Computational Pathology
Whole-slide image analysis, tumor segmentation, tissue phenotyping, and quantitative pathology tools for more precise biomedical interpretation.
Research
Visual intelligence for biomedical problems that matter.
VIBE Lab develops AI methods for medical image analysis with a strong focus on practical biomedical use. We work on segmentation, representation learning, computational pathology, multimodal biomedical AI, and usable research software that can help models move from papers into real workflows.
Featured Challenge
Cross-Organ and Cross-Scanner Adenocarcinoma Segmentation (COSAS 2024)
COSAS 2024 sits directly inside our interest in robust medical image analysis: models must segment adenocarcinoma across changes in organ context, scanner acquisition, staining appearance, and data domain. We use this challenge as a research anchor for domain generalization, pathology segmentation, and reliable evaluation under real-world variation.
Research Directions
View publicationsBiomedical Image Analysis
Robust denoising, segmentation, microscopy analysis, and visual measurement methods designed around real data quality and acquisition constraints.
Multimodal Biomedical AI
Learning from images together with reports, clinical context, molecular signals, and metadata to make biomedical AI more grounded and useful.
Usable AI Software
We value aesthetics as part of research impact: software should be clear, reliable, easy to inspect, and pleasant enough that people actually use it.
Project Pipeline
Datasets & SoftwareStart from pathology, clinical, and biomedical questions where visual AI can remove friction.
Build methods that are robust across datasets, organs, scanners, and acquisition variation.
Release reusable datasets, models, interfaces, and demos when projects are ready to share.