PDCOVIDNet: a parallel-dilated convolutional neural network architecture for detecting COVID-19 from chest X-ray images

Abstract

The COVID-19 pandemic continues to severely undermine the prosperity of the global health system. To combat this pandemic, effective screening techniques for infected patients are indispensable. There is no doubt that the use of chest X-ray images for radiological assessment is one of the essential screening techniques. Some of the early studies revealed that the patient’s chest X-ray images showed abnormalities, which is natural for patients infected with COVID-19. In this paper, we proposed a parallel-dilated convolutional neural network (CNN) based COVID-19 detection system from chest X-ray images, named as Parallel-Dilated COVIDNet (PDCOVIDNet). First, the publicly available chest X-ray collection fully preloaded and enhanced, and then classified by the proposed method. Differing convolution dilation rate in a parallel form demonstrates the proof-of-principle for using PDCOVIDNet to extract radiological features for COVID-19 detection. Accordingly, we have assisted our method with two visualization methods, which are specifically designed to increase understanding of the key components associated with COVID-19 infection. Both visualization methods compute gradients for a given image category related to feature maps of the last convolutional layer to create a class-discriminative region. In our experiment, we used a total of 2905 chest X-ray images, comprising three cases (such as COVID-19, normal, and viral pneumonia), and empirical evaluations revealed that the proposed method extracted more significant features expeditiously related to suspected disease. The experimental results demonstrate that our proposed method significantly improves performance metrics: the accuracy, precision, recall and F1 scores reach , , and , respectively, which is comparable or enhanced compared with the state-of-the-art methods. We believe that our contribution can support resistance to COVID-19, and will adopt for COVID-19 screening in AI-based systems.

What Can COVID-19 Teach Us about Using AI in Pandemics?

“…Upside notwithstanding, neither the CDC nor the American College of Radiology (ACR) mentioned AI-guided image analysis and its potential to discern image features beyond traditional human-guided analysis [ 51 ]. Several other models for AI-augmented analysis of portable X-rays were proposed due to the higher accessibility of X-ray machines vs. CT scanners [ 52 , 53 , 54 ]. For several years, AI has been applied to image analysis, but even more significant progress was made when large, publicly available data were made available to the scientific community [ 12 ].…”

ECOVNet: a highly effective ensemble based deep learning model for detecting COVID-19

“…An advanced custom CNN architecture, COVID-Net ( Wang, Lin & Wong, 2020 ) was implemented and tested using a large COVID-19 benchmark, but due to the large number of parameters, the computational overhead of this model is high. Another CNN-based modular architecture, named PDCOVIDNet, is proposed by Chowdhury, Rahman & Kabir (2020) , which consists of a parallel stack of multi-layer filter blocks in a cascade with a classification and visualization block. The authors reported the model was effective when compared with a number of well-known CNN architectures and showed precision and recall of 96.58% and 96.59%, respectively.…”

Peer Review #2 of "A novel perceptual two layer image fusion using deep learning for imbalanced COVID-19 dataset (v0.1)"

“…Moreover in used multi-resolution convolutional networks to learn the features and employed four different fusion methods that are CNN, Committee, late, and Full fusion strategies for lung classification and the results obtained are 95.01%, 97.17%, 97.92%, and 98.23% respectively. Parallel-dilated convolutional neural network (PDCNN) based COVID-19 classification system from chest X-ray images is presented by (Chowdhury et al, 2020) generated features are fused into the CNN network to produce the final prediction. They used 2905 chest X-ray images representing COVID-19, Normal, and Pneumonia cases with a reasonable accuracy reached to 96.58%.…”