Automatic scoring of COVID-19 severity in X-ray imaging based on a novel deep learning workflow

Abstract

In this study, we propose a two-stage workflow used for the segmentation and scoring of lung diseases. The workflow inherits quantification, qualification, and visual assessment of lung diseases on X-ray images estimated by radiologists and clinicians. It requires the fulfillment of two core stages devoted to lung and disease segmentation as well as an additional post-processing stage devoted to scoring. The latter integrated block is utilized, mainly, for the estimation of segment scores and computes the overall severity score of a patient. The models of the proposed workflow were trained and tested on four publicly available X-ray datasets of COVID-19 patients and two X-ray datasets of patients with no pulmonary pathology. Based on a combined dataset consisting of 580 COVID-19 patients and 784 patients with no disorders, our best-performing algorithm is based on a combination of DeepLabV3 + , for lung segmentation, and MA-Net, for disease segmentation. The proposed algorithms’ mean absolute error (MAE) of 0.30 is significantly reduced in comparison to established COVID-19 algorithms; BS-net and COVID-Net-S, possessing MAEs of 2.52 and 1.83 respectively. Moreover, the proposed two-stage workflow was not only more accurate but also computationally efficient, it was approximately 11 times faster than the mentioned methods. In summary, we proposed an accurate, time-efficient, and versatile approach for segmentation and scoring of lung diseases illustrated for COVID-19 and with broader future applications for pneumonia, tuberculosis, pneumothorax, amongst others.

Generative adversarial network for automatic quantification of Coronavirus disease 2019 pneumonia on chest radiographs

“…There were studies in interest in developing artificial intelligence to score the severity of COVID-19 pneumonia on chest radiographs and trying to prove that severity scored by artificial intelligence correlates with adverse outcomes in COVID-19 patients [18] , [19] , [20] , [21] . However, in some of those studies, COVID-19 pneumonia in the subdiaphragmatic and retrocardiac areas were neglected.…”

“…Notably, the radiologic extent was fully quantifiable on CT images using deep neural networks, highlighting the benefit of radiologic quantification in COVID-19. Deep neural networks were also applied to chest radiographs in COVID-19, but most research has focused on the automatic detection of COVID-19 [13] , [14] , [15] , whereas only a small proportion of work has investigated semi-quantitative assessments [16] , [17] , [18] , [19] , [20] , [21] , [22] . If pneumonia extent is fully quantifiable on chest radiographs, similar to CT images, the role of chest radiographs in managing COVID-19 would be maximized because it would be possible to identify and monitor at-risk patients regardless of image readers’ experience or variability in image interpretation across readers, particularly in resource-limited settings [8] .…”

Augmentation strategies for an imbalanced learning problem on a novel COVID-19 severity dataset

“…Some notable work regarding severity prediction with various machine learning models has been done on tabular data (clinical data, demographic data, etc.) 23 – 27 or image data 28 – 38 or a combination of both 39 – 41 .…”

Predicting intensive care need for COVID-19 patients using deep learning on chest radiography

“…[9][10][11][12] Various AI/ML methods have been developed to assess the severity/extent of disease [13][14][15][16] and predict the prognosis of the disease, 17 as well as for patient management in therapeutic treatment planning and monitoring patients' response. 13,18 Image-based studies of long-term COVID-19 effects on other organs, including the heart and brain, are also underway. 19 Accurate prognosis prediction for COVID-19 patients is crucial not only for implementing appropriate treatment for individual patients, but also for optimizing medical resource allocation during the pandemic.…”

“…AI/ML algorithms have been developed to differentiate COVID-19 pneumonia from non-COVID-19 pneumonia when RT-PCR is not readily available 9 – 12 Various AI/ML methods have been developed to assess the severity/extent of disease 13 – 16 and predict the prognosis of the disease, 17 as well as for patient management in therapeutic treatment planning and monitoring patients’ response 13 , 18 . Image-based studies of long-term COVID-19 effects on other organs, including the heart and brain, are also underway 19 …”