3D U-Net for segmentation of COVID-19 associated pulmonary infiltrates using transfer learning: State-of-the-art results on affordable hardware

TL;DR

This study demonstrates that transfer learning enables efficient training of 3D U-Net models for COVID-19 lung infiltrate segmentation on limited hardware, achieving state-of-the-art results with reduced training time.

Contribution

The paper introduces a transfer learning approach using a pretrained 3D ResNet encoder to train a 3D U-Net for pulmonary infiltrate segmentation on limited hardware.

Findings

Achieved a mean Dice score of 0.679 on the tuning dataset.

Achieved a Dice score of 0.648 on the Corona Cases dataset.

Achieved a Dice score of 0.405 on the MosMed dataset.

Abstract

Segmentation of pulmonary infiltrates can help assess severity of COVID-19, but manual segmentation is labor and time-intensive. Using neural networks to segment pulmonary infiltrates would enable automation of this task. However, training a 3D U-Net from computed tomography (CT) data is time- and resource-intensive. In this work, we therefore developed and tested a solution on how transfer learning can be used to train state-of-the-art segmentation models on limited hardware and in shorter time. We use the recently published RSNA International COVID-19 Open Radiology Database (RICORD) to train a fully three-dimensional U-Net architecture using an 18-layer 3D ResNet, pretrained on the Kinetics-400 dataset as encoder. The generalization of the model was then tested on two openly available datasets of patients with COVID-19, who received chest CTs (Corona Cases and MosMed datasets). Our model performed comparable to previously published 3D U-Net architectures, achieving a mean Dice score of 0.679 on the tuning dataset, 0.648 on the Coronacases dataset and 0.405 on the MosMed dataset. Notably, these results were achieved with shorter training time on a single GPU with less memory available than the GPUs used in previous studies.