Need Support to Build a Model Similar to MedGemma (Medical Imaging + Vision-Language Model)

I’m working on a medical imaging AI project and trying to build a model similar to MedGemma a vision-language model that can interpret DICOM images (CT/MRI/X-ray) and generate structured radiology-style reports.

If anyone has worked on something similar or has useful resources, papers, GitHub repos, or technical advice, I’d really appreciate it.
Your support will help me move toward building a clean, ethical, and open solution for medical imaging AI.

Thanks 🙏