VTMA

Administrative platform for veterinary thermography. Upload thermal images, AI analyses patterns, receive AAT-compliant reports in 5 minutes instead of 45. Use the assistant for further evaluation.

Comment

Inspiration

My mother's cousin is a veterinary thermographer who spends a lot of time per report. I built VTMA to reduce this to 5 minutes.

What it does

Analyzes thermographic images with Google Gemini 2.0 Flash, detects temperature asymmetries ≥1°C, searches similar cases via MongoDB vector search, generates AAT-compliant veterinary reports. Dutch and English.

How we built it

Next.js 15 + React 19 frontend. Gemini 2.0 Flash for image analysis. MongoDB Atlas vector search for knowledge retrieval. Google text-embedding-004 for semantic search.

Challenges we ran into

MongoDB vector index configuration. AAT compliance formatting.

Accomplishments that we're proud of

Fast report generation. AAT compliance. An veterinary thermographer that is exited for this product to be fully usable. Bilingual semantic search.

What we learned

Medical accuracy matters more than features. Vector embeddings capture clinical nuances. Domain expertise essential for prompt engineering.

What's next for VTMA

Direct FLIR camera integration. Multi-visit temperature tracking. Veterinary EMR exports. Expand to other animals beyond horses.

Built With

Share this project:

Updates