You need to agree to share your contact information to access this model

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

Log in or Sign Up to review the conditions and access this model content.

DeepCAN-SR-swinViT-T1-CanonAug

Canine Brain MRI Super-Resolution (SwinUNETR-SR) — T1 + Canon-robust domain randomization

⚠️ Research / preview checkpoint. Trained with simulated Canon degradation on SKY (GE-family) T1 data — no real Canon scans.

A T1-weighted, Canon-robust adaptation of hwonheo/DeepCAN-SR-swinViT. LoRA (rank 32, α 64) on the swinViT encoder (backbone frozen, decoder + output head trained) with canon_aug LR-only domain randomization. Keeps its LoRA structure and loads into SwinUNETRSR(use_lora=True) (drop-in).

Domain randomization (canon_aug)

The LR input only is degraded during training — anisotropic through-plane blur (σ_z 1–3), contrast gamma (0.6–1.6), mild bias field, Gaussian noise — HR target untouched. Validation uses deterministic degradation (σ_z=2.0, γ=1.3) so val_psnr tracks Canon-restoration.

Performance

val_psnr 32.58 dB / val_ssim 0.9352 on the Canon-degraded validation set (early-stopped @ epoch 26).

Training

Base DeepCAN-SR-swinViT-T1 (T1-adapted; itself LoRA-adapted from DeepCAN-SR-swinViT T2)
Method LoRA (r=32, α=64) on swinViT + canon_aug; encoder frozen, decoder + out + adapters trained
Data 30 T1 HR subjects → 64³ LR→HR pairs @ 0.5 mm
Optimizer AdamW, LR 5e-5, weight decay 1e-5
Schedule cosine, 100 epochs (early-stopped @ 26)
Loss Combined L1 + SSIM (0.1) + gradient (0.05)
W&B https://wandb.ai/heohwon/DeepCAN-SegSR-public/runs/akmu1w5s

Usage

from huggingface_hub import snapshot_download
snapshot_download(repo_id="hwonheo/DeepCAN-SR-swinViT-T1-CA",
                  local_dir="src/checkpoint/DeepCAN-SR-swinViT-T1-CanonAug-LoRA")

from src.inference.models.sr_inferencer import SRInferencer
sr = SRInferencer(
    checkpoint_path="src/checkpoint/DeepCAN-SR-swinViT-T1-CanonAug-LoRA/DeepCAN-SR-swinViT-T1-CanonAug-LoRA.pth",
    device="cuda")

Known limitations

  • Trained on GE-family (SKY) T1 with simulated Canon degradation, not real Canon scans; small T1 corpus (30 subjects).

License

Research use only — see LICENSE. Contact: Hwon Heo, PhD (heohwon@gmail.com), BMC lab, Asan Medical Center.

Downloads last month
-
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support