Overview
Research
My research focuses on building reliable and explainable AI systems for medical image analysis. I work at the intersection of computer vision, deep learning, and clinical AI—designing models and evaluation methods that are practically useful in healthcare settings.
Current Research Focus
Medical Image Analysis
Developing deep learning models for disease detection and diagnosis from chest X-rays, ultrasound images, and other medical imaging modalities. Focus on reliable, generalizable models that perform under clinical constraints.
Open Questions
- →How can hybrid CNN-ViT models improve disease detection in limited-label medical settings?
- →What augmentation strategies are most effective for medical image classification under data scarcity?
Explainable AI for Healthcare
Investigating methods to make deep learning decisions interpretable for clinicians, including evaluation of saliency maps, attention visualization, and quantitative assessment of explanation quality.
Open Questions
- →How can saliency maps be rigorously evaluated against clinician annotations?
- →What XAI methods are most aligned with clinical reasoning for radiology AI?
Hybrid Architectures
Designing architectures that combine local feature extraction from CNNs with global context modeling from Vision Transformers, optimized for performance on small-to-medium medical datasets.
Open Questions
- →What fusion strategies work best for CNN-ViT hybrids in medical imaging?
- →How can attention mechanisms in ViTs be regularized to focus on clinically relevant regions?
Multimodal Medical AI
Exploring how combining multiple imaging modalities and clinical metadata can improve diagnostic performance and robustness in real-world clinical settings.
Open Questions
- →How can cross-modal attention improve joint reasoning across imaging modalities?
- →How does integrating clinical text with imaging features affect diagnostic accuracy?
Long-term PhD Research Direction
My long-term research vision is to develop AI systems that physicians can genuinely trust and use in clinical practice. Current deep learning models often achieve high accuracy on benchmark datasets but fail to generalize robustly or explain their decisions in ways clinicians can interpret.
I am motivated by three core research questions: How can we build hybrid architectures—combining the local feature extraction of CNNs with the global context modeling of Vision Transformers—that perform reliably even under limited labeled data? How can explainability methods be meaningfully evaluated so that saliency maps actually correspond to clinically relevant regions? And how can multimodal learning across imaging modalities improve diagnostic accuracy for diseases where single-modality data is insufficient?
During my PhD, I intend to pursue these questions in the context of chest X-ray analysis, ultrasound imaging, and potentially retinal imaging—domains where AI has clear potential to augment clinical workflows and improve patient outcomes in resource-limited settings.
Research Timeline
Hybrid CNN-ViT for Medical Imaging
Designing and evaluating hybrid architectures for tuberculosis and pneumonia detection on chest X-rays. Investigating XAI evaluation frameworks.
Real-time Object Detection Systems
Built and deployed YOLOv8-based emergency event detection API. Explored domain adaptation for surveillance use cases.
Medical Ultrasound Classification
Developed kidney disease classification models from ultrasound images. Explored transfer learning under limited annotated data.
ASEAN-India Hackathon — First Prize
Competed in and won the ASEAN-India Hackathon 2021 with an AI-based solution addressing a regional challenge.