COVID Detection From X-Ray Images using Self Attention Mechanism (Transformers)

Project Title

Project Area of Specialization

Artificial Intelligence

Project Summary

COVID-19 has become a global pandemic issue, it has bad effects on the health, economy and population of the world. As Chest X-Ray is emerging as a valuable diagnostic tool for clinical management of COVID-19. Artificial intelligence (AI) has the power to detect and aid to the fastest evolution of X Rays for detection of COVID-19 findings. So, we are working on the autonomous COVID-19 detection from X-Rays Images using well know models of deep learning, we will us multinational datasets and train those using different techniques and algorithms of deep learning. And we will implement this using the web Application. This will be helpful for the society, beneficial for the country and can reduce the cost of testing the people for COVID-19, because a simple detection application will tell the patients are either COVID-19 positive or not, also it will diagnose for lungs diseases. This is developed because almost 90% people are using mobile phones and they can easily access the web application and test themselves by giving an X-Ray image as input, the system will also detect other lungs diseases too.

AI has emerged in everyday life so that AI-based algorithms can readily identify X-Rays scans with COVID-19 associated any other diseases, as well as distinguish non-COVID-19 patients with high specificity in diverse patient populations.

Project Objectives

After the completion of this project we will be able to solve the problem of testing the COVID-19 Patients, this project also covers many other aspects including:

Easy use of web application to test whether the person is COVID-19 positive or negative.
Time saving, the most important feature of our project, because in the hospital centers there are many
long queues that wait for their turn to go for the testing.
Inexpensive web application compared to Test kit prices, our project can save the cost of test kits and
can give the results accurately and comparatively better ones.
24/7 Availability anyone can use it anywhere.

Project Implementation Method

The above procedure can be summarized into 7 stages as (Major);
Stage1: To design a prototype for the classification of X-Ray image.
Satge2: Collect the athentic data from the hospitals and create a dataset of the X-ray Images.
Stage3: Research the best models for the classification and detection, and use one of them. (Our main objective is to implemet the self-attention mechanism i.e Transformers)
Stage4: Training the model
Stage5: Testing the model
Stage6: Evaluate the results
Stage7: Comparative analysis of the results obtained from the model.

Implementation in Web Framework (Django)
Satge1: Web Architecture using Django,
Stage2: HTML & CSS architecture and layout designing
Stage3: Python for backend and also some of the Javascript scripts.
Stage4: Dashboard that represents the Gloabl data and Pakistan's Data and Visualization.
Stage4: Database Design and Integeration

Benefits of the Project

Addition in literature of COVID-19, and resarch on Self Attention Mechanism, because this is a generic and a recent problem. Mainly this project is intended to analyze all the above-mentioned characteristics as the proposed results and analysis would bebeneficial for achieving the best results in classifying the COVID-19 from X-Ray Images. The implementation of this system will decrease the test kit costs. This system will play a vital role in revealing the most facts about the COVID-19, and will add up something in the COVID-19 Literature.

Technical Details of Final Deliverable

This project aims to implement a fully functional software of smart surveillance system in which image is classified as COVID-19 Positive or Negative. In this project a model is trained and tested on different datasets for classification of images. One short Learning will be used that can be trained on low dataset and can be effective in identifying images. Web scraping will also be used to visualize the COVID informatics from reliable resources. It will automatically classify the images. Test kits cost will be reduced with that system. In this project, different models will be used and comparative analysis will be done on these model to check the best performing model on given conditions.

The NVIDIA® Jetson Xavier NX™ Developer Kit includes a power-efficient, compact Jetson Xavier NX module for AI edge devices. It benefits from new cloud-native support, and accelerates the NVIDIA software stack in as little as 10 W with more than 10x the performance of its widely adopted predecessor Jetson TX2. This lets you run multiple neural networks in parallel for applications like image classification, object detection, segmentation, and speech processing.