Research Article
Survey on Chest X-Ray Based Classification Model Using Deep Transfer Learning for Covid-19 Detection
@INPROCEEDINGS{10.4108/eai.28-4-2025.2358014, author={R P Narmadha and P M Anurag and K Nithish and M Senthil Kumar}, title={Survey on Chest X-Ray Based Classification Model Using Deep Transfer Learning for Covid-19 Detection}, proceedings={Proceedings of the 4th International Conference on Information Technology, Civil Innovation, Science, and Management, ICITSM 2025, 28-29 April 2025, Tiruchengode, Tamil Nadu, India, Part II}, publisher={EAI}, proceedings_a={ICITSM PART II}, year={2025}, month={10}, keywords={identification of covid-19 classifier ensembles chest x-ray vgg19 ensemble methods deep transfer learning deep cnn}, doi={10.4108/eai.28-4-2025.2358014} }- R P Narmadha
P M Anurag
K Nithish
M Senthil Kumar
Year: 2025
Survey on Chest X-Ray Based Classification Model Using Deep Transfer Learning for Covid-19 Detection
ICITSM PART II
EAI
DOI: 10.4108/eai.28-4-2025.2358014
Abstract
The urgent requirement for fast and accurate diagnostic tools has been emphasized by the COVID-19 pandemic, in particular for the role of CXR in the detection of lung diseases such as COVID guise pneumonia and the COVID. It has led to advances in more powerful machine learning (ML) and deep learning (DL) methods as traditional systems continued to grapple with complexity and inefficiency. To improve feature extraction quality from CXRs, here the VGG19 neural network is applied. Finally SVM and RF classification methods are applied. Ensemble models like stacking, XGBoost and CatBoost are used to improve prediction performance by pooling collective strength. A new sum fusion method is introduced where decision probabilities of the classifiers are combined, resulting in both enhanced accuracy, sensitivity, and robustness and reduced bias. The model is evaluated on large datasets and outperforms the conventional approaches with better precision for diagnosis of lung diseases. Crucially, the model is not specific to COVID-19; it can identify other life-threatening lung diseases, such as pneumonia that needs to be treated promptly. By increasing the speed and the accuracy of the diagnostic workflows, this method has a strong potential for wider clinical and healthcare system's applications, both in pandemic and in routine conditions, leading finally to a benefit also for the patient in terms of a faster and more accurate analysis.
