Voluma Bio

Voluma converts a single micro-CT image into trustworthy 3D metric results, and each measurement has a paper trail to back it up. A scientist chooses a workflow and loads an image. AI agents handle bone segmentation, region and landmark creation, measurements, and self-validation. The scientist reviews results visually (3D, slices) and types a correction. Voluma is a research tool, not a medical device. Demo: https://meta.voluma.bio GitHub: https://github.com/voluma-bio/voluma Models: https://huggingface.co/VolumaBio/mouse-knee-growth-plate Judge account: test@voluma.bio / VolumaDev-Test-2026! A single automated run produced 29 measurements, nearly matching the lab's values. - 1 h 49 min, no human present, 29 measurements: on a real knee image, it measured the knee-joint growth centre (IIOC), the cartilage sublayer of the bone plate, the trabeculae and the cortex, the osteophytes, and the femur diameter. - Near the lab's result: IIOC volume is within 1.3% (2.949 vs 2.988 mm³) of a gold-standard manual tracing; femur diameter is within 0.08 mm of the lab's value. - Validated by a lab expert: the bone segmentation was good, and the tibia alignment was correct, "The image looks good." - Traceable: each measurement includes a report with the files, regions, and validation images. A "To check" label marks cases the agents couldn't solve on their own. Why it matters: manual bone metrics slow osteoarthritis research. Osteoarthritis affects ~600 million people, and none of the drugs that are in development have been approved to