: Solution Description
Identification of CORONA has become a primary concern for us at this moment. We propose an AI solution to detect the probable COVID-19 disease from X-ray images. We also present a data collection tool that results in a Radiology Database of COVID-19 cases with X-ray along with the case details history of Bangladeshi patients and at the same time, makes the detection accuracy of our solution better over time. We are detecting Corona from X-ray images rather than the CT Scan images as the expenses of CT is much higher.
We use deep learning to develop our model that uses a multi-size filter that extracts features of various dimensions. It extracts general features (for example 5x5 filters) and local features (for example 1x1 filters) at the same time. Our solution prevents gradient vanishing problems so that it can help to train deep layers. We use a deep network instead of a small one to provide versatility for feature extraction.