Difference between revisions of "Publications/puybareau.18.brainles"

From LRDE

 
(One intermediate revision by the same user not shown)
Line 3: Line 3:
 
| date = 2018-11-05
 
| date = 2018-11-05
 
| authors = Élodie Puybareau, Guillaume Tochon, Joseph Chazalon, Jonathan Fabrizio
 
| authors = Élodie Puybareau, Guillaume Tochon, Joseph Chazalon, Jonathan Fabrizio
| title = Segmentation of Gliomas and Prediction of Patient Overall Survival: A Simple and Fast Procedure.
+
| title = Segmentation of Gliomas and Prediction of Patient Overall Survival: A Simple and Fast Procedure
 
| booktitle = Proceedings of the Workshop on Brain Lesions (BrainLes)in conjunction with MICCAI
 
| booktitle = Proceedings of the Workshop on Brain Lesions (BrainLes)in conjunction with MICCAI
 
| series = Lecture Notes in Computer Science
 
| series = Lecture Notes in Computer Science
  +
| volume = 11384
  +
| pages = 199 to 209
 
| publisher = Springer
 
| publisher = Springer
 
| abstract = In this paper, we propose a fast automatic method that seg- ments glioma without any manual assistance, using a fully convolutional network (FCN) and transfer learning. From this segmentation, we predict the patient overall survival using only the results of the segmentation and a home made atlas. The FCN is the base network of VGG-16pretrained on ImageNet for natural image classificationand fine tuned with the training dataset of the MICCAI 2018 BraTS Challenge. It relies on the "pseudo-3D" method published at ICIP 2017, which allows for segmenting objects from 2D color images which contain 3D information of MRI volumes. For each n th slice of the volume to segment, we consider three images, corresponding to the (n-1)th, nthand (n-1)th slices of the original volume. These three gray-level 2D images are assembled to form a 2D RGB color image (one image per channel). This image is the input of the FCN to obtain a 2D segmentation of the n th slice. We process all slices, then stack the results to form the 3D output segmentation. With such a technique, the segmentation of a 3D volume takes only a few seconds. The prediction is based on Random Forests, and has the advantage of not being dependant of the acquisition modality, making it robust to inter-base data.
 
| abstract = In this paper, we propose a fast automatic method that seg- ments glioma without any manual assistance, using a fully convolutional network (FCN) and transfer learning. From this segmentation, we predict the patient overall survival using only the results of the segmentation and a home made atlas. The FCN is the base network of VGG-16pretrained on ImageNet for natural image classificationand fine tuned with the training dataset of the MICCAI 2018 BraTS Challenge. It relies on the "pseudo-3D" method published at ICIP 2017, which allows for segmenting objects from 2D color images which contain 3D information of MRI volumes. For each n th slice of the volume to segment, we consider three images, corresponding to the (n-1)th, nthand (n-1)th slices of the original volume. These three gray-level 2D images are assembled to form a 2D RGB color image (one image per channel). This image is the input of the FCN to obtain a 2D segmentation of the n th slice. We process all slices, then stack the results to form the 3D output segmentation. With such a technique, the segmentation of a 3D volume takes only a few seconds. The prediction is based on Random Forests, and has the advantage of not being dependant of the acquisition modality, making it robust to inter-base data.
  +
| lrdepaper = http://www.lrde.epita.fr/dload/papers/puybareau.18.brainles.pdf
 
| lrdeprojects = Olena
 
| lrdeprojects = Olena
 
| lrdekeywords = Image
 
| lrdekeywords = Image
Line 13: Line 16:
 
| type = inproceedings
 
| type = inproceedings
 
| id = puybareau.18.brainles
 
| id = puybareau.18.brainles
  +
| identifier = doi:10.1007/978-3-030-11726-9_18
 
| bibtex =
 
| bibtex =
 
@InProceedings<nowiki>{</nowiki> puybareau.18.brainles,
 
@InProceedings<nowiki>{</nowiki> puybareau.18.brainles,
Line 18: Line 22:
 
Chazalon and Jonathan Fabrizio<nowiki>}</nowiki>,
 
Chazalon and Jonathan Fabrizio<nowiki>}</nowiki>,
 
title = <nowiki>{</nowiki>Segmentation of Gliomas and Prediction of Patient Overall
 
title = <nowiki>{</nowiki>Segmentation of Gliomas and Prediction of Patient Overall
Survival: <nowiki>{</nowiki>A<nowiki>}</nowiki> Simple and Fast Procedure.<nowiki>}</nowiki>,
+
Survival: <nowiki>{</nowiki>A<nowiki>}</nowiki> Simple and Fast Procedure<nowiki>}</nowiki>,
 
booktitle = <nowiki>{</nowiki>Proceedings of the Workshop on Brain Lesions (BrainLes),
 
