XRay-Eye

Android Application
Web Application
Model Accuracy Graph Results
Wireframe

Inspiration

There are over 20 Million of them every year, and a staggering 40% of them are chest X-rays, accounting for over 8 Million. Scientific studies have argued that clinically major errors in radiology have a 2 to 20 percent chance of occurring. Other research has also pointed that mammographs for detecting cancer can be misdiagnosed 61% of the time.

In the industry, there has also been an increasing number of vacancies in radiology positions. There is a shortage of over 1,00,000 Radiologists worldwide which is also predicted to increase!

In light of this information, “the increasing vacancy in radiologist positions, and current events surrounding the world, such as Covid-19 pandemic, we asked ourselves: How can we fill up this gap, speed up the process and prioritize vulnerable patients coming to the Hospitals for X-ray diagnostic testing, while aiding health professionals in performing their duties accurately and consistently?”

What it does

We have developed our app called “XRay-Eye” as a Decision support tool that uses machine learning to expedite the way frontline care workers identify lung-related abnormalities, which are typically associated with conditions such as Covid-19 infections, Pneumonia and other diseases.

Our app empowers clinicians-from emergency doctors to nurses to administrative staff to Radiologists by providing immediate diagnosis of a chest radiograph, followed by showing the results in terms of percentage confidence level enhancing their ability to form an accurate diagnosis at the time treatment is prescribed.

How we built it

We have used machine learning's deep learning application of neural network in training our Image classifier model using 1000 Open source Samples of each Covid-19, Pneumonia and Normal X-rays with accuracy close to 95% as shown in attached images.

Challenges we ran into

The biggest challenge was to find the right Datasets each for Covid-19, Pneumonia and Normal Chest X-rays. Then second one was to Train our machine learning model close to its accuracy with minimal Loss as shown in the attached Graph Results and last challenge was to Integrate our Trained model in two platforms such as Android App and Web Application.

Accomplishments that we're proud of

The biggest Accomplishment was that we were able to Train our machine learning model to its 95% accuracy in matter of 2 days though the open source data we got is not that big(1000 samples each). Secondly, we were able to make functional prototypes on two platforms such as Android and Web Application.

What we learned

Honestly, during this competition, we learned a lot about Product Ideation in solving the Real world problem of speeding up the Testing process in situations like Covid-19 Pandemic using Machine learning Technology.

What's next for XRay-Eye

We are expecting to Win this competition to get a chance to meet Expert Entrepreneurs in person Office hours to learn and expand our Idea to Startup.

Please check our full 4:35 min Demo Pitch Video including the Functional Apps demo starting at** [03:38 mins]** with the link : [https://vimeo.com/424477958]

Thank you,

Team XRay-Eye