Deep Learning Based Covid-19 Diagnosis Using Lung Images

In 2019, a new virus named corona virus had changed the life of every individual in the world. As the number of covid positive cases had been increased, it causes a very big pressure in the medical field. To overcome this situation, we are in need of some algorithm which predicts whether the person is affected by COVID-19 or not. The most known deep learning method is used to detect whether the person’s lung is affected by covid or not. In this project, lung CT images are segmented and then it is given into the simple convolutional neural network. The image segmentation techniques followed are canny edge detection, thresholding technique and U-Net algorithm. From these techniques the better one is chosen and its result is pushed into the convolutional neural network. By segmenting the image and then predicting whether the lung is affected by covid or not increases the accuracy rate and the accuracy value is 95%.