COVID Detection and Severity Prediction with 3D-ConvNeXt and Custom Pretrainings

TL;DR

This paper presents a 3D-ConvNeXt based neural network for COVID detection and severity prediction from lung CT scans, incorporating custom pretraining methods, achieving top rankings in COVID challenges.

Contribution

It adapts the ConvNeXt model for 3D CT data and introduces specialized pretraining techniques to enhance model performance.

Findings

Ranked 2nd in COVID19 Severity Detection Challenge

Ranked 3rd in COVID19 Detection Challenge

Improved model performance with custom pretraining methods

Abstract

Since COVID strongly affects the respiratory system, lung CT-scans can be used for the analysis of a patients health. We introduce a neural network for the prediction of the severity of lung damage and the detection of a COVID-infection using three-dimensional CT-data. Therefore, we adapt the recent ConvNeXt model to process three-dimensional data. Furthermore, we design and analyze different pretraining methods specifically designed to improve the models ability to handle three-dimensional CT-data. We rank 2nd in the 1st COVID19 Severity Detection Challenge and 3rd in the 2nd COVID19 Detection Challenge.