MedeHealth/paligemma-chest-xray-vision
MedeHealth/paligemma-chest-xray-vision is a 4.3 billion parameter vision-language model fine-tuned for generating structured Portuguese chest X-ray reports. Built upon google/medgemma-4b-it, it specializes in analyzing medical images to produce detailed radiology reports including TÉCNICA, ACHADOS, and IMPRESSÃO sections. This model is optimized for medical image report generation, specifically identifying common chest pathologies in Brazilian Portuguese.
Loading preview...
Overview
MedeHealth/paligemma-chest-xray-vision is a specialized 4.3 billion parameter vision-language model designed for medical image report generation. It is fine-tuned from the google/medgemma-4b-it base model using QLoRA (4-bit quantization) to process chest X-ray images and generate structured radiology reports in Brazilian Portuguese.
Key Capabilities
- Medical Image Analysis: Analyzes chest X-ray images to identify relevant features.
- Structured Report Generation: Produces detailed radiology reports with specific sections: TÉCNICA (Technique), ACHADOS (Findings), and IMPRESSÃO (Impression).
- Pathology Identification: Capable of identifying common chest pathologies within the X-ray images.
- Language Specificity: Optimized for Portuguese (Brazilian) medical terminology and reporting standards.
Good For
- Research and Educational Purposes: Ideal for academic studies and learning environments focused on medical imaging and AI.
- Assisted Radiology Reporting: Can serve as a tool to generate initial drafts of chest X-ray reports, which require subsequent review and validation by a radiologist.
Limitations
This model is intended for research and educational use and should not be used as the sole basis for clinical decisions. All generated reports require professional radiologist review and validation.