Exploiting Shared Knowledge from Non-COVID Lesions for Annotation-Efficient COVID-19 CT Lung Infection Segmentation

• Main Authors: Liao, Q. (Author), Xing, J. (Author), Yuan, L. (Author), Zhang, J. (Author), Zhang, Y. (Author), Zhu, H. (Author)
• Format: Article
• Language: English
• Published: Institute of Electrical and Electronics Engineers Inc. 2021
• Subjects: Article; benchmarking; Benchmarking; Biological organs; clinical article; collaborative learning; Collaborative learning; Collaborative learning model; computed tomography; computer assisted tomography; Computerized tomography; controlled study; coronavirus disease 2019; Coronaviruses; COVID-19; deep learning; Deep learning; diagnostic imaging; Evaluation metrics; feature extraction; Feature level; few-shot learning; ground glass opacity; human; Humans; image segmentation; knowledge; Knowledge management; knowledge transfer; Large dataset; learning algorithm; Learning systems; liver tumor; lung; Lung; lung infection; Lung infection; lung infection segmentation; lung lesion; lung nodule; pleura effusion; quantitative analysis; SARS-CoV-2; sensitivity and specificity; Signal encoding; Similarity coefficients; State-of-the-art performance; Tomography, X-Ray Computed; Viruses; x-ray computed tomography
• Online Access: View Fulltext in Publisher

 The novel Coronavirus disease (COVID-19) is a highly contagious virus and has spread all over the world, posing an extremely serious threat to all countries. Automatic lung infection segmentation from computed tomography (CT) plays an important role in the quantitative analysis of COVID-19. However, the major challenge lies in the inadequacy of annotated COVID-19 datasets. Currently, there are several public non-COVID lung lesion segmentation datasets, providing the potential for generalizing useful information to the related COVID-19 segmentation task. In this paper, we propose a novel relation-driven collaborative learning model to exploit shared knowledge from non-COVID lesions for annotation-efficient COVID-19 CT lung infection segmentation. The model consists of a general encoder to capture general lung lesion features based on multiple non-COVID lesions, and a target encoder to focus on task-specific features based on COVID-19 infections. We develop a collaborative learning scheme to regularize feature-level relation consistency of given input and encourage the model to learn more general and discriminative representation of COVID-19 infections. Extensive experiments demonstrate that trained with limited COVID-19 data, exploiting shared knowledge from non-COVID lesions can further improve state-of-the-art performance with up to 3.0% in dice similarity coefficient and 4.2% in normalized surface dice. In addition, experimental results on large scale 2D dataset with CT slices show that our method significantly outperforms cutting-edge segmentation methods metrics. Our method promotes new insights into annotation-efficient deep learning and illustrates strong potential for real-world applications in the global fight against COVID-19 in the absence of sufficient high-quality annotations. © 2013 IEEE.