Robustness and Domain Shift
- Robustness under domain shift
- Staining, microscope, scanner, and illumination variability
- Lab and site variation
MICCAI 2026 Workshop
Toward Reliable AI in Hematology: Robust Red Blood Cell and Parasitemia Analysis
A MICCAI 2026 workshop on robust, fair, and clinically reliable AI for blood smear analysis
Red blood cell analysis and parasitemia detection are important tasks in clinical hematology and computational pathology. This workshop focuses on robust and trustworthy AI methods that can generalize across real-world domain shifts, support standardized evaluation, and accelerate clinically meaningful deployment.
Held in conjunction with MICCAI 2026. Exact workshop date and venue details are to be announced.
Reliability Focus
Domain shift, fairness, reproducibility
About
Building dependable hematology AI
AI methods for red blood cell analysis and blood-borne parasite detection have advanced quickly, yet reliability remains a major
Topics of Interest
A focused discussion space for rigorous methods, meaningful evaluation, and translational thinking across hematology image
Reliable Predictions
- Generalization and uncertainty estimation
- Calibration and selective prediction
- Bias measurement and mitigation
Learning Paradigms
- Self-supervised learning for cytology
- Foundation models for blood image analysis
- Promptable and interactive segmentation
Phenotyping and Morphology
- RBC morphology analysis
- Anemia-related phenotypes
- Dysmorphisms and hemoglobinopathies
- Sickle cell analysis
Parasitemia Analysis
- Parasitemia detection
- Malaria and other blood-borne parasites
Benchmarks and Translation
- Benchmarks and reproducibility
- Standardized evaluation protocols
- Translational and workflow considerations
Why This Workshop
Why this conversation matters now
Reliable AI in the real world
Performance on curated datasets is not enough. The field needs methods that remain dependable across clinical settings and deployment conditions.
Robust hematology microscopy
Blood smear analysis is highly sensitive to acquisition and preparation variability, making robustness and calibration central technical challenges.
Clinical translation
Better tools can support diagnosis, triage, and laboratory workflow only when they align with clinical needs and practical constraints.
Reproducibility and benchmarks
Shared evaluation protocols and rigorous benchmarks are key to comparing methods fairly and moving the field forward credibly.
Program
Preliminary Program
A focused 120-minute format designed to balance invited context, selected research talks, and discussion.
Total duration: 120 minutes
5 min
Welcome and scope
Opening remarks and workshop framing.
45 min
Invited talk
Featured invited presentation.
5 min
Q&A
Audience discussion following the invited talk.
3 x 10 min
Top-3 oral presentations
Selected peer-reviewed contributions presented orally.
Up to 6 x 1 min
Poster spotlight pitches
Short previews introducing poster contributions.
25 min
Poster discussion / networking
Interactive exchange around posters and emerging ideas.
4 min
Closing remarks and next steps
Wrap-up, outlook, and community follow-up.
Speakers
Invited Speaker TBD
The invited speaker is still being confirmed and will be announced once finalized.
Only one 45-minute invited talk will be scheduled in the quarter-day format. Final invited speaker selection is subject to availability.
TBD
Speaker details will be published here as soon as the invitation process is completed.
Call for Papers
Call for Papers
We welcome submissions on robust RBC morphology analysis, parasitemia detection, foundation models, self-supervised learning, fairness and generalization, reproducibility and benchmark design, and clinically relevant hematology AI.
Submitted papers will undergo peer review, with selected contributions featured as oral presentations and posters.
Submission will be handled via EasyChair. The submission link and full instructions will be shared once available.
Detailed submission instructions and important dates remain to be announced. Please stay tuned for updates after April 10, 2026.
Organizers
Organizing Team
The workshop is coordinated by an interdisciplinary team spanning academic research, translational AI, and biomedical image analysis.
General Chair
Dr. Andrea Loddo
University of Cagliari
Program Chair · Webmaster
Luca Zedda
University of Cagliari
Clinical and Translational Advisor
Dr. Ahmed Elnady Elsafty
PathOlOgics, LLC
Senior Advisory Chair
Dr. Carsten Marr
Helmholtz Munich
Program Chair
Dr. Ario Sadafi
Helmholtz Munich
Webmaster
Davide Antonio Mura
University of Cagliari
Keynote and Outreach Chair
Dr. Cecilia Di Ruberto
University of Cagliari
Contact
Get in touch
For workshop-related inquiries, please contact the corresponding organizer.
Corresponding organizer
Dr. Andrea Loddo
Dr. Andrea Loddo — andrea.loddo@unica.it
Webmasters: Luca Zedda, Davide Antonio Mura