Automatic Segmentation of Novel Coronavirus Pneumonia Lesions in CT Images Utilizing Deep-Supervised Ensemble Learning Network

Abstract

Background: The novel coronavirus disease 2019 (COVID-19) has been spread widely in the world, causing a huge threat to the living environment of people.Objective: Under CT imaging, the structure features of COVID-19 lesions are complicated and varied greatly in different cases. To accurately locate COVID-19 lesions and assist doctors to make the best diagnosis and treatment plan, a deep-supervised ensemble learning network is presented for COVID-19 lesion segmentation in CT images.Methods: Since a large number of COVID-19 CT images and the corresponding lesion annotations are difficult to obtain, a transfer learning strategy is employed to make up for the shortcoming and alleviate the overfitting problem. Based on the reality that traditional single deep learning framework is difficult to extract complicated and varied COVID-19 lesion features effectively that may cause some lesions to be undetected. To overcome the problem, a deep-supervised ensemble learning network is presented to combine with local and global features for COVID-19 lesion segmentation.Results: The performance of the proposed method was validated in experiments with a publicly available dataset. Compared with manual annotations, the proposed method acquired a high intersection over union (IoU) of 0.7279 and a low Hausdorff distance (H) of 92.4604.Conclusion: A deep-supervised ensemble learning network was presented for coronavirus pneumonia lesion segmentation in CT images. The effectiveness of the proposed method was verified by visual inspection and quantitative evaluation. Experimental results indicated that the proposed method has a good performance in COVID-19 lesion segmentation.

Identifying the optimal deep learning architecture and parameters for automatic beam aperture definition in 3D radiotherapy

“…Segmentation on 2D x-ray images is also frequently described in the literature; for instance in lungs, [38][39][40] various phantom or human anatomic structures, 41 and recently in COVID-19 lesions. 42 However, to the best of our knowledge, no previous study has extensively evaluated selection of architecture or hyperparameters as we have described here; thus, translation across datasets or anatomic sites may be restricted.…”

Lung lobe segmentation in computed tomography images based on multi‐feature fusion and ensemble learning framework

“…Anatomically, the left lung is subdivided into two lobes (upper and lower lobes), whereas the right lung is subdivided into three lobes (upper, middle, and lower lobes), 2 as described in Figure 1. Clinically, lung lobe characteristics are important pathological indexes in assessing the severity of lung diseases and making treatment planning, 4,5 it can be used to help doctors diagnose emphysema, post‐primary tuberculosis, silicosis, idiopathic pulmonary fibrosis, chronic obstructive pulmonary disease and Covid‐19 6,7 . However, manual segmentation of lung lobes in computed tomography (CT) images is time‐consuming, tedious, and impracticable due to indistinguishable pulmonary arteries and veins, blurring pulmonary fissures, and unpredictable pathological deformation.…”