TY - GEN
T1 - Dilated convolutional neural networks for cardiovascular MR segmentation in congenital heart disease
AU - Wolterink, Jelmer M.
AU - Leiner, Tim
AU - Viergever, Max A.
AU - Išgum, Ivana
N1 - Funding Information:
We gratefully acknowledge the support of NVIDIA Corporation with the donation of the Tesla K40 GPU used for this research.
Publisher Copyright:
© Springer International Publishing AG 2017.
PY - 2017
Y1 - 2017
N2 - We propose an automatic method using dilated convolutional neural networks (CNNs) for segmentation of the myocardium and blood pool in cardiovascular MR (CMR) of patients with congenital heart disease (CHD). Ten training and ten test CMR scans cropped to an ROI around the heart were provided in the MICCAI 2016 HVSMR challenge. A dilated CNNwith a receptive field of 131×131 voxels was trained for myocardium and blood pool segmentation in axial, sagittal and coronal image slices. Performance was evaluated within the HVSMR challenge. Automatic segmentation of the test scans resulted in Dice indices of 0.80 ± 0.06 and 0.93 ± 0.02, average distances to boundaries of 0.96 ± 0.31 and 0.89 ± 0.24 mm, and Hausdorff distances of 6.13 ± 3.76 and 7.07 ± 3.01mm for the myocardium and blood pool, respectively. Segmentation took 41.5 ± 14.7 s per scan. In conclusion, dilated CNNs trained on a small set of CMR images of CHD patients showing large anatomical variability provide accurate myocardium and blood pool segmentations.
AB - We propose an automatic method using dilated convolutional neural networks (CNNs) for segmentation of the myocardium and blood pool in cardiovascular MR (CMR) of patients with congenital heart disease (CHD). Ten training and ten test CMR scans cropped to an ROI around the heart were provided in the MICCAI 2016 HVSMR challenge. A dilated CNNwith a receptive field of 131×131 voxels was trained for myocardium and blood pool segmentation in axial, sagittal and coronal image slices. Performance was evaluated within the HVSMR challenge. Automatic segmentation of the test scans resulted in Dice indices of 0.80 ± 0.06 and 0.93 ± 0.02, average distances to boundaries of 0.96 ± 0.31 and 0.89 ± 0.24 mm, and Hausdorff distances of 6.13 ± 3.76 and 7.07 ± 3.01mm for the myocardium and blood pool, respectively. Segmentation took 41.5 ± 14.7 s per scan. In conclusion, dilated CNNs trained on a small set of CMR images of CHD patients showing large anatomical variability provide accurate myocardium and blood pool segmentations.
KW - Cardiovascular MR
KW - Congenital heart disease
KW - Deep learning
KW - Dilated convolutional neural networks
KW - Medical image segmentation
UR - http://www.scopus.com/inward/record.url?scp=85011307871&partnerID=8YFLogxK
U2 - 10.1007/978-3-319-52280-7_9
DO - 10.1007/978-3-319-52280-7_9
M3 - Conference contribution
AN - SCOPUS:85011307871
SN - 9783319522791
VL - 10129 LNCS
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 95
EP - 102
BT - Reconstruction, Segmentation, and Analysis of Medical Images - 1st International Workshops, RAMBO 2016 and HVSMR 2016 Held in Conjunction with MICCAI 2016, Revised Selected Papers
PB - Springer-Verlag
T2 - 1st International Workshops on Reconstruction and Analysis of Moving Body Organs, RAMBO 2016 and 1st International Workshops on Whole-Heart and Great Vessel Segmentation from 3D Cardiovascular MRI in Congenital Heart Disease, HVSMR 2016 Held in Conjunction with 19th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2016
Y2 - 17 October 2016 through 21 October 2016
ER -