A deep learning approach to segmentation of the developing cortex in fetal brain MRI with minimal manual labeling

Ahmed E. Fetit, Amir Alansary, Lucilio Cordero-Grande, John Cupitt, Alice B. Davidson, A. David Edwards, Joseph V. Hajnal, Emer Hughes, Konstantinos Kamnitsas, Vanessa Kyriakopoulou, Antonios Makropoulos, Prachi A. Patkee, Anthony N. Price, Mary A. Rutherford, Daniel Rueckert
Proceedings of the Third Conference on Medical Imaging with Deep Learning, PMLR 121:241-261, 2020.

Abstract

We developed an automated system based on deep neural networks for fast and sensitive 3D image segmentation of cortical gray matter from fetal brain MRI. The lack of extensive/publicly available annotations presented a key challenge, as large amounts of labeled data are typically required for training sensitive models with deep learning. To address this, we: (i) generated preliminary tissue labels using the {\em Draw-EM} algorithm, which uses Expectation-Maximization and was originally designed for tissue segmentation in the neonatal domain; and (ii) employed a human-in-the-loop approach, whereby an expert fetal imaging annotator assessed and refined the performance of the model. By using a hybrid approach that combined automatically generated labels with manual refinements by an expert, we amplified the utility of ground truth annotations while immensely reducing their cost (283 slices). The deep learning system was developed, refined, and validated on 249 3D T2-weighted scans obtained from the {\em Developing Human Connectome Project}’s fetal cohort, acquired at 3T. Analysis of the system showed that it is invariant to gestational age at scan, as it generalized well to a wide age range (21 � 38 weeks) despite variations in cortical morphology and intensity across the fetal distribution. It was also found to be invariant to intensities in regions surrounding the brain (amniotic fluid), which often present a major obstacle to the processing of neuroimaging data in the fetal domain.

Cite this Paper


BibTeX
@InProceedings{pmlr-v121-fetit20a, title = {A deep learning approach to segmentation of the developing cortex in fetal brain MRI with minimal manual labeling}, author = {Fetit, Ahmed E. and Alansary, Amir and Cordero-Grande, Lucilio and Cupitt, John and Davidson, Alice B. and Edwards, A. David and Hajnal, Joseph V. and Hughes, Emer and Kamnitsas, Konstantinos and Kyriakopoulou, Vanessa and Makropoulos, Antonios and Patkee, Prachi A. and Price, Anthony N. and Rutherford, Mary A. and Rueckert, Daniel}, booktitle = {Proceedings of the Third Conference on Medical Imaging with Deep Learning}, pages = {241--261}, year = {2020}, editor = {Arbel, Tal and Ben Ayed, Ismail and de Bruijne, Marleen and Descoteaux, Maxime and Lombaert, Herve and Pal, Christopher}, volume = {121}, series = {Proceedings of Machine Learning Research}, month = {06--08 Jul}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v121/fetit20a/fetit20a.pdf}, url = {https://proceedings.mlr.press/v121/fetit20a.html}, abstract = {We developed an automated system based on deep neural networks for fast and sensitive 3D image segmentation of cortical gray matter from fetal brain MRI. The lack of extensive/publicly available annotations presented a key challenge, as large amounts of labeled data are typically required for training sensitive models with deep learning. To address this, we: (i) generated preliminary tissue labels using the {\em Draw-EM} algorithm, which uses Expectation-Maximization and was originally designed for tissue segmentation in the neonatal domain; and (ii) employed a human-in-the-loop approach, whereby an expert fetal imaging annotator assessed and refined the performance of the model. By using a hybrid approach that combined automatically generated labels with manual refinements by an expert, we amplified the utility of ground truth annotations while immensely reducing their cost (283 slices). The deep learning system was developed, refined, and validated on 249 3D T2-weighted scans obtained from the {\em Developing Human Connectome Project}’s fetal cohort, acquired at 3T. Analysis of the system showed that it is invariant to gestational age at scan, as it generalized well to a wide age range (21 � 38 weeks) despite variations in cortical morphology and intensity across the fetal distribution. It was also found to be invariant to intensities in regions surrounding the brain (amniotic fluid), which often present a major obstacle to the processing of neuroimaging data in the fetal domain.} }
Endnote
%0 Conference Paper %T A deep learning approach to segmentation of the developing cortex in fetal brain MRI with minimal manual labeling %A Ahmed E. Fetit %A Amir Alansary %A Lucilio Cordero-Grande %A John Cupitt %A Alice B. Davidson %A A. David Edwards %A Joseph V. Hajnal %A Emer Hughes %A Konstantinos Kamnitsas %A Vanessa Kyriakopoulou %A Antonios Makropoulos %A Prachi A. Patkee %A Anthony N. Price %A Mary A. Rutherford %A Daniel Rueckert %B Proceedings of the Third Conference on Medical Imaging with Deep Learning %C Proceedings of Machine Learning Research %D 2020 %E Tal Arbel %E Ismail Ben Ayed %E Marleen de Bruijne %E Maxime Descoteaux %E Herve Lombaert %E Christopher Pal %F pmlr-v121-fetit20a %I PMLR %P 241--261 %U https://proceedings.mlr.press/v121/fetit20a.html %V 121 %X We developed an automated system based on deep neural networks for fast and sensitive 3D image segmentation of cortical gray matter from fetal brain MRI. The lack of extensive/publicly available annotations presented a key challenge, as large amounts of labeled data are typically required for training sensitive models with deep learning. To address this, we: (i) generated preliminary tissue labels using the {\em Draw-EM} algorithm, which uses Expectation-Maximization and was originally designed for tissue segmentation in the neonatal domain; and (ii) employed a human-in-the-loop approach, whereby an expert fetal imaging annotator assessed and refined the performance of the model. By using a hybrid approach that combined automatically generated labels with manual refinements by an expert, we amplified the utility of ground truth annotations while immensely reducing their cost (283 slices). The deep learning system was developed, refined, and validated on 249 3D T2-weighted scans obtained from the {\em Developing Human Connectome Project}’s fetal cohort, acquired at 3T. Analysis of the system showed that it is invariant to gestational age at scan, as it generalized well to a wide age range (21 � 38 weeks) despite variations in cortical morphology and intensity across the fetal distribution. It was also found to be invariant to intensities in regions surrounding the brain (amniotic fluid), which often present a major obstacle to the processing of neuroimaging data in the fetal domain.
APA
Fetit, A.E., Alansary, A., Cordero-Grande, L., Cupitt, J., Davidson, A.B., Edwards, A.D., Hajnal, J.V., Hughes, E., Kamnitsas, K., Kyriakopoulou, V., Makropoulos, A., Patkee, P.A., Price, A.N., Rutherford, M.A. & Rueckert, D.. (2020). A deep learning approach to segmentation of the developing cortex in fetal brain MRI with minimal manual labeling. Proceedings of the Third Conference on Medical Imaging with Deep Learning, in Proceedings of Machine Learning Research 121:241-261 Available from https://proceedings.mlr.press/v121/fetit20a.html.

Related Material