COVID-CXNet: Detecting COVID-19 in frontal chest X-ray images using deep learning

Abstract

One of the primary clinical observations for screening the novel coronavirus is capturing a chest x-ray image. In most patients, a chest x-ray contains abnormalities, such as consolidation, resulting from COVID-19 viral pneumonia. In this study, research is conducted on efficiently detecting imaging features of this type of pneumonia using deep convolutional neural networks in a large dataset. It is demonstrated that simple models, alongside the majority of pretrained networks in the literature, focus on irrelevant features for decision-making. In this paper, numerous chest x-ray images from several sources are collected, and one of the largest publicly accessible datasets is prepared. Finally, using the transfer learning paradigm, the well-known CheXNet model is utilized to develop COVID-CXNet. This powerful model is capable of detecting the novel coronavirus pneumonia based on relevant and meaningful features with precise localization. COVID-CXNet is a step towards a fully automated and robust COVID-19 detection system.

Keywords: COVID-19; CheXNet; Chest X-ray; Convolutional neural networks; Imaging features.

Fig. 1

Randomly selected frontal CXR images from different sources

Fig. 2

Dataset distribution. COVIDGR dataset with 426 images has the largest contribution in building our dataset

Fig. 3

Descriptive charts for categorical variables. a x-ray image view, b patient sex, c PCR test result, and d patient chance of survival

Fig. 4

Most reported symptoms from COVID patients

Fig. 5

Different image enhancement methods. a is the main image, b is the image with histogram equalization (HE), c is adaptive histogram equalization (AHE) applied on the image, and d is the image with contrast limited AHE

Fig. 6

BEASF with different hyperparameter values compared with original image and CLAHE

Fig. 7

The segmentation approach based on the U-Net

Fig. 8

A high-level illustration of the base model

Fig. 9

Co-occurrence of different CXR findings as a circular diagram by [54]

Fig. 10

COVID-CXNet model architecture based on the DenseNet-121 feature extractor as the backbone

Fig. 11

Learning Curves for base model trained on (a) 450 images, and (b) 600 images

Fig. 12

Grad-CAM heatmaps of the base model for 6 images of positive class. Important regions are wrong in most images, while classification scores are notably high

Fig. 13

Results of the base model trained on 3,400 images; a training loss changes, b training accuracy score changes, and c receiver operating characteristic (ROC) plot

Fig. 14

Model interpretability visualization by (a) Grad-CAM and (b) LIME image explanation

Fig. 15

DenseNet-121 fine-tuning curve over 10 epochs

Fig. 16

CheXNet probabilities of different classes for (a) a COVID-19 positive case, and (b) a normal case

Fig. 17

Grad-CAM visualization of the proposed model over sample cases

Fig. 18

Comparison between the CheXNet and the proposed model; a is the image with patchy opacities in the upper left zone, b and c are heatmaps of the CheXNet and the proposed COVID-CXNet, respectively

Fig. 19

Text removal effect on model results. Images on the right have dates and signs, which are concealed in the images on the left

Fig. 20

Grad-CAM visualization of the proposed model, trained with lung-segmented CXRs, over sample cases

Fig. 21

COVID-CXNet multiclass classification visualization results

Fig. 22

Comparison of learning curves for (a) training loss and (b) validation loss. Note that curves of the validation loss are smoothed for better intuition

Fig. 23

Grad-CAMs from COVID-CXNet

Fig. 24

Grad-CAMs from COVID-CXNet with lung segmentation module

Fig. 25

Grad-CAMs from multiclass COVID-CXNet