Short Summary & Block Diagram

This repository serves as a companion to the paper entitled "Enhanced Atrial Fibrillation (AF) Detection via Data Augmentation with Diffusion Model," authored by A. Vashagh, A. Akhoondkazemi, S J. Zahabi, and D. Shafie, which has been accepted for publication in IEEE by the 2023 International Conference on Computer and Knowledge Engineering (ICCKE) in Mashhad, Iran. Our project's main objective was to improve atrial fibrillation detection by transforming a 1-D image into a 2-D image and employing a data augmentation technique based on the Diffusion Model. The provided block diagram presents an overview of our methodology. For a comprehensive understanding, we encourage you to explore our full paper.

System Details

• System Model: ASUS TUF Dash F15
• Processor: 12th Gen Intel(R) Core(TM) i7-12650H 2.30 GHz
• Installed RAM: 32.0 GB (31.6 GB usable)
• Graphics Card: NVIDIA GeForce RTX 3070 Laptop GPU

Requirements

Matlab

Access to Matlab program and command line

Python 3.11.4

Install the requirements using

pip install -r requirements.txt

Code Structure

The majority of the core MATLAB code was originally written as part of the 2017 PhysioNet contest by the BlackSwan group, who were the winners of the competition. The authors of the paper contributed the Python code and some additional MATLAB scripting.

R-R Extraction

To expedite the development of our algorithm, we utilized the R-peak detection code from the BlackSwan group. In the initial phase of our work, we extracted the R-peaks and stored them in a MATLAB cell using the 'RrExtraction.m' script.

Preprocessing and Image-Construction

Subsequently, we loaded the R-R intervals in Python and proceeded to generate the preprocessed Poincaré images. Below are a few examples of the preprocessed images. The first example represents an atrial fibrillation (AF) image, while the second example illustrates a normal image.

### Image augmentation and classification. In this part, we used a CNN to classify the data to AF and not-af. Both of these steps are combined in **/jupyter-notebook/augment-and-classify.ipynb**

Note: The outputs of our last execution are also visible in /jupyter-notebook/augment-and-classify.ipynb

Code Execution

Matlab code

After cloning the project, execute the following line in your matlab environment.

Matlab_scripts/RrExtraction

which will output the following .mat files. These two files will be passed on to the python code in the next section

• NormalRPeaks
• AfRPeaks

Python code

python ./preprocessing.py

python ./augment-and-classify.py