TY - GEN
T1 - Deep Learning for Quality Control of Subcortical Brain 3D Shape Models
AU - Petrov, Dmitry
AU - Gutman, Boris A.
AU - Kuznetsov, Egor
AU - Ching, Christopher R.K.
AU - Alpert, Kathryn
AU - Zavaliangos-Petropulu, Artemis
AU - Isaev, Dmitry
AU - Turner, Jessica A.
AU - van Erp, Theo G.M.
AU - Wang, Lei
AU - Schmaal, Lianne
AU - Veltman, Dick
AU - Thompson, Paul M.
N1 - Funding Information:
Acknowledgements. This work was funded in part by NIH BD2K grant U54 EB020403 and Russian Science Foundation grant 17-11-01390.
Publisher Copyright:
© 2018, Springer Nature Switzerland AG.
PY - 2018
Y1 - 2018
N2 - We present several deep learning models for assessing the morphometric fidelity of deep grey matter region models extracted from brain MRI. We test three different convolutional neural net architectures (VGGNet, ResNet and Inception) over 2D maps of geometric features. Further, we present a novel geometry feature augmentation technique based on parametric spherical mapping. Finally, we present an approach for model decision visualization, allowing human raters to see the areas of subcortical shapes most likely to be deemed of failing quality by the machine. Our training data is comprised of 5200 subjects from the ENIGMA Schizophrenia MRI cohorts, and our test dataset contains 1500 subjects from the ENIGMA Major Depressive Disorder cohorts. Our final models reduce human rater time by 46–70%. ResNet outperforms VGGNet and Inception for all of our predictive tasks.
AB - We present several deep learning models for assessing the morphometric fidelity of deep grey matter region models extracted from brain MRI. We test three different convolutional neural net architectures (VGGNet, ResNet and Inception) over 2D maps of geometric features. Further, we present a novel geometry feature augmentation technique based on parametric spherical mapping. Finally, we present an approach for model decision visualization, allowing human raters to see the areas of subcortical shapes most likely to be deemed of failing quality by the machine. Our training data is comprised of 5200 subjects from the ENIGMA Schizophrenia MRI cohorts, and our test dataset contains 1500 subjects from the ENIGMA Major Depressive Disorder cohorts. Our final models reduce human rater time by 46–70%. ResNet outperforms VGGNet and Inception for all of our predictive tasks.
KW - Deep learning
KW - Quality checking
KW - Subcortical shape analysis
UR - http://www.scopus.com/inward/record.url?scp=85057390144&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85057390144&partnerID=8YFLogxK
U2 - 10.1007/978-3-030-04747-4_25
DO - 10.1007/978-3-030-04747-4_25
M3 - Conference contribution
AN - SCOPUS:85057390144
SN - 9783030047467
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 268
EP - 276
BT - Shape in Medical Imaging - International Workshop, ShapeMI 2018, Held in Conjunction with MICCAI 2018, Proceedings
A2 - Lombaert, Hervé
A2 - Paniagua, Beatriz
A2 - Egger, Bernhard
A2 - Lüthi, Marcel
A2 - Reuter, Martin
A2 - Wachinger, Christian
PB - Springer Verlag
T2 - International Workshop on Shape in Medical Imaging, ShapeMI 2018 held in conjunction with 21st International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2018
Y2 - 20 September 2018 through 20 September 2018
ER -