The following model is a convolutional neural network (CNN) that seeks to detect COVID-19 from a chest radiograph (CXR). Model weights can be downloaded from https://drive.google.com/drive/folders/1BdK0IZNSq0mU9SVsk5e4--QpLHyS_6pl?usp=sharing. To use the COVID-19 diagnosis model, run the line below. On a standard CPU, predictions should take less than 10 seconds to generate.
python run.py [PATH_TO_DCM_FILE]Training and testing scripts assume a CUDA device is installed. Images must be padded and resized to 224x224 prior to model training and testing.
For all inquiries, please contact chris_kim@alumni.brown.edu. Methodology and model performance are available in the article An automated COVID-19 triage pipeline using artificial intelligence based on chest radiographs and clinical data, available online at https://www.nature.com/articles/s41746-021-00546-w