booktitle = <nowiki>{</nowiki>Proceedings of the Workshop on Brain Lesions (BrainLes),
 
in conjunction with MICCAI<nowiki>}</nowiki>,
 
in conjunction with MICCAI<nowiki>}</nowiki>,
 
year = 2018,
 
year = 2018,
 
series = <nowiki>{</nowiki>Lecture Notes in Computer Science<nowiki>}</nowiki>,
 
series = <nowiki>{</nowiki>Lecture Notes in Computer Science<nowiki>}</nowiki>,
  +
volume = <nowiki>{</nowiki>11384<nowiki>}</nowiki>,
  +
pages = <nowiki>{</nowiki>199--209<nowiki>}</nowiki>,
 
publisher = <nowiki>{</nowiki>Springer<nowiki>}</nowiki>,
 
publisher = <nowiki>{</nowiki>Springer<nowiki>}</nowiki>,
 
abstract = <nowiki>{</nowiki>In this paper, we propose a fast automatic method that
 
abstract = <nowiki>{</nowiki>In this paper, we propose a fast automatic method that
Line 46: Line 52:
 
prediction is based on Random Forests, and has the
 
prediction is based on Random Forests, and has the
 
advantage of not being dependant of the acquisition
 
advantage of not being dependant of the acquisition
modality, making it robust to inter-base data.<nowiki>}</nowiki>
+
modality, making it robust to inter-base data.<nowiki>}</nowiki>,
  +
doi = <nowiki>{</nowiki>10.1007/978-3-030-11726-9_18<nowiki>}</nowiki>
 
<nowiki>}</nowiki>
 
<nowiki>}</nowiki>
   

Latest revision as of 16:01, 27 May 2021

Abstract

In this paper, we propose a fast automatic method that seg- ments glioma without any manual assistance, using a fully convolutional network (FCN) and transfer learning. From this segmentation, we predict the patient overall survival using only the results of the segmentation and a home made atlas. The FCN is the base network of VGG-16pretrained on ImageNet for natural image classificationand fine tuned with the training dataset of the MICCAI 2018 BraTS Challenge. It relies on the "pseudo-3D" method published at ICIP 2017, which allows for segmenting objects from 2D color images which contain 3D information of MRI volumes. For each n th slice of the volume to segment, we consider three images, corresponding to the (n-1)th, nthand (n-1)th slices of the original volume. These three gray-level 2D images are assembled to form a 2D RGB color image (one image per channel). This image is the input of the FCN to obtain a 2D segmentation of the n th slice. We process all slices, then stack the results to form the 3D output segmentation. With such a technique, the segmentation of a 3D volume takes only a few seconds. The prediction is based on Random Forests, and has the advantage of not being dependant of the acquisition modality, making it robust to inter-base data.

Documents

Bibtex (lrde.bib)

@InProceedings{	  puybareau.18.brainles,
  author	= {\'Elodie Puybareau and Guillaume Tochon and Joseph
		  Chazalon and Jonathan Fabrizio},
  title		= {Segmentation of Gliomas and Prediction of Patient Overall
		  Survival: {A} Simple and Fast Procedure},
  booktitle	= {Proceedings of the Workshop on Brain Lesions (BrainLes),
		  in conjunction with MICCAI},
  year		= 2018,
  series	= {Lecture Notes in Computer Science},
  volume	= {11384},
  pages		= {199--209},
  publisher	= {Springer},
  abstract	= {In this paper, we propose a fast automatic method that
		  seg- ments glioma without any manual assistance, using a
		  fully convolutional network (FCN) and transfer learning.
		  From this segmentation, we predict the patient overall
		  survival using only the results of the segmentation and a
		  home made atlas. The FCN is the base network of VGG-16,
		  pretrained on ImageNet for natural image classification,
		  and fine tuned with the training dataset of the MICCAI 2018
		  BraTS Challenge. It relies on the "pseudo-3D" method
		  published at ICIP 2017, which allows for segmenting objects
		  from 2D color images which contain 3D information of MRI
		  volumes. For each n th slice of the volume to segment, we
		  consider three images, corresponding to the (n-1)th, nth,
		  and (n-1)th slices of the original volume. These three
		  gray-level 2D images are assembled to form a 2D RGB color
		  image (one image per channel). This image is the input of
		  the FCN to obtain a 2D segmentation of the n th slice. We
		  process all slices, then stack the results to form the 3D
		  output segmentation. With such a technique, the
		  segmentation of a 3D volume takes only a few seconds. The
		  prediction is based on Random Forests, and has the
		  advantage of not being dependant of the acquisition
		  modality, making it robust to inter-base data.},
  doi		= {10.1007/978-3-030-11726-9_18}
}