# Numan Saeed

> Assistant Teaching Professor at the Mohamed bin Zayed University of Artificial Intelligence (MBZUAI). MBZUAI's first PhD graduate. Research on medical imaging, multimodal and vision-language models, and AI for healthcare and climate.

Personal website: https://www.numansaeed.com/

## About

Numan Saeed is an Assistant Teaching Professor at MBZUAI (Abu Dhabi, UAE), where he teaches graduate courses in computer vision and machine learning while leading an active research program. He was the first PhD graduate of MBZUAI. His work focuses on building accessible, impactful AI for real-world challenges in healthcare and climate, with an emphasis on medical image analysis, echocardiography, fetal ultrasound, vision-language foundation models, survival/prognosis prediction, and continual learning. He leads the HECKTOR head-and-neck tumor challenge and serves as an Area Chair for MICCAI 2026.

- Role: Assistant Teaching Professor, MBZUAI
- Teaching: Applied Computer Vision (MAAI7602, Fall 2026), Medical Multimodal Vision (CV8501, Fall 2025)
- Fields: Artificial Intelligence, Medical Imaging, Multimodal Learning, Computer Vision, Vision-Language Models, Echocardiography, AI for Healthcare, AI for Climate
- Faculty profile: https://mbzuai.ac.ae/study/faculty/numan-saeed/
- ORCID: https://orcid.org/0000-0002-6326-6434
- Google Scholar: https://scholar.google.ae/citations?user=VMPEU20AAAAJ&hl=en
- GitHub: https://github.com/numanai
- Hugging Face: https://huggingface.co/numansaeed
- LinkedIn: https://www.linkedin.com/in/numansaeed-pk/
- CV: https://www.numansaeed.com/cv.pdf

