Imran Razzak

Associate Professor, Computational Biology

MBZUAI, Abu Dhabi, UAE

Focus: Personalized Medicine • Medical Imaging • Multimodal Clinical AI

Editorial: Associate Editor — IEEE TNNLS, Neural Networks, IEEE TCSS, IEEE JBHI

Short biography

Imran Razzak is an Associate Professor at MBZUAI (Abu Dhabi, UAE). His research centers on Personalized Medicine and Medical Imaging, with emphasis on building multimodal AI systems that combine visual understanding, clinical reasoning, and spatial grounding to support decision-making in healthcare.

Medical AI at a glance

Illustrative medical visualization highlighting multimodal reasoning and spatial grounding.

Illustrative medical visualization (decorative) multimodal • reasoning • spatial grounding

CT • Chest

Findings → localization → summary

MRI • Brain

Reasoning + report generation

Pathology • Slide

Cross-modality grounding

Ophthalmology • Fundus

VQA + diagnosis support

Research Interests

My work focuses on Personalized Medicine and Medical Imaging, building models that unify multimodal understanding (vision + language), clinical reasoning, and spatial grounding across diverse imaging modalities. I’m especially interested in trustworthy medical AI (robustness, uncertainty, and hallucination reduction), longitudinal modeling, and end-to-end clinical workflows.

Open positions: Looking for Visiting Students and Remote Internships on medical VLM and reasoning.

PhD students and postdocs for Medical Foundation Models, VLMs for clinical reasoning, and personalized/longitudinal healthcare AI. Please email your CV + a short research proposal.

What’s new?

🚀 We released MedMO (4B & 8B) — a generalist medical AI foundation model

MedMO (4B & 8B) unifies multimodal understanding, clinical reasoning, and spatial grounding across diverse medical imaging modalities. Unlike task-specific systems, MedMO delivers end-to-end medical imaging intelligence within a single unified architecture.

🔍 Core capabilities

Understanding & reasoning: VQA, text-based medical QA, diagnostic reasoning
Report intelligence: Radiology report generation and summarization
Spatial grounding: Disease localization with precise bounding boxes
Cross-modality localization: Radiology, pathology, ophthalmology, microscopy

Resources

📄 Paper 💻 Project Page 🤗 Hugging Face 💾 GitHub

[Feb, 2026] 4 papers on Medical VLM reasoning and hallucination accepted in CVPR.
[Jan, 2026] 4 papers accepted in The Web Conference.
📄 Split RAG
[Jan, 2026] 1 paper accepted in ICLR — Read our work.
[Jan, 2026] We’re excited to welcome Abhijit to the lab!
[Oct, 2025] 4 papers accepted in AAAI (2 Oral). See you in Singapore!
[Sep, 2025] 6 papers accepted in NeurIPS (1 Spotlight) and 1 NeurIPS workshop paper.
📄 Inference-time 📄 LLM 📄 Ophthalmic Surgery 📄 Mediation Pathways 📄 SSL 📄 UniViT
[Aug, 2025] 2 papers accepted in EMNLP.
[Aug, 2025] Our team ranked 1st in AIM 2025 Rip Current Segmentation (RipSeg) Challenge Report.
[July, 2025] 1 paper accepted in ACM MM and one in CIKM.
[July, 2025] We released OphNet-3D, the first extensive RGB-D dynamic 3D reconstruction dataset for ophthalmic surgery.
[April, 2025] 2 papers accepted in MICCAI and 2 papers in IEEE SMC.
[March, 2025] Paper accepted in CVPR (ORAL) and one in ACL (ORAL).
[March, 2025] 1 paper accepted in NPJ Digital Medicine.
[March, 2025] Our work "Retinal Pathways to Asthma Diagnosis: AI-Driven Insight" accepted in ARVO.