MRSegmentator-KonfAI

KonfAI-accelerated adaptation of MRSegmentator — multi-organ MRI segmentation (40 structures), built with KonfAI.

🧩 Model

Model	Modality	Labels	Ensemble
MRSegmentator	MRI	40	5

3D residual UNet · patch [96, 128, 160] · resampled to 1.5 mm.

🚀 Usage

pip install mrsegmentator-konfai
mrsegmentator-konfai segment -i input_mr.nii.gz -o output/

• Generic runner: konfai-apps infer VBoussot/MRSegmentator-KonfAI:MRSegmentator -i input_mr.nii.gz -o output/
• Interactive: SlicerKonfAI — the ⚙ Advanced dialog overrides patch size and batch size.

⚡ Performance & VRAM

Same input, same weights (5-fold ensemble), same PyTorch build (cu13.0), single NVIDIA RTX PRO 5000 (24 GB). Peak RAM = process-tree resident set; peak VRAM = over baseline.

Case (voxels)	Tool	Time	Peak RAM	Peak VRAM
S — 249 × 246 × 246	KonfAI	14 s	6.0 GB	13.0 GB
	Original	26 s	8.6 GB	3.7 GB
M — 533 × 390 × 177	KonfAI	25 s	7.5 GB	15.7 GB
	Original	65 s	14.6 GB	5.3 GB
L — 512 × 512 × 531	KonfAI	120 s	6.2 GB	16.7 GB
	Original	192 s	37.5 GB	14.6 GB

1.6–2.6× faster, 1.4–6.0× less host RAM, byte-identical to the CPU reassembly path. The GPU-resident accumulator trades more VRAM for the speed and low host RAM, while streaming keeps it bounded — on the large case host RAM stays at 6.2 GB where the original grows to 37.5 GB. The batch size is auto-selected from your free VRAM; override with --patch-size / --batch-size.

🔗 Links

• 🧠 Original MRSegmentator: github.com/hhaentze/MRSegmentator
• 🧠 KonfAI: github.com/vboussot/KonfAI
• 📦 PyPI: mrsegmentator-konfai