A New Classification Model Based on Transfer Learning of DCNN and Stacknet for Fast Classification of Pneumonia Through X-Ray Images

A New Classification Model Based on Transfer Learning of DCNN and Stacknet for Fast Classification of Pneumonia Through X-Ray Images

Jalal Rabbah, Mohammed Ridouani, Larbi Hassouni
Copyright: © 2023 |Pages: 23
DOI: 10.4018/IJRQEH.326765
Article PDF Download
Open access articles are freely available for download


Coronavirus has spread worldwide, with over 688 million confirmed cases and 6.8 million deaths. The results could be important as containment restrictions begin to be relaxed and we are not immune to new strains. They underscore the need to introduce increasingly effective techniques to deal with such a spread and help identify new infections more quickly, at a reasonable cost and with a minimum error rate. Machine learning models constitute a new approach, used increasingly in this field. In this proposed work, the authors built a classification model named CovStacknet based on StackNet metamodeling methodology combined with the deep convolutional neural network as the basis for feature extraction from x-ray images. Firstly, the proposed model used VGG16 as a transfer learning of deep convolutional neural networks and achieved an accuracy score of 98%. Secondly, the proposed model is extended to evaluate four other deep convolutional neural networks, ResNet-50, Inception-V3, MobileNet-V2 and DenseNet, and ResNet-50, has achieved the best performance.
Article Preview


Coronavirus is an infectious disease caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). Considering the degree of its spread worldwide, World Health Organization Director-General declared the 2019-nCoV outbreak a public health emergency of international concern on 30 January 2020 and a pandemic on 11 March 2020. The World Health Organization leads international coordination to country readiness to research and innovation to limit transmission, provide early care, communicate critical information, and minimize social and economic impacts. The choice of Covid-19 to name the novel coronavirus disease is to guard against using other names that might be inaccurate or stigmatizing. Since the early weeks of the pandemic, thousands of researchers tried to stop the Covid -19 outbreak by developing easy-to-apply diagnostics, accelerating existing vaccine candidates, and preventing infection. Common symptoms include fever, cough, fatigue, shortness of breath, and loss of sense of smell. Complications may include pneumonia and acute respiratory distress syndrome. Until now, Covid ‐19 has had no immunization or treatment. However, numerous continuous clinical preliminaries are assessing potential medicines. More than 688 million Covid-19 infected cases were confirmed in more than 230 countries until March 2023, including more than 6.8 million deaths, 660 million recovered, and 20 million active cases (World Health Organization, 2023). The standard test for current infection with SARS-CoV-2 uses RNA testing of respiratory secretions collected using a nasopharyngeal swab, though it is possible to test other samples. This test uses real-time reverse transcription polymerase chain response (i.e., RT-PCR), which detects the presence of viral RNA fragments (Wang et al., 2020). The test procedure is manual and is facing many issues, namely the shortage of tests available worldwide, it may fail to identify infected patients without symptoms and is a time-consuming process (Zhou et al., 2020). Because of the primary involvement of the respiratory system, chest computerized tomography (CT) is strongly recommended in suspected Covid-19 cases for both initial evaluation and follow-up.

Chest radiographs (CXR) are of little diagnostic value in the early stages, whereas CT findings may be present before symptom onset. Ground glass opacities (GGO) pattern is the most common finding in Covid-19 infections. They are usually multifocal, bilateral, and peripheral. Still, in the early phase of the disease, the GGO may present as a unifocal lesion, most commonly located in the inferior lobe of the right lung (Chassagnon et al., 2020). There are widespread bilateral GGO with a posterior predominance. Sometimes there are thickened interlobular and intralobular lines in combination with a ground glass pattern. CXR images can assist in the early detection of suspected Covid-19 cases, but the overlap with other infectious and inflammatory lung diseases can lead to misdiagnosis. It is, therefore, essential to take the necessary precautions to avoid the financial and health consequences of such a false screening. The call to artificial intelligence has become critical, more particularly in times of crisis such as the Covid-19 pandemic. Indeed, the increased growth of detected and suspected cases has implied a demand far exceeding the capacity of health institutions, even for the best health systems in the world, such as the United States, Italy or Spain.

Complete Article List

Search this Journal:
Volume 13: 1 Issue (2024)
Volume 12: 2 Issues (2023)
Volume 11: 4 Issues (2022)
Volume 10: 4 Issues (2021)
Volume 9: 4 Issues (2020)
Volume 8: 4 Issues (2019)
Volume 7: 4 Issues (2018)
Volume 6: 4 Issues (2017)
Volume 5: 4 Issues (2016)
Volume 4: 4 Issues (2015)
Volume 3: 4 Issues (2014)
Volume 2: 4 Issues (2013)
Volume 1: 4 Issues (2012)
View Complete Journal Contents Listing