Note: This research is currently in progress. Hence, much details will not be discussed here. The above image is representative only and a reference link to the owner is given

The goal of this research is to provide a better covid-19 detection using chest x-ray images of the patients. This research is inspired from and continuation of the work: COVID-Net

Significance: The COVID-19 global pandemic continues to have devastating effect around the world. The only step the world has to fight this and future pandemics is to massively increase testing for the disease acroos the world. However testing capacities across the world has been very limited. Initially, to scale up testing capacity, we have wasted valuable time when we could saved countless lives. The complicated, costly and time consuming testing process was a major detterance in our fight against the pandemic. Hence there is mojor requirement to have a fast, cheap, reliable and highly scalable testing method using the existing infrastructure, especially in a country like India. Deep learning methods like ours would prove hugely beneficial in a situation like this as first step in our fight.

Current research and their shortcomings: The major shortcoming currently for any research in this area is the availability of data. Most works currently have been tested on the small dataset available along with similar datasets like the rsna-pneumonia dataset. Our research plans to develop a better model to work on smaller training datasets and test on larger dataset

Approach: We are using an attention based model to make it more sensitive to covid-19 features and be better able learn these features. Our model has achieved 2% more sensitivity than the base model when trained on 467 covid-19 images and 150 test images.