Introducing NV-Reason-CT Open 3D CT VLM for Radiologist Chain-of-Thought Reasoning

Summary

NVIDIA released NV-Reason-CT, a vision language model for 3D CT analysis that processes volumes as true 3D input rather than independent 2D slices. The model generates structured diagnostic reports covering 30 chest and 29 abdominal abnormalities and emulates radiologist chain-of-thought reasoning with systematic anatomical reviews and uncertainty articulation. Its architecture pairs a Qwen3.5-4B language model with a full 3D ViT encoder that processes volumes resampled to 192³ voxels at 2mm isotropic resolution using 8x8x8 patch tokens and 3D MRoPE for spatial token relationships. Training uses supervised fine-tuning on roughly 550,000 QA examples from CT-RATE, NIH CT, and CancerVerse followed by Group Relative Policy Optimization that refines reasoning quality based on diagnostic accuracy. NV-Reason-CT is an open research foundation not a cleared clinical product and sits within NVIDIA’s Medical AI ecosystem alongside NV-Reason-CXR, NV-Segment-CTMR, and NV-Generate-CTMR.

Takeaways

1. 1

   3D VLMs require native volumetric processing because flattening a CT volume into 2D slices loses the through-plane spatial relationships that define structures like masses and infiltrates.

2. 2

   NV-Reason-CT’s architecture uses a full 3D ViT encoder with 8x8x8 patches and 3D MRoPE to preserve volumetric token relationships, unlike models that adapt 2D encoders by treating slices independently.

3. 3

   The training pipeline combines supervised fine-tuning on 550,000 QA examples with GRPO reinforcement learning to refine chain-of-thought reasoning quality based on diagnostic accuracy.

FAQ