erasmusmc-bigr/gemma3-4b-lirads-features-lora
Updated • 1
AI & ML interests
None defined yet.
Recent Activity
View all activity
Organization Card
Erasmus MC — Biomedical Imaging Group Rotterdam (BIGR)
The Biomedical Imaging Group Rotterdam (BIGR) is a research group at Erasmus MC developing AI and machine learning methods for medical image analysis. Our work spans quantitative imaging, radiomics, and deep learning for clinical decision support across multiple organ systems and imaging modalities.
🌐 bigr.nl
Models
LI-RADS Imaging Feature Classification (Multiphase Liver MRI)
Fine-tuned vision-language model (LoRA) adapters for classifying LI-RADS imaging features on multiphase liver MRI, with lexicon-guided explanations. Developed as part of an EMMBIOME master's thesis by Amal Lahchim (2026).
| Model | Base model |
|---|---|
| medgemma-4b-lirads-features-lora | Google MedGemma 4B |
| gemma3-4b-lirads-features-lora | Google Gemma 3 4B |
Related resources: Pipeline code · LI-RADS Lexicon (Zenodo)
Contact
Dr. Martijn Starmans · Assistant Professor · Erasmus MC, Rotterdam m.starmans@erasmusmc.nl
datasets 0
None public yet