Abstract:Infection segmentation on chest CT plays an important role in the quantitative analysis of COVID-19. Developing automatic segmentation tools in a short period with limited labelled images has become an urgent need. Pseudo label-based semi-supervised method is a promising way to leverage unlabelled data to improve segmentation performance. Existing methods usually obtain pseudo labels by first training a network with limited labelled images and then inferring unlabelled images. However, these methods may generate obviously inaccurate labels and degrade the subsequent training process. To address these challenges, in this paper, an active contour regularized semi-supervised learning framework was proposed to automatically segment infections with few labelled images. The active contour regularization was realized by the region-scalable fitting (RSF) model which is embedded to the loss function of the network to regularize and refine the pseudo labels of the unlabelled images. We further designed a splitting method to separately optimize the RSF regularization term and the segmentation loss term with iterative convolution-thresholding method and stochastic gradient descent, respectively, which enable fast optimization of each term. Furthermore, we built a statistical atlas to show the infection spatial distribution. Extensive experiments on a small public dataset and a large scale dataset showed that the proposed method outperforms state-of-the-art methods with up to 5% in dice similarity coefficient and normalized surface dice, 10% in relative absolute volume difference and 8 mm in 95% Hausdorff distance. Moreover, we observed that the infections tend to occur at the dorsal subpleural lung and posterior basal segments that are not mentioned in current radiology reports and are meaningful to advance our understanding of COVID-19.

Segmenting Lung Fields in CT Image Using Legendre Moments and Active Contour

Segmentation for CT Image Based on Improved Level-Set Approach

A Region Based Active Contour Method for X-Ray Lung Segmentation Using Prior Shape and Low Level Features

Global Optimal Hybrid Geometric Active Contour for Automated Lung Segmentation on CT Images

Automatic Segmentation of Liver from CT Scans with CCP-TSPM Algorithm

Automatic Approach for Lung Segmentation with Juxta-Pleural Nodules from Thoracic CT Based on Contour Tracing and Correction.

3D Segmentation of the Lung Based on the Neighbor Information and Curvature

Automatic Method for Lung Segmentation with Juxta-Pleural Nodules from Thoracic CT Based on Border Separation and Correction.

3-D Segmentation Of Lung Nodules In Ct Images Based On Improved Level Set Method

Auto-segmentation Method for Lung Parenchyma of CT Images

A Visually Guided Framework for Lung Segmentation and Visualization in Chest CT Images

Automated Lung Segmentation and Smoothing Techniques for Inclusion of Juxtapleural Nodules and Pulmonary Vessels on Chest CT Images

Automated Lung Segmentation for Chest CT Images Used for Computer Aided Diagnostics

A 3D Segmentation and Visualization Scheme for Solid and Non-solid Lung Lesions Based on Gaussian Filtering Regularized Level Set

Multi-stage learning for robust lung segmentation in challenging CT volumes.

Lung Segmentation From Ct With Severe Pathologies Using Anatomical Constraints

Active contour regularized semi-supervised learning for COVID-19 CT infection segmentation with limited annotations

A 3D segmentation method of lung parenchyma based on CT image sequences

An Optimal Level Sets Method for Lung Nodules Detection in CT Images

A PET-CT Lung Tumor Segmentation Method Based on Active Contour Model

A Fast Automatic Method Of Lung Segmentation In Ct Images Using Mathematical Morphology