Irfan Sadiq
Rahat
AI Researcher · Medical Imaging · Trustworthy AI
I build deep learning systems that clinicians can actually trust — not just models with high benchmark accuracy, but systems with formal uncertainty guarantees, interpretable decisions, and robustness to real-world clinical constraints. B.Tech CSE (AI & ML), VIT-AP University, India.
Active Research
Current Projects
MultiModal-SAM-BraTS
Dual-branch LoRA adaptation of SAM for multi-modal brain tumor segmentation (T1/T2/FLAIR MRI). Cross-modal attention fusion with missing-modality robustness. Adds <5% parameters over frozen SAM.
ConformalCXR
Post-hoc RAPS conformal prediction on CheXagent VLM for chest X-ray diagnosis with provable per-class coverage guarantees. Zero retraining required. Evaluated on CheXpert + NIH ChestX-ray14.
Research Output
Publications
Advancing Breast Ultrasound Diagnostics Through Hybrid Deep Learning Models
A Step Towards Automated Haematology: DL Models for Blood Cell Detection and Classification
Exploring DL Models for Accurate Alzheimer's Disease Classification Based on MRI Imaging
From Pixels to Pathology: The Power of CNNs in Detecting Tuberculosis
Convolutional Neural Networks in Malaria Diagnosis: A Study on Cell Image Classification
Deep Learning in Medical Imaging: A Case Study on Lung Tissue Classification
Enhancing Agricultural Sustainability with Deep Learning: Cauliflower Disease Classification
Using Deep Learning and Machine Learning: Real-Time Discernment and Diagnostics of Rice-Leaf Diseases in Bangladesh
Unraveling the Heterogeneity of Lower-Grade Gliomas: Deep Learning-Assisted FLAIR Segmentation and Genomic Analysis
Cassava Syndrome Scan: A Pioneering Deep Learning System for Accurate Cassava Leaf Disease Classification
Water Quality Assessment Through Predictive Machine Learning
Open Source
GitHub Repositories
MultiModal-SAM-BraTS
ResearchDual-branch LoRA-SAM for multi-modal brain tumor segmentation. Target: MICCAI 2026.
ConformalCXR
ResearchConformal prediction on CheXagent VLM for uncertainty-aware chest X-ray diagnosis. Target: IEEE TMI 2026.
MS-DSCCNet-BrainTumor
PublishedMulti-scale depthwise separable CNN with channel attention for brain tumor classification. IEEE DELCON 2025, Paper #234.
PneuNet-PediatricPneumonia
PublishedLightweight multi-scale attention CNN for pediatric pneumonia detection. High-sensitivity design. IEEE DELCON 2025, Paper #235.
TeaLeafNet-GWO
Nature 2025CNN with residual blocks for tea leaf disease detection. Published in Scientific Reports (Nature), IF 3.9. 99% accuracy.
Gallery
Activities & Highlights
Club Leadership & Events
Certificates & Awards
Your images are already in the repo root as personal.png, club1.png–club6.png, cert1.png–cert10.png ✓
Background
Academic Journey
Recognition
Awards & Highlights
Technical
Skills
Deep Learning
Languages & Tools
Medical AI
Get in Touch
Contact
Open to collaborations & opportunities
I am actively seeking funded MS/PhD positions in AI for healthcare, trustworthy AI,
and medical image analysis — particularly at labs working on foundation model adaptation,
uncertainty quantification, or clinical AI systems.
I am also open to research collaborations on medical imaging AI, conformal prediction,
or precision agriculture AI.
Based in Dhaka, Bangladesh. Available for remote collaboration globally.