As we continue to face the rapid increase in confirmed Coronavirus cases around the world, we created an AI-based pneumonia detection platform for COVID-19. The system is able to automatically detect high-risk patients with pneumonia that will then send information to doctors. With that information, the doctors are then able to make follow-up decisions and provide a treatment plan after the diagnosis. In specific, doctors from the Department of Medical Imaging provided us thousands of positive and negative chest x-rays for pneumonia as a training set. Our system has already been tested with and adopted by doctors at the NCKU Hospital. The system achieved 95% accuracy to detect the pneumonia symptom, based on 1400 test images.

AI-based disease detection is usually considered as a pattern classification task. In general, it uses a classification model with Convolutional Block Attention Module (CBAM) as a common way to handle it. Instead of observing the last feature map in the model, we design the image segmentation stage as our target task. With radiologists’ manually labeled images as the target, we utilize the segmentation model as the first stage to identify potential hot spot areas. The following stage with a simple threshold to determine if any pixel is predicted as positive(diseased). If there existed at least one positive pixel, the image is considered as infected cases.

We aim to help with the COVID-19 pandemic around the world. In this project, we use AI technology to determine if the patient is infected with pneumonia from a chest x-ray. Our system assists doctors with diagnosis to find out confirmed cases earlier, leading to a decrease in the risk of community spread. If the lack of medical personnel is serious, our system may be the first line helper for filtering COVID-19 suspected cases. Inexperienced doctors or medical personnel can learn the infected patterns of COVID-19 on chest x-ray while they obtain the initial prediction from our system.