VISTA3D Aorta Fine-tuned Checkpoint

This model repository stores the fine-tuned VISTA3D/NV-Segment-CT checkpoint used in the accompanying GitHub repository:

https://github.com/BZNR3/vista3d-aorta-finetune

Files

  • model_finetune_aorta_seed42.pt

SHA256:

3e4906c6aaf456f2a4ec4da6543b0e02ae3eec02803ef3535b43b6d0c19a779f

Training Summary

  • Base model: nvidia/NV-Segment-CT
  • Base checkpoint file: vista3d_pretrained_model/model.pt
  • Local aorta label mapping: 1 -> 6
  • Seed: 42
  • Epochs: 20
  • Best validation Dice: 0.8767 at epoch 7

Split

  • Train: AD-005, AD-006, AD-007, AD-008, AD-011, AD-012, AD-014, AD-015, AD-019, AD-020
  • Validation: AD-002, AD-013
  • Held-out test: AD-001, AD-003, AD-004, AD-009, AD-016, AD-021

Held-out Test

  • Mean Dice: 0.9123
  • Median Dice: 0.9092
  • Min Dice: 0.8274
  • Max Dice: 0.9683

The raw CTA images and labels are not included in this model repository.

Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support

Model tree for BZNR3/vista3d-aorta-finetune-checkpoint

Finetuned
(3)
this model