Abstract
In the last five years, advances in processing power and computational efficiency in graphical processing units have catalyzed dozens of deep neural network segmentation algorithms for a variety of target tissues and malignancies. However, few of these algorithms preconfigure any biological context of their chosen segmentation tissues, instead relying on the neural network's optimizer to develop such associations de novo. We present a novel method for applying deep neural networks to the problem of glioma tissue segmentation that takes into account the structured nature of gliomas-edematous tissue surrounding mutually-exclusive regions of enhancing and non-enhancing tumor. We trained separate deep neural networks with a 3D U-Net architecture in a tree structure to create segmentations for edema, non-enhancing tumor, and enhancing tumor regions. Specifically, training was configured such that the whole tumor region including edema was predicted first, and its output segmentation was fed as input into separate models to predict enhancing and non-enhancing tumor. We trained our model on publicly available pre- and post-contrast T1 images, T2 images, and FLAIR images, and validated our trained model on patient data from an ongoing clinical trial.
| Original language | English (US) |
|---|---|
| Title of host publication | Medical Imaging 2018 |
| Subtitle of host publication | Image Processing |
| Publisher | SPIE |
| Volume | 10574 |
| ISBN (Electronic) | 9781510616370 |
| DOIs | |
| State | Published - Jan 1 2018 |
| Externally published | Yes |
| Event | Medical Imaging 2018: Image Processing - Houston, United States Duration: Feb 11 2018 → Feb 13 2018 |
Other
| Other | Medical Imaging 2018: Image Processing |
|---|---|
| Country | United States |
| City | Houston |
| Period | 2/11/18 → 2/13/18 |
Fingerprint
Keywords
- Deep Learning
- Glioma
- Neural Networks
- Segmentation
ASJC Scopus subject areas
- Electronic, Optical and Magnetic Materials
- Biomaterials
- Atomic and Molecular Physics, and Optics
- Radiology Nuclear Medicine and imaging
Cite this
Sequential neural networks for biologically informed glioma segmentation. / Beers, Andrew; Chang, Ken; Brown, James; Gerstner, Elizabeth; Rosen, Bruce; Kalpathy-Cramer, Jayashree.
Medical Imaging 2018: Image Processing. Vol. 10574 SPIE, 2018. 1057433.Research output: Chapter in Book/Report/Conference proceeding › Conference contribution
}
TY - GEN
T1 - Sequential neural networks for biologically informed glioma segmentation
AU - Beers, Andrew
AU - Chang, Ken
AU - Brown, James
AU - Gerstner, Elizabeth
AU - Rosen, Bruce
AU - Kalpathy-Cramer, Jayashree
PY - 2018/1/1
Y1 - 2018/1/1
N2 - In the last five years, advances in processing power and computational efficiency in graphical processing units have catalyzed dozens of deep neural network segmentation algorithms for a variety of target tissues and malignancies. However, few of these algorithms preconfigure any biological context of their chosen segmentation tissues, instead relying on the neural network's optimizer to develop such associations de novo. We present a novel method for applying deep neural networks to the problem of glioma tissue segmentation that takes into account the structured nature of gliomas-edematous tissue surrounding mutually-exclusive regions of enhancing and non-enhancing tumor. We trained separate deep neural networks with a 3D U-Net architecture in a tree structure to create segmentations for edema, non-enhancing tumor, and enhancing tumor regions. Specifically, training was configured such that the whole tumor region including edema was predicted first, and its output segmentation was fed as input into separate models to predict enhancing and non-enhancing tumor. We trained our model on publicly available pre- and post-contrast T1 images, T2 images, and FLAIR images, and validated our trained model on patient data from an ongoing clinical trial.
AB - In the last five years, advances in processing power and computational efficiency in graphical processing units have catalyzed dozens of deep neural network segmentation algorithms for a variety of target tissues and malignancies. However, few of these algorithms preconfigure any biological context of their chosen segmentation tissues, instead relying on the neural network's optimizer to develop such associations de novo. We present a novel method for applying deep neural networks to the problem of glioma tissue segmentation that takes into account the structured nature of gliomas-edematous tissue surrounding mutually-exclusive regions of enhancing and non-enhancing tumor. We trained separate deep neural networks with a 3D U-Net architecture in a tree structure to create segmentations for edema, non-enhancing tumor, and enhancing tumor regions. Specifically, training was configured such that the whole tumor region including edema was predicted first, and its output segmentation was fed as input into separate models to predict enhancing and non-enhancing tumor. We trained our model on publicly available pre- and post-contrast T1 images, T2 images, and FLAIR images, and validated our trained model on patient data from an ongoing clinical trial.
KW - Deep Learning
KW - Glioma
KW - Neural Networks
KW - Segmentation
UR - http://www.scopus.com/inward/record.url?scp=85047349114&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85047349114&partnerID=8YFLogxK
U2 - 10.1117/12.2293941
DO - 10.1117/12.2293941
M3 - Conference contribution
AN - SCOPUS:85047349114
VL - 10574
BT - Medical Imaging 2018
PB - SPIE
ER -