M&Ms-winning nnU-Net ensemble

Task114_heart_mnms

DKFZ (Full, Isensee, Jager, Maier-Hein)

Cardiac MRI

Filter catalog by Modality:
Cardiac MRI

Cardiac chamber segmentation

Filter catalog by Disease / Trait:
Structural Heart & Cardiomyopathy

Segmentation

Filter catalog by Task Type:
Segmentation & Detection

Hybrid

Filter catalog by Architecture:
Hybrid / Multi-branch

PyTorch

Filter catalog by Framework:
PyTorch

CC BY 4.0

Filter catalog by License:
Open — Attribution

file_copy

View model files code View code

Ensemble of ten self-configuring nnU-Net models (five 2D, five 3D) that segments the left ventricle, right ventricle, and myocardium from short-axis cardiac cine MRI. Won the 2020 M&Ms challenge, a multi-centre, multi-vendor, multi-disease benchmark spanning scanners from four vendors and three countries, demonstrating strong generalization across acquisition protocols. Developed by DKFZ, the group behind the widely used nnU-Net framework.

memory Specifications

gavel License

CC BY 4.0

check_small Open source check_small Commercial use OK check_small Modifications allowed Attribution required No share-alike requirement

License for model weights only. Associated code may be licensed seperately, check code source for specific terms.

description Publication

Studying Robustness of Semantic Segmentation Under Domain Shift in cardiac MRI open_in_new

2021 · original paper

database Training & evaluation data

M&Ms 2020 (cardiac MRI) open_in_new

train

public 350 subjects · Spain; Germany; Canada

350 patients with hypertrophic and dilated cardiomyopathies plus healthy controls, scanned at clinical centres in Spain, Germany, and Canada.

science Capabilities & performance