## Selected Publications

- **EchoSonar-R: A Multi-View Reasoning-Enabled Model for Disease Classification and Report Generation in Echocardiography** (European Conference on Computer Vision (ECCV), 2026). Darya Taratynova, Ahmed Aly, Numan Saeed, Mohammad Yaqub. [Paper](https://arxiv.org/pdf/2606.28164)
- **MobileFetalCLIP: Selective Repulsive KD for Fetal Ultrasound** (arXiv preprint, 2026). Numan Saeed, Fadillah Adamsyah Maani, Mohammad Yaqub. [Paper](https://arxiv.org/pdf/2603.05421) · [Code](https://github.com/numanai/MobileFetalCLIP) · [Website](https://www.numansaeed.com/mobilefetalclip/)
- **FETAL-GAUGE: A Benchmark for Assessing Vision-Language Models in Fetal Ultrasound** (International Conference on Learning Representations (ICLR), 2026). Hussain Alasmawi, Numan Saeed, Mohammad Yaqub. [Paper](https://arxiv.org/pdf/2512.22278) · [Code](https://github.com/BioMedIA-MBZUAI/FETAL-GAUGE)
- **DuPLUS: Dual-Prompt Vision-Language Framework for Universal Medical Image Segmentation and Prognosis** (IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), 2026). Numan Saeed, Tausifa Jan Saleem, Fadillah Adamsyah Maani, Muhammad F. Ridzuan, Hu Wang, Mohammad Yaqub. [Paper](https://arxiv.org/pdf/2510.03483) · [Code](https://github.com/BioMedIA-MBZUAI/DuPLUS)
- **CardioBench: Do Echocardiography Foundation Models Generalize Beyond the Lab?** (International Joint Conference on Artificial Intelligence (IJCAI), 2026). Darya Taratynova, Ahmed Aly, Numan Saeed, Mohammad Yaqub. [Paper](https://arxiv.org/pdf/2510.00520)
- **A Multimodal and Multi-centric Head and Neck Cancer Dataset for Segmentation, Diagnosis and Outcome Prediction** (Under Review at Nature Data Overview, 2025). Numan Saeed, Salma Hassan, Shahad Hardan, Ahmed Aly, Darya Taratynova, Umair Nawaz, Ufaq Khan, Muhammad Ridzuan, Vincent Andrearczyk, Adrien Depeursinge, Yutong Xie, et al.. [Paper](https://arxiv.org/pdf/2509.00367) · [Code](https://github.com/BioMedIA-MBZUAI/HECKTOR2025)
- **FetalCLIP: A Visual-Language Foundation Model for Fetal Ultrasound Image Analysis** (npj Digital Medicine (Nature Portfolio), 2026). Fadillah Maani, Numan Saeed, Tausifa Jan Saleem, Zaid Farooq, Hussain Alasmawi, Werner Diehl, Ameera Mohammad, Gareth Waring, Saudabi Valappil, Leanne Bricker, Mohammad Yaqub. [Paper](https://arxiv.org/pdf/2502.14807) · [Code](https://github.com/biomedia-mbzuai/fetalclip) · [Website](https://biomedia-mbzuai.github.io/FetalCLIP/)
- **SurvRNC: Learning Ordered Representations for Survival Prediction using Rank-N-Contrast** (International Conference on Medical Image Computing and Computer-Assisted Intervention (MICCAI), 2024). Numan Saeed, Muhammad Ridzuan, Fadillah Adamsyah Maani, Hussain Alasmawi, Karthik Nandakumar, Mohammad Yaqub. [Paper](https://arxiv.org/pdf/2403.10603) · [Code](https://github.com/numanai/SurvRNC)
- **PEMMA: Parameter-Efficient Multi-Modal Adaptation for Medical Image Segmentation** (International Conference on Medical Image Computing and Computer-Assisted Intervention (MICCAI), 2024). Nada Saadi, Numan Saeed, Mohammad Yaqub, Karthik Nandakumar. [Paper](https://arxiv.org/pdf/2404.13704)
- **CoReEcho: Continuous Representation Learning for 2D+ time Echocardiography Analysis** (International Conference on Medical Image Computing and Computer-Assisted Intervention (MICCAI), 2024). Fadillah Adamsyah Maani, Numan Saeed, Aleksandr Matsun, Mohammad Yaqub. [Paper](https://arxiv.org/pdf/2403.10164) · [Code](https://github.com/BioMedIA-MBZUAI/CoReEcho)
- **HuLP: Human-in-the-Loop for Prognosis** (International Conference on Medical Image Computing and Computer-Assisted Intervention (MICCAI), 2024). Muhammad Ridzuan, Mai Kassem, Numan Saeed, Ikboljon Sobirov, Mohammad Yaqub. [Paper](https://arxiv.org/pdf/2403.10164) · [Code](https://github.com/BioMedIA-MBZUAI/HuLP)
- **MGMT promoter methylation status prediction using MRI scans? An extensive experimental evaluation of deep learning models** (Medical Image Analysis (MedIA), 2023). Numan Saeed, Muhammad Ridzuan, Hussain Alasmawi, Ikboljon Sobirov, Mohammad Yaqub. [Paper](https://arxiv.org/pdf/2304.00774)

## Recent News

- 2026-08-01 - Promoted to Assistant Teaching Professor at MBZUAI: 🎓 Promoted to Assistant Teaching Professor at MBZUAI! I will continue leading my research program in medical imaging and multimodal learning alongside graduate teaching, and will be teaching Applied Computer Vision (MAAI7602) this Fall. (https://mbzuai.ac.ae/study/faculty/numan-saeed/)
- 2026-07-03 - EchoSonar-R Accepted at ECCV 2026: 🎉 Our paper "EchoSonar-R: A Multi-View Reasoning-Enabled Model for Disease Classification and Report Generation in Echocardiography" is accepted at ECCV 2026! EchoSonar-R is a multi-view reasoning vision-language model for joint disease classification and clinical report generation from echocardiography. (https://arxiv.org/abs/2606.28164)
- 2026-06-10 - 2 Papers Accepted at CVPR 2026 (Findings): 🎉 Two papers accepted to CVPR 2026 (Findings): CarePilot, a multi-agent framework for long-horizon computer task automation, and AnatomiX, an anatomy-aware grounded multimodal large language model for chest imaging.
- 2026-06-05 - FetalCLIP Accepted at npj Digital Medicine: 🎉 Our paper "FetalCLIP: A Visual-Language Foundation Model for Fetal Ultrasound Image Analysis" is accepted at npj Digital Medicine (Nature Portfolio)! FetalCLIP is the first vision-language foundation model for fetal ultrasound, pre-trained on 210,035 image-text pairs and achieving strong performance across classification, gestational age estimation, CHD detection, and segmentation.
- 2026-04-29 - CardioBench Accepted at IJCAI 2026: 🎉 Our paper "CardioBench: Do Echocardiography Foundation Models Generalize Beyond the Lab?" is accepted at IJCAI 2026! The benchmark unifies 8 datasets across 4 regression and 5 classification tasks for evaluating echocardiography foundation models with reproducible evaluation pipelines. (https://arxiv.org/pdf/2510.00520)
- 2026-03-06 - MobileFetalCLIP Website, Code, and Weights Released: 🚀 Released MobileFetalCLIP, a mobile fetal ultrasound model distilled from FetalCLIP for handheld deployment. The project page, paper, code, and model weights are now publicly available.
- 2026-01-31 - Selected as Area Chair for MICCAI 2026: 🎖️ Honored to be selected as an Area Chair for MICCAI 2026, the premier international conference on Medical Image Computing and Computer Assisted Intervention. (https://conferences.miccai.org/2026/en/default.asp)
- 2026-01-25 - FETAL-GAUGE Benchmark Accepted at ICLR 2026: 🎉 Our paper "FETAL-GAUGE: A Benchmark for Assessing Vision-Language Models in Fetal Ultrasound" is accepted to ICLR 2026! The benchmark comprises 42,000+ images and 93,000+ question-answer pairs spanning anatomical plane identification, visual grounding, fetal orientation, clinical view conformity, and diagnosis tasks.
- 2025-11-13 - 2 Papers Accepted at WACV 2026: 🎉 Two papers accepted to WACV 2026: MAFM³ for modular adaptation of foundation models using lightweight "skill set" components for multi-domain medical AI, and DuPLUS, a dual-prompt vision-language framework for universal segmentation across 10 datasets (30+ organs) with prognosis capabilities achieving C-index of 0.69.
- 2025-10-01 - CardioBench: Standardized Benchmark for Echocardiography FMs: 🫀 Released CardioBench, a comprehensive benchmark unifying 8 datasets across 4 regression and 5 classification tasks for evaluating echocardiography foundation models, with reproducible evaluation pipelines addressing key challenges in cardiac ultrasound. (https://arxiv.org/pdf/2510.00520)
- 2025-09-23 - HECKTOR 2025 Challenge Concluded Successfully at MICCAI: 🎯 Successfully concluded the 4th HECKTOR Challenge at MICCAI 2025 in Daejeon with winners across 3 tasks: MedAI (Segmentation - won NVIDIA DGX SPARK!), SIMS-LIFE (Recurrence Prediction), and Interstellar (HPV Diagnosis). Special thanks to NVIDIA for sponsorship. (https://www.linkedin.com/feed/update/urn:li:activity:7376503202642526209/)
- 2025-09-01 - HECKTOR Dataset Released: 1123 Head & Neck Cancer PET/CT Studies: 🗂️ Released a multimodal dataset of 1123 FDG-PET/CT studies from 10 international centers with expert annotations, radiotherapy dose maps, and clinical metadata for tumor segmentation, survival prediction, and HPV classification.
- 2025-07-25 - Survey Paper Accepted at ACM Computing Surveys: 📚 Comprehensive survey on continual learning in medical imaging accepted at ACM Computing Surveys. The paper addresses catastrophic forgetting and distribution shifts, providing taxonomy, practical insights, and future directions for the field.
- 2025-07-20 - Paper Accepted at MLHC 2025: 📄 FIVA accepted to MLHC 2025: A federated learning approach for universal CT segmentation that uses Bayesian inverse-variance aggregation for privacy-preserving collaboration with uncertainty quantification for clinical reliability.
- 2025-07-15 - 2 Papers Accepted at BMVC 2025: 🎉 Two papers accepted to BMVC 2025: TPA for fetal CHD classification achieving 85.40% F1 score using temporal prompt alignment, and InMerge for enhancing robustness of medical imaging models through in-model merging of convolutional kernels.

## Feeds

- RSS: https://www.numansaeed.com/rss.xml
- JSON Feed: https://www.numansaeed.com/feed.json

_Last generated: 2026-08-06._