A deep-learning-based framework for severity assessment of COVID-19 with CT images

Abstract

Millions of positive COVID-19 patients are suffering from the pandemic around the world, a critical step in the management and treatment is severity assessment, which is quite challenging with the limited medical resources. Currently, several artificial intelligence systems have been developed for the severity assessment. However, imprecise severity assessment and insufficient data are still obstacles. To address these issues, we proposed a novel deep-learning-based framework for the fine-grained severity assessment using 3D CT scans, by jointly performing lung segmentation and lesion segmentation. The main innovations in the proposed framework include: 1) decomposing 3D CT scan into multi-view slices for reducing the complexity of 3D model, 2) integrating prior knowledge (dual-Siamese channels and clinical metadata) into our model for improving the model performance. We evaluated the proposed method on 1301 CT scans of 449 COVID-19 cases collected by us, our method achieved an accuracy of 86.7% for four-way classification, with the sensitivities of 92%, 78%, 95%, 89% for four stages. Moreover, ablation study demonstrated the effectiveness of the major components in our model. This indicates that our method may contribute a potential solution to severity assessment of COVID-19 patients using CT images and clinical metadata.

CT-based severity assessment for COVID-19 using weakly supervised non-local CNN

“… [25] Attention MIL 0.771 0.407 0.166 0.160 11 Li et al. [26] Multi-view Dual-Siamese CNN 0.802 0.392 0.154 0.147 12 Ouyang et al. [49] Dual sampling attention 0.813 0.386 0.149 0.145 13 Proposed work Non-local squeeze attention CNN 0.843 0.019 0.139 0.133 …”

“…The dual Siamese network proposed by Li et al. resulted in a high 81.3% R 2 score and 0.145 MAE as it analyzed multiple views, yet it requires complementary information from other clinical markers to assess accurately [26] . Of the attention models, dual sampler attention induced size-aware sampling and attention refinement, therefore, has matched performance closer to the proposed work [49] .…”

“…To optimize 3D CNN complexity for fine-grained severity assessment, Li et al. introduced an input slicing based on multi-view slicing [26] . Furthermore, the performance was enhanced by utilizing dual-Siamese channels and clinical metadata for prior knowledge transfer.…”

RESCOVIDTCNnet: A residual neural network-based framework for COVID-19 detection using TCN and EWT with chest X-ray images

“…The danger of COVID-19 disease and its spread prompted researchers to develop many automatic diagnostic methods using X-ray images. Many traditional machine learning techniques have been presented in the literature for the early diagnosis of COVID-19 ( Heidari et al, 2020 , Li et al, 2021 , Sharifrazi et al, 2021 , Júnior et al, 2021 , Khan et al, 2021 , Fan et al, 2021 , Karthik et al, 2021 ). The convolutional neural network (CNN) ( Le Cun et al, 2015 ), support vector machine (SVM) ( Cortes and Vapnik, 1995 ), residual exemplar local binary pattern (ResExLBP), iterative ReliefF ( Li et al, 2021 ), and Sobel filter ( Sobel and Feldman, 1968 ) have been used.…”

“…Recently, several studies were adapted to diagnose COVID diseases based on chest X-ray images. Transfer learning approaches relying on (CNN) can be used for classification, feature extraction ( Li et al, 2021 , Sharifrazi et al, 2021 , Minaee et al, 2020 , Demir, 2021 ), and transfer learning ( Zhang et al, 2020 , Abraham and Nair, 2020 , Jin et al, 2021 , Quan et al, 2021 , Das et al, 2021 , Ozcan, 2021 ). The summary of studies developed for the automated detection of COVID-19 using X-ray images is summarized in Table 1 .…”