Objective: To help improve radiologists’ efficacy of disease diagnosis in reading computed tomography (CT) images, this study aims to investigate the feasibility of applying a modified deep learning (DL) method as a new strategy to automatically segment disease-infected regions and predict disease severity. Methods: We employed a public dataset acquired from 20 COVID-19 patients, which includes manually annotated lung and infections masks, to train a new ensembled DL model that combines five customized residual attention U-Net models to segment disease infected regions followed by a Feature Pyramid Network model to predict disease severity stage. To test the potential clinical utility of the new DL model, we conducted an observer comparison study. First, we collected another set of CT images acquired from 80 COVID-19 patients and process images using the new DL model. Second, we asked two chest radiologists to read images of each CT scan and report the estimated percentage of the disease-infected lung volume and disease severity level. Third, we also asked radiologists to rate acceptance of DL model-generated segmentation results using a 5-scale rating method. Results: Data analysis results show that agreement of disease severity classification between the DL model and radiologists is >90% in 45 testing cases. Furthermore, >73% of cases received a high rating score (≥4) from two radiologists. Conclusion: This study demonstrates the feasibility of developing a new DL model to automatically segment disease-infected regions and quantitatively predict disease severity, which may help avoid tedious effort and inter-reader variability in subjective assessment of disease severity in future clinical practice.

No abstract

Quantification of infected lung volume using computed tomography (CT) images can play a critical role in predicting the severity of pulmonary infectious disease. Manual segmentation of infected areas from several CT image slices, however, is not efficient and viable in clinical practice. To assist clinicians in overcoming this challenge, we developed a new method to automatically segment and quantify the percentage of the infected lung volume. First, we used a public dataset of 20 COVID-19 patients, which consists of manually annotated lung and infection masks, to train a new joint deep learning (DL) model for lung and infection segmentation. As for lung segmentation, a Mask-RCNN model was applied to the lung volume with a novel postprocessing technique. Following that, an ensemble model with a customized residual attention UNet model and feature pyramid network (FPN) models was employed for infection segmentation. Next, we assembled another set of 80 CT scans of Covid-19 patients. Two chest radiologists manually evaluated each CT scan and reported the infected lung volume percentage using a customized graphical user interface (GUI). The developed DL-model was also employed to process these CT images. Then, we compared the agreement between the radiologist (manual) and model-based (automated) percentages of diseased regions. Additionally, the GUI was used to let radiologists rate acceptance of the DL-model generated segmentation results. Analyzing the results demonstrate that the agreement between manual and automated segmentation is >95% in 28 testing cases. Furthermore, >53% of testing cases received the top assessment rating scores from two radiologists (between 4-5 score). Thus, this study illustrates the feasibility of developing a DL-model based automated tool to effectively provide quantitative evaluation of infected lung regions to assist in improving the efficiency of radiologists in infection diagnosis.