Computed tomography–based COVID–19 triage through a deep neural network using mask–weighted global average pooling

Background There is an urgent need to find an effective and accurate method for triaging coronavirus disease 2019 (COVID-19) patients from millions or billions of people. Therefore, this study aimed to develop a novel deep-learning approach for COVID-19 triage based on chest computed tomography (CT) images, including normal, pneumonia, and COVID-19 cases. Methods A total of 2,809 chest CT scans (1,105 COVID-19, 854 normal, and 850 non-3COVID-19 pneumonia cases) were acquired for this study and classified into the training set (n = 2,329) and test set (n = 480). A U-net-based convolutional neural network was used for lung segmentation, and a mask-weighted global average pooling (GAP) method was proposed for the deep neural network to improve the performance of COVID-19 classification between COVID-19 and normal or common pneumonia cases. Results The results for lung segmentation reached a dice value of 96.5% on 30 independent CT scans. The performance of the mask-weighted GAP method achieved the COVID-19 triage with a sensitivity of 96.5% and specificity of 87.8% using the testing dataset. The mask-weighted GAP method demonstrated 0.9% and 2% improvements in sensitivity and specificity, respectively, compared with the normal GAP. In addition, fusion images between the CT images and the highlighted area from the deep learning model using the Grad-CAM method, indicating the lesion region detected using the deep learning method, were drawn and could also be confirmed by radiologists. Conclusions This study proposed a mask-weighted GAP-based deep learning method and obtained promising results for COVID-19 triage based on chest CT images. Furthermore, it can be considered a convenient tool to assist doctors in diagnosing COVID-19.

Computed tomography-based COVID-19 triage through a deep neural network using mask-weighted global average pooling.

Background: There is an urgent need to find an effective and accurate method for triaging coronavirus disease 2019 (COVID-19) patients from millions or billions of people. Therefore, this study aimed to develop a novel deep-learning approach for COVID-19 triage based on chest computed tomography (CT) images, including normal, pneumonia, and COVID-19 cases. Methods: A total of 2,809 chest CT scans (1,105 COVID-19, 854 normal, and 850 non-3COVID-19 pneumonia cases) were acquired for this study and classified into the training set (n = 2,329) and test set (n = 480). A U-net-based convolutional neural network was used for lung segmentation, and a mask-weighted global average pooling (GAP) method was proposed for the deep neural network to improve the performance of COVID-19 classification between COVID-19 and normal or common pneumonia cases. Results: The results for lung segmentation reached a dice value of 96.5% on 30 independent CT scans. The performance of the mask-weighted GAP method achieved the COVID-19 triage with a sensitivity of 96.5% and specificity of 87.8% using the testing dataset. The mask-weighted GAP method demonstrated 0.9% and 2% improvements in sensitivity and specificity, respectively, compared with the normal GAP. In addition, fusion images between the CT images and the highlighted area from the deep learning model using the Grad-CAM method, indicating the lesion region detected using the deep learning method, were drawn and could also be confirmed by radiologists. Conclusions: This study proposed a mask-weighted GAP-based deep learning method and obtained promising results for COVID-19 triage based on chest CT images. Furthermore, it can be considered a convenient tool to assist doctors in diagnosing COVID-19.

Outbreak of Mycoplasma pneumoniae at a military academy.

In 2019, an outbreak of Mycoplasma pneumoniae occurred at a military academy in China. The attack rate (10.08%,60/595) was significantly different among the units. High-intensity training and crowded environments to which cadets are exposed are the high risk factors for the outbreak of M. pneumoniae. In-time prevention and control measures effectively controlled the spread of the epidemic.

Clinical and computed tomographic imaging features of novel coronavirus pneumonia caused by SARS-CoV-2.

PURPOSE: To investigate the clinical and imaging characteristics of computed tomography (CT) in novel coronavirus pneumonia (NCP) caused by SARS-CoV-2. MATERIALS AND METHODS: A retrospective analysis was performed on the imaging findings of patients confirmed with COVID-19 pneumonia who had chest CT scanning and treatment after disease onset. The clinical and imaging data were analyzed. RESULTS: Fifty patients were enrolled, including mild type in nine, common in 28, severe in 10 and critically severe in the rest three. Mild patients (29 years) were significantly (P<0.03) younger than either common (44.5 years) or severe (54.7) and critically severe (65.7 years) patients, and common patients were also significantly (P<0.03) younger than severe and critically severe patients. Mild patients had low to moderate fever (<39.1 °C), 49 (98%) patients had normal or slightly reduced leukocyte count, 14 (28%) had decreased counts of lymphocytes, and 26 (52%) patients had increased C-reactive protein. Nine mild patients were negative in CT imaging. For all the other types of NCP, the lesion was in the right upper lobe in 30 cases, right middle lobe in 22, right lower lobe in 39, left upper lobe in 33 and left lower lobe in 36. The lesion was primarily located in the peripheral area under the pleura with possible extension towards the pulmonary hilum. Symmetrical lesions were seen in 26 cases and asymmetrical in 15. The density of lesion was mostly uneven with ground glass opacity as the primary presentation accompanied by partial consolidation and fibrosis. CONCLUSION: CT imaging presentations of NCP are mostly patchy ground glass opacities in the peripheral areas under the pleura with partial consolidation which will be absorbed with formation of fibrotic stripes if improved. CT scanning provides important bases for early diagnosis and treatment of NCP.

Exploring an Integrative Therapy for Treating COVID-19: A Randomized Controlled Trial.

OBJECTIVES: To develop a new Chinese medicine (CM)-based drug and to evaluate its safety and effect for suppressing acute respiratory distress syndrome (ARDS) in COVID-19 patients. METHODS: A putative ARDS-suppressing drug Keguan-1 was first developed and then evaluated by a randomized, controlled two-arm trial. The two arms of the trial consist of a control therapy (alpha interferon inhalation, 50 µg twice daily; and lopinavir/ritonavir, 400 and 100 mg twice daily, respectively) and a testing therapy (control therapy plus Keguan-1 19.4 g twice daily) by random number table at 1:1 ratio with 24 cases each group. After 2-week treatment, adverse events, time to fever resolution, ARDS development, and lung injury on newly diagnosed COVID-19 patients were assessed. RESULTS: An analysis of the data from the first 30 participants showed that the control arm and the testing arm did not exhibit any significant differences in terms of adverse events. Based on this result, the study was expanded to include a total of 48 participants (24 cases each arm). The results show that compared with the control arm, the testing arm exhibited a significant improvement in time to fever resolution (P=0.035), and a significant reduction in the development of ARDS (P=0.048). CONCLUSIONS: Keguan-1-based integrative therapy was safe and superior to the standard therapy in suppressing the development of ARDS in COVID-19 patients. (Trial registration No. NCT04251871 at www.clinicaltrials.gov ).

Persistent Viral Presence Determines the Clinical Course of the Disease in COVID-19.

BACKGROUND: The clinical management of coronavirus disease 2019 (COVID-19) is dependent on understanding the underlying factors that contribute to the disease severity. In the absence of effective antiviral therapies, other host immunomodulatory therapies such as targeting inflammatory response are currently being used without clear evidence of their effectiveness. Because inflammation is an essential component of host antiviral mechanisms, therapies targeting inflammation may adversely affect viral clearance and disease outcome. OBJECTIVE: To understand whether the persistent presence of the virus is a key determinant in the disease severity during COVID-19 and to determine whether the viral reactivation in some patients is associated with infectious viral particles. METHODS: The data for patients were available including the onset of the disease, duration of viral persistence, measurements of inflammatory markers such as IL-6 and C-reactive protein, chest imaging, disease symptoms, and their durations among others. Follow-up tests were performed to determine whether the viral negative status persists after their recovery. RESULTS: Our data show that patients with persistent viral presence (>16 days) have more severe disease outcomes including extensive lung involvement and requirement of respiratory support. Two patients who died of COVID-19 were virus-positive at the time of their death. Four patients demonstrated virus-positive status on the follow-up tests, and these patient samples were sent to viral culture facility where virus culture could not be established. CONCLUSIONS: These data suggest that viral persistence is the key determining factor of the disease severity. Therapies that may impair the viral clearance may impair the host recovery from COVID-19.