# Difference between revisions of "Publications/puybareau.18.stacom"

### From LRDE

(Created page with "{{Publication | published = true | date = 2018-10-25 | authors = Élodie Puybareau, Zhou Zhao, Younes Khoudli, Yongchao Xu, Thierry Géraud | title = Left Atrial Segmentation...") |
|||

Line 2: | Line 2: | ||

| published = true | | published = true | ||

| date = 2018-10-25 | | date = 2018-10-25 | ||

− | | authors = Élodie Puybareau, Zhou Zhao, Younes Khoudli, Yongchao Xu, Thierry Géraud | + | | authors = Élodie