AnatomyGen: Deep Anatomy Generation From Dense Representation With Applications in Mandible Synthesis

Amir H. Abdi, Heather Borgard, Purang Abolmaesumi, Sidney Fels
Proceedings of The 2nd International Conference on Medical Imaging with Deep Learning, PMLR 102:4-14, 2019.

Abstract

This work is an effort in human anatomy synthesis using deep models. Here, we introduce a deterministic deep convolutional architecture to generate human anatomies represented as 3D binarized occupancy maps (voxel-grids). The shape generation process is constrained by the 3D coordinates of a small set of landmarks selected on the surface of the anatomy. The proposed learning framework is empirically tested on the mandible bone where it was able to reconstruct the anatomies from landmark coordinates with the average landmark-to-surface error of 1.42 mm. Moreover, the model was able to linearly interpolate in the $\mathbb{Z}$-space and smoothly morph a given 3D anatomy to another. The proposed approach can potentially be used in semi-automated segmentation with manual landmark selection as well as biomechanical modeling. Our main contribution is to demonstrate that deep convolutional architectures can generate high fidelity complex human anatomies from abstract representations.

Cite this Paper


BibTeX
@InProceedings{pmlr-v102-abdi19a, title = {AnatomyGen: Deep Anatomy Generation From Dense Representation With Applications in Mandible Synthesis}, author = {Abdi, {Amir H.} and Borgard, Heather and Abolmaesumi, Purang and Fels, Sidney}, booktitle = {Proceedings of The 2nd International Conference on Medical Imaging with Deep Learning}, pages = {4--14}, year = {2019}, editor = {Cardoso, M. Jorge and Feragen, Aasa and Glocker, Ben and Konukoglu, Ender and Oguz, Ipek and Unal, Gozde and Vercauteren, Tom}, volume = {102}, series = {Proceedings of Machine Learning Research}, month = {08--10 Jul}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v102/abdi19a/abdi19a.pdf}, url = { http://proceedings.mlr.press/v102/abdi19a.html }, abstract = {This work is an effort in human anatomy synthesis using deep models. Here, we introduce a deterministic deep convolutional architecture to generate human anatomies represented as 3D binarized occupancy maps (voxel-grids). The shape generation process is constrained by the 3D coordinates of a small set of landmarks selected on the surface of the anatomy. The proposed learning framework is empirically tested on the mandible bone where it was able to reconstruct the anatomies from landmark coordinates with the average landmark-to-surface error of 1.42 mm. Moreover, the model was able to linearly interpolate in the $\mathbb{Z}$-space and smoothly morph a given 3D anatomy to another. The proposed approach can potentially be used in semi-automated segmentation with manual landmark selection as well as biomechanical modeling. Our main contribution is to demonstrate that deep convolutional architectures can generate high fidelity complex human anatomies from abstract representations.} }
Endnote
%0 Conference Paper %T AnatomyGen: Deep Anatomy Generation From Dense Representation With Applications in Mandible Synthesis %A Amir H. Abdi %A Heather Borgard %A Purang Abolmaesumi %A Sidney Fels %B Proceedings of The 2nd International Conference on Medical Imaging with Deep Learning %C Proceedings of Machine Learning Research %D 2019 %E M. Jorge Cardoso %E Aasa Feragen %E Ben Glocker %E Ender Konukoglu %E Ipek Oguz %E Gozde Unal %E Tom Vercauteren %F pmlr-v102-abdi19a %I PMLR %P 4--14 %U http://proceedings.mlr.press/v102/abdi19a.html %V 102 %X This work is an effort in human anatomy synthesis using deep models. Here, we introduce a deterministic deep convolutional architecture to generate human anatomies represented as 3D binarized occupancy maps (voxel-grids). The shape generation process is constrained by the 3D coordinates of a small set of landmarks selected on the surface of the anatomy. The proposed learning framework is empirically tested on the mandible bone where it was able to reconstruct the anatomies from landmark coordinates with the average landmark-to-surface error of 1.42 mm. Moreover, the model was able to linearly interpolate in the $\mathbb{Z}$-space and smoothly morph a given 3D anatomy to another. The proposed approach can potentially be used in semi-automated segmentation with manual landmark selection as well as biomechanical modeling. Our main contribution is to demonstrate that deep convolutional architectures can generate high fidelity complex human anatomies from abstract representations.
APA
Abdi, A.H., Borgard, H., Abolmaesumi, P. & Fels, S.. (2019). AnatomyGen: Deep Anatomy Generation From Dense Representation With Applications in Mandible Synthesis. Proceedings of The 2nd International Conference on Medical Imaging with Deep Learning, in Proceedings of Machine Learning Research 102:4-14 Available from http://proceedings.mlr.press/v102/abdi19a.html .

Related Material