Neural Networks
Computer Vision
Deep Learning
Medical Image Analysis
Data Mining
Undergraduate Thesis Work -
Title: COV19X-Net: A Convolutional Neural Network-based method for classifying COVID-19 in chest X-ray images
Supervised By: Dr. Mir Md. Jahangir Kabir, Professor, RUET
Work Description: This study presents the COV19X-Net model, which is based on the InceptionResNetV2 architecture, a pre-trained Convolutional Neural Network initially trained on the ImageNet dataset. COV19X-Net is designed to analyze chest X-rays, which reveal ground-glass opacities and other COVID-19-related lung features. The model was trained and evaluated using three public chest X-ray datasets, where COVID-19 is categorized as one of four classes. To the best of our knowledge, there had been very few studies (until the time of this research work) comparing automated methods for classifying these four chest X-ray classes across different datasets. Therefore, we assessed the performance of COV19X-Net against previous works using various metrics such as accuracy, precision, recall, and F1 score. The model demonstrated superior performance on unprocessed test data, with accuracy rates of 86.36%, 94.42%, and 91.88% on the three datasets.
Contributions:
Ensured the stability of the proposed model by performing implementation on three different sized datasets.
The proposed model was able to generate results that were satisfactory in reference to the multiclass classification which was a prior objective at the beginning of the research work.
The results of this research laid the groundwork for a more effective healthcare system.
Ensured a fair comparison and satisfactory improved performance with an existing state of the art approach.
Limitations:
The performance could be evaluated on more number of datasets.
Slightly irregular performance on datasets with different sizes and sources.
The work is not yet completely ready to be executed in healthcare systems.
Published Articles:
Md. Jamil Uddin, Mir Md. Jahangir Kabir. (2023). "COV19X-Net: A Convolutional Neural Network-based method for classifying COVID-19 in chest X-ray images." In Proceedings of the 26th International Conference on Computer and Information Technology (ICCIT 2023), Cox’s Bazar, Bangladesh. 📄
Accepted Conference Papers:
"An Ensemble Deep Learning Framework for Two-Stage Diabetic Retinopathy Diagnosis and Severity Screening."
Authors: Ahmed Noorim, Raina Nusrat Jahan, Md. Sabbir Al Ahsan, Sourav Adhikari, Md. Jamil Uddin.
Presented at the 10th International Conference on Information and Communication Technology for Intelligent Systems (ICTIS 2025),New York, USA, May 2025.
"Net Load Forecasting for Prosumers Using a Comparative and Ensemble Approach with Feature Selection"
Authors: Sadman Fayez, Md. Jamil Uddin, Md. Sabbir Al Ahsan, Mir Md. Jahangir Kabir
Accepted for presentation at the 28th International Conference on Computer and Information Technology, Cox's Bazar, Bangladesh, December 2025.
"Predicting Stock Movements with a Sentiment-Infused Model Case Studies on the Dhaka Stock Exchange"
Authors: Rahidur Rahman, Md. Sabbir Al Ahsan, Md. Jamil Uddin
Accepted for presentation at the 28th International Conference on Computer and Information Technology, Cox's Bazar, Bangladesh, December 2025.