COVID Detection System with X-ray Image Data Using Deep Learning

Project Title

COVID Detection System with X-ray Image Data Using Deep Learning

Project Area of Specialization Artificial IntelligenceProject Summary

COVID-19 detection is an extremely significant research theme nowadays, due to the prevailing pandemic situation due to Coronavirus in the world. In this study, we will develop a COVID detection system by training a model using Deep Learning techniques. We will incorporate a publicly available dataset which contains hundreds of chest X-ray images of COVID patients. In particular, the model will be able to detect COVID positive cases and survival rates. It is expected that the model will be tuned, trained, and tested on the acquired dataset with expected high accuracy, specificity, and sensitivity results.

Project Objectives

i. To classify the data by detecting COVID and non-COVID subjects for the acquired dataset.

ii. To classify the data by determining the survival rates of the patients.

iii. To select the most efficient model for the COVID detection.

iv. To quantify the performance of model based on prior information or features, or based on information provided beforehand or not given at all.

Project Implementation Method

 The study will be carried out on the basis of following steps:

i. Background study of COVID disease, and its implications by exploring the online research repositories.

ii. Background study of Deep Learning, and its various tools and applications in automation systems using the online research sources. This also includes the exploration of different DL models and their characteristics.

iii. Finding an online dataset comprising of COVID patients X-ray images, and learning to check its features.

iv. Performing cross-validation of data to ascertain which model can be effectively applied to the data, and finalizing the right model for the dataset.

v. Tuning the model (e.g., CNN) by preprocessing the data as well as adjusting the weights of the model according to the dataset. For instance, the survival rate data should exclude all those data rows (as in table given in Figure 2) where the survival information of the patients is missing.

vi. Dividing the data into training & testing groups, and training the model on the training dataset. For instance, 70% of the dataset can comprise of data allocated for training purpose only; whereas, the remainder 30% can be utilized for testing.

vii. Testing the data after the model is trained to ascertain the accuracy of the model.

viii. Visualizing the acquired results using different criteria such as, specificity or True Positive Rate (TPR) & sensitivity or True Negative Rate (TNR), precision or Positive Predictive Values (PPVs), and recall.

ix. Write the thesis.

Benefits of the Project

COVID detection system will be used as initial screening to better treat the COVID patients by timely detecting and screening the presence of disease by means of X-rays. This will reduce the workload of clinical physicians and will add up to their efficiency.

Technical Details of Final Deliverable

The expected deliverables at the end of the project are:

• A well-compiled and compressive thesis report.

• Publication/submission of research output of the project.

Final Deliverable of the Project Software SystemCore Industry MedicalOther Industries Health Core Technology Artificial Intelligence(AI)Other Technologies Big DataSustainable Development Goals Good Health and Well-Being for PeopleRequired Resources

Item Name	Type	No. of Units	Per Unit Cost (in Rs)	Total (in Rs)
			Total in (Rs)	50000
Course for deep learning	Equipment	3	4000	12000
For Accessing Paid Research Articles	Equipment	5	2000	10000
Printing Documents and Thesis	Miscellaneous	1	10000	10000
Google Colab GPU (Month)	Equipment	9	2000	18000