Researcher at Rajshahi University of Engineering and Technology
My research interests have developed through working on a range of applied research problems rather than a single application area. During my undergraduate studies in Electrical and Computer Engineering at Rajshahi University of Engineering and Technology (RUET), I became interested in how artificial intelligence and machine learning could be used to address practical problems across different fields. Therefore, I worked on projects in public health, agriculture, and medical imaging, which eventually led me toward research on human diseases and healthcare. My first research project was carried out during the COVID-19 pandemic. Motivated by the need for low-cost and deployable solutions for public health screening, I worked on an IoT-based mask detection system for crowded environments. The project combined a Raspberry Pi with deep learning to develop a practical screening system. Alongside this, while exploring agricultural applications of computer vision, I found that many existing approaches for grape leaf disease classification relied on computationally intensive models. This motivated me to develop a lightweight CNN-SVM model and use explainable AI to make its predictions more understandable. Both projects were published and presented at ICAEEE 2024. My interest in healthcare research grew further through my undergraduate thesis on Alzheimer’s disease detection from MRI images. I was particularly interested in the potential of deep learning to identify disease-related patterns in brain images, while recognizing that the reasoning behind such predictions can be difficult to understand. To address this, I developed a lightweight CNN and used SHAP, LIME, and Grad-CAM to examine its predictions and identify important brain regions. This work was published and presented at ICCIT 2024 and strengthened my interest in interpretable AI for medical imaging. I also worked on an integrated framework for consistent UAV remote sensing in agriculture. While studying existing work, I noticed that factors such as illumination, flight parameters, sensor configuration, and positioning were often considered separately, although their combined effects could influence the reliability of collected data. This motivated me to investigate these factors together and develop a framework for more consistent aerial data collection. The work was accepted at IEEE TENCON 2026.