Receive a weekly summary and discussion of the top papers of the week by leading researchers in the field.

ArXiv Preprint

Since COVID strongly affects the respiratory system, lung CT scans can be used for the analysis of a patients health. We introduce an neural network for the prediction of the severity of lung damage and the detection of infection using three-dimensional CT-scans. Therefore, we adapt the recent ConvNeXt model to process three-dimensional data. Furthermore, we introduce different pretraining methods specifically adjusted to improve the models ability to handle three-dimensional CT-data. In order to test the performance of our model, we participate in the 2nd COV19D Competition for severity prediction and infection detection.

Daniel Kienzle, Julian Lorenz, Robin Schön, Katja Ludwig, Rainer Lienhart

2022-06-30