Dual Multi-scale Mean Teacher Network for Semi-supervised Infection Segmentation in Chest CT Volume for COVID-19 (preprint)

Automated detecting lung infections from computed tomography (CT) data plays an important role for combating COVID-19. However, there are still some challenges for developing AI system. 1) Most current COVID-19 infection segmentation methods mainly relied on 2D CT images, which lack 3D sequential constraint. 2) Existing 3D CT segmentation methods focus on single-scale representations, which do not achieve the multiple level receptive field sizes on 3D volume. 3) The emergent breaking out of COVID-19 makes it hard to annotate sufficient CT volumes for training deep model. To address these issues, we first build a multiple dimensional-attention convolutional neural network (MDA-CNN) to aggregate multi-scale information along different dimension of input feature maps and impose supervision on multiple predictions from different CNN layers. Second, we assign this MDA-CNN as a basic network into a novel dual multi-scale mean teacher network (DM${^2}$T-Net) for semi-supervised COVID-19 lung infection segmentation on CT volumes by leveraging unlabeled data and exploring the multi-scale information. Our DM${^2}$T-Net encourages multiple predictions at different CNN layers from the student and teacher networks to be consistent for computing a multi-scale consistency loss on unlabeled data, which is then added to the supervised loss on the labeled data from multiple predictions of MDA-CNN. Third, we collect two COVID-19 segmentation datasets to evaluate our method. The experimental results show that our network consistently outperforms the compared state-of-the-art methods.

Ultra-High-Resolution CT Follow-Up in Patients with Imported Early-Stage Coronavirus Disease 2019 (COVID-19) Related Pneumonia (preprint)

Background: An ongoing outbreak of mystery pneumonia in Wuhan was caused by coronavirus disease 2019 (COVID-19). The infectious disease has spread globally and become a major threat to public health. Purpose: We aim to investigate the ultra-high-resolution CT (UHR-CT) findings of imported COVID-19 related pneumonia from the initial diagnosis to early-phase follow-up. Methods: This retrospective study included confirmed cases with early-stage COVID-19 related pneumonia imported from the epicenter. Initial and early-phase follow-up UHR-CT scans (within 5 days) were reviewed for characterizing the radiological findings. The normalized total volumes of ground-glass opacities (GGOs) and consolidations were calculated and compared during the radiological follow-up by artificial-intelligence-based methods. Results: Eleven patients (3 males and 8 females, aged 32-74 years) with confirmed COVID-19 were evaluated. Subpleural GGOs with inter/intralobular septal thickening were typical imaging findings. Other diagnostic CT features included distinct margins (8/11, 73%), pleural retraction or thickening (7/11, 64%), intralesional vasodilatation (6/11, 55%). Normalized volumes of pulmonary GGOs (p=0.003) and consolidations (p=0.003) significantly increased during the CT follow-up. Conclusions: The abnormalities of GGOs with peripleural distribution, consolidated areas, septal thickening, pleural involvement and intralesional vasodilatation on UHR-CT indicate the diagnosis of COVID-19. COVID-19 cases could manifest significantly progressed GGOs and consolidations with increased volume during the early-phase CT follow-up.