Cloud Based AI Grader for Radiologists

Abstract

Our product is designed for the biomedical industry which would aid radiologists and doctors in the timely and efficient diagnosis of illnesses before they become fatal. The end product of our project is a web application whose target audience is the community of radiologists as well as the doctors. Our app is going to make up for the deficiency of time, limitation of workforce with respect to the abundant number of chest xrays, urgent need of reports which also at times leads to the misdiagnosis of the ailment. Our solution is based on the automation of manual reading and interpretation of chest xrays by radiologists and sometimes doctors with the help of a cloud based platform which has been developed using deep learning and image processing algorithms for the analysis of chest xrays. AI Grader for Radiologists takes a chest xray image as an input from the end user which would pass from 5 different models that includes segmentation of various anatomical structures present in chest, detection of opacity, detection of external devices, detection of fourteen chest diseases in lungs as well as the model of report generation. The output would consist of the generated chest xray report with severity of the disease graded as well as the CTR calculated and displayed. We have trained all the 5 models using datasets acquired from trusted online resources. With the prevalence of chest related diseases and the threat they pose to the society, our solution is going to serve as calm in the chaos