Myocardial infarction (MI)-induced changes in the morphological and temporal wave features are extracted from the derived VCG using spline approximation.

After the feature extraction, a classifier based on multilayer perceptron network (MLP) is used for MI classification.

Experiments on PTB diagnostic database demonstrate that the proposed system achieved satisfactory performance to differentiating MI patients from healthy subjects and to localizing the infarcted area.

Details of this study please refer to 👉 https://doi.org/10.3390/s20247246

This project implement only the latter part of the proposed system. (The other part will be added soon)

To automatically download the records from PTD database, please refer to 👉 https://github.com/yuhung1206/Auto_download_PTB

1. Spline-based feature extraction spline

   • Main program: SplineFeatureExtract.m

   • Sub programs: SplineFit_1lead_ECG.m, SplineFit_2lead_ECG.m, SplineFit_3lead_ECG.m, pan_tompkin.m

     

     Code	function
     SplineFit_klead_ECG.m	extract feature from k leads & generate label for learning
     pan_tompkin.m [1][2]	R-peak detection
     plotATM.m	Load .mat and .info files

   • Get function "plotATM.m" from physionet website
     👉 https://archive.physionet.org/physiotools/matlab/plotATM.m

   • Get function "pan_tompkin.m"
     👉 https://www.mathworks.com/matlabcentral/fileexchange/45840-complete-pan-tompkins-implementation-ecg-qrs-detector

   • Set the lead_Num as [13,14,15] to extract features from VCG (Vx, Vy, Vz)
     

   • Set Output Name & directory to store features
     

   • Input filename make sure that input filename is correct
     

   • Choose how many leads to extract:
     SplineFit_1lead_ECG.m -> 1 lead features(18 dimension)
     SplineFit_2lead_ECG.m -> 2 lead features(35 dimension)
     SplineFit_3lead_ECG.m -> 3 lead features(52 dimension)

     

     • The setting to extract features from Vx+Vy leads is shown below:
       and select the function SplineFit_2lead_ECG.m
       

   • Start extraction
     

2. Multi-Layer-Perceptron MLP
   Note: input are features rather than signals

   • Main program: Classification_SMOTE.m

   • Sub programs: mySMOTE.m

     Code	function
     Classification_SMOTE.m	FNN for classification
     mySMOTE.m	deal with the imbalanced database

   • Get mySMOTE.m from [3] 👉 https://www.mathworks.com/matlabcentral/fileexchange/70315-smote-over-sampling

     

   • make sure InputName is correct:
     

   • the classification performance will be stored in total variable:
     Because it is 12-type classification, the size of total is [12x12].

   • start training
     
     

[1] Sedghamiz. H, "Matlab Implementation of Pan Tompkins ECG QRS detector.", March 2014. https://www.researchgate.net/publication/313673153_Matlab_Implementation_of_Pan_Tompkins_ECG_QRS_detect
[2] PAN.J, TOMPKINS. W.J,"A Real-Time QRS Detection Algorithm" IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. BME-32, NO. 3, MARCH 1985.
[3] Abhishek Gupta (2021). SMOTE-over-Sampling (https://github.com/earthat/SMOTE-over-Sampling), GitHub. Retrieved September 11, 2021.