Deep-Learning-Based Estimation of Spatial QRS-T angle

Welcome to the DeepLearningSpatialQRS-Tangle repository! 🤓

If you're here, you probably saw our open-access research paper: "Deep-Learning-Based Estimation of the Spatial QRS-T Angle from Reduced-Lead ECGs", published on 20^th July 2022. In this repository, you'll be able to find code and any relevant documentation mentioned in the paper.

When using this resource, please cite:

  Santos Rodrigues A, Augustauskas R, Lukoševičius M, Laguna P, Marozas V. 
  Deep-Learning-Based Estimation of the Spatial QRS-T Angle from Reduced-Lead ECGs. Sensors. 
  2022; 22(14):5414. https://doi.org/10.3390/s22145414

A few important notes before we start:

• The code is written in MATLAB and Python. The required versions and toolboxes are specified in each section.
• Some of the code was implemented by other authors, who are stated/cited accordingly throughout this README.md and the .m files. We provide all relevant citations in the Refs.bib file to simplify referencing and citation.
• The prefix "fn" in a filename indicates a MATLAB function (fnFunctionName.m). In every function file, you'll find:
  • A description of the function;
  • Any applicable literature;
  • Summary of input/output variables;
  • A list of additional .mfiles.

It looks roughly like this:

function [out_A] = fnFunctionName(in_A, in_B)

Expand here for details

function [out_A] = fnFunctionName(in_A, in_B)
  %FNFUNCTIONNAME is an example to describe the code syntax used. This is the function description.
    % Any relavant literature is listed as:
    %     [1]: Author 1, Author 2. Title. Journal. Year. DOI.
    %     [2]: Author 1, Author 2. Title. Journal. Year. DOI. 
    %------------------------------------------------------------
    %   INPUTS:
    %------------------------------------------------------------
    %     1) in_A: This is variable of size MxN.
    %     2) in_B: This is another variable.
    %------------------------------------------------------------
    %   OUTPUTS:
    %------------------------------------------------------------
    %     1) out_A: This is variable.
    %------------------------------------------------------------
    %   NECESSARY FUNCTIONS:
    %------------------------------------------------------------
    %     1) fnAnotherFunction.m
    %------------------------------------------------------------
    % Written by: Author (email)
    % Updated: YYYY-MM-DD.
    %------------------------------------------------------------
    % According to [1]:
    out_A = in_A + in_B; 
end

---

Data

Physionet's PTB-XL dataset is the 12-lead clinical ECG dataset used in our research work. You can download the dataset here.

When using this database, please cite:

  Wagner, P., Strodthoff, N., Bousseljot, R.-D., Kreiseler, D., Lunze, F.I., Samek, W., Schaeffter, T. (2020), 
  PTB-XL: A Large Publicly Available ECG Dataset. Scientific Data. https://doi.org/10.1038/s41597-020-0495-6

(and cite this one too)

  Goldberger, A., Amaral, L., Glass, L., Hausdorff, J., Ivanov, P. C., Mark, R., ... & Stanley, H. E. (2000). 
  PhysioBank, PhysioToolkit, and PhysioNet: Components of a new research resource for complex physiologic signals. 
  Circulation [Online]. 101 (23), pp. e215–e220.

---

Training and Validation Sets

Section 4.3. of our research paper describes the data splitting process into training and validation datasets. Each dataset is disclosed on the files Training_Dataset.csv and Validation_Dataset.csv in the folder Split Datasets. Only the high-quality recordings suitable for analysis (see Section 4.1.1.) are listed. In each table, you'll find:

ecg_id	patient_id	sex	scp_codes	class	qrst_angle	qrst_range
16968	2089	1	{'NORM': 100.0, 'SR': 0.0}	["Norm"]	21.081	[20:25[
...	...	...	...	...	...	...
1427	20044	0	{'ASMI': 100.0, 'ISCAL': 100.0, 'SR': 0.0}	["MI","STTC"]	152.1103	[150:155[
...	...	...	...	...	...	...

The attributes ecg_id, patient_id, sex and scp_codes correspond to the same attributes in the file ptbxl_database.csv provided by the original documentation of the PTB-XL dataset, whereas qrst_angle is the estimated QRS-T angle $\alpha$ (°) using the conventional approach and qrst_range is the range of $\alpha$ = [0:5:180]° as described in the paper.

You can easily load the .csv files in the MATLAB environment with the readtable function:

  Train_Data  = readtable("Training_Dataset.csv");
  Val_Data    = readtable("Validation_Dataset.csv");

---

How to read the raw ECG files in MATLAB?

After downloading the raw files from the PTB-XL dataset page, you can open the .dat files in MATLAB:

  % Switch the environment/directory
  cd([YOUR_DATA_DIR])
  % Read file
  fid        = fopen([recording_name, '.dat']);
  % Get the raw ECG and convert to volts
  ecg_raw    = transpose(fread(fid, [12 inf], 'int16'))./1000;
  % Close file
  fclose(fid);
  % Get back to working environment
  cd(WORKING_DIR);

Alternatively, you can use our function fnFileReaderECG.m. You can also use Physionet's WFDB Software Package, but we do not recommend it (it is excruciantingly slow).