now you can run
heal_t_call(DATA_PATH,NAME_OUTPUT_folder,METHOD_SELECTOR,FS,{[Fp1 Fz1],[Fp2 Fz2]},[inc1,inc2],overlap,WIN,S_sel);- You can see the same notifications that appears in the log file from bash but in this case from the command window. (Please see below)
Before running this code be sure adding the following dependencies folders in the thirdparty directory
- eeglab: in this code we only used runica functions - please do the same, subsequently download the last eeglabversion from https://sccn.ucsd.edu/eeglab/downloadtoolbox.php
- MFOCUSS: This code library calculate a sparse spectrum reconstruction for stationary signals, download this from Zhiling Zhang code repository http://dsp.ucsd.edu/~zhilin/Software.html
- MATLAB > R2016a
- bash
15. MATLAB_PATH: this is the matlab binary path defined by user (i.e., in Matlab /usr/local/../bin/matlab and in mac /Applications/Matlab.../bin/matlab)
IF YOU RUN HEAL-T FROM BASH:
Take into accoun the folder tree description:
Main -> main code, i.e, pwd or a folder that includes .m
functionsHR -> auxiliary functions for HR processing and some functions necessary from MFOCUSS baseline evaluation and the first HEAL-T using time overlap
Data -> defined by user
eeglab -> appears in thirdparty folder please unzip it before run the code (i.e. EEGlab ICA)
sh runHEALT.sh DATA_FOLDER 0 1 32 0.7 2.5 0.7 3.5 0 0 1 50 10 fileo.out MATLAB_PATHYour input BVP and ACC files with the corresponding time series should be included in your Data folder as:
- bvp.txt
- acc.txt
bvp and acc files with the input time series must contain at least 2 columns (one -> unix_time, second->non-normalized values). In case of the accelerometer the file should include the time column and the three axis columns [t ; x ; y ; z] in this order. You can create this as a custom csv file delimited by colon.
** OUTPUT: outputfile will be generated in the given log folder, with the name and the ID of each process as prefix
The output files will be formatted as follows_
-
The average HR output file will be a .csv file with with three columns: Col1: window time [index], Col2: HR [bpm], and Col3: HR smooth [bpm]
-
The output file containing the IBI peak-to-peak will be another .csv file with three columns Col1: Time [s] (take into account your sample frequency), Col2: HR values [bpm] , Col3: HR smoothed values [bpm] (after Smoothing spline).
e.g. PID_IBIHEALPEAK.txt
Please review the output in the log folder and you can check the following items when the script runs successfully:
- The code is capable to read inputs when the log folder is already updated by the corresponding method ("File Exist!" string appears)
- Code will end when the string "Process PID has ended successfully" appears.
Note1: change the DATA_PATH as you desire : please take into account that inside this folder you should have a file called bvp.txt in which you should define your bvp with the same parameters and instructions explained above
Note2: Please run this code from the main directory and any input, data, or output data folder should be defined by user respecting the folder tree.
Some outputs time-series calculated from ICASSP2015 cup (http://archive.signalprocessingsociety.org/community/sp-cup/ieee-sp-cup-2015/) data are:
The Bland-Altman plots obtained from the ICASSP2015 data were:
Please read the paper Torres, J. M. M., Ghosh, A., Stepanov, E. A., & Riccardi, G. (2016, August). Heal-t: An efficient ppg-based heart-rate and ibi estimation method during physical exercise. In 2016 24th European Signal Processing Conference (EUSIPCO) (pp. 1438-1442). IEEE (https://ieeexplore.ieee.org/document/7760486) for more details
- Juan Manuel Mayor Torrres, Arindam Ghosh, Evgeny Stepanov, and Giuseppe Riccardi
See also the list of contributors who participated in this project.
see the LICENCE.md file for details
University of Trento - Signals and Interactive Sistems Lab members and alumni (http://sisl.disi.unitn.it/)