ElectroCardioGram(ECG) Image inference

Providing ElectroCardioGram- ECG interpretation using computer vision, to help reduce heart attack deaths

Omdena Local Chapter: Morocco


The Problem

The objective of the project is to use the ECG data to build a computer vision system that will be able to learn from cardio specialist doctors to perform the classification of a new patient based on his/her ECG image. Providing ElectroCardioGram (ECG) interpretation to general doctors on time using AI (Computer vision) will enable them to properly assess the patient and alert them in case of heart attack symptoms. This is extremely helpful in regions where medical staff is limited and access is very difficult.

This solution has the potential to serve a large population in the rural regions in Morocco but also for other African countries lacking such specialties.

The Project Outcomes

The participants will be able to apply AI for one of the most recurrent scenarios in Healthcare and could be also extending the case study to other radio diagrams or patient imagery data. E-health is just starting to develop AI and thus the project will bring a big opportunity to practice and increase the impact.

1. Understanding Medical Image Data

2. Image Processing

3. Image Classification & Segmentation

3. Computer Vision Models

Web Application Link: ECG Web App

Source Code:https://github.com/OmdenaAI/omdena-morocco-heartrisk


My Tasks as a collaborator: 

- Being a collaborator, helped acquiring the ECG Images dataset.
- I was the Lead task:

  1. Preprocessing these images along with other collaborators.
  2. Implementing the web application and deploying it.

- I created a VGG model to give Inference about the ECG Image whether it is MI or Normal. with an accuracy of 95%.
- I created the web application using Flask, and deploy it to the AWS EC2 instance, and used Apache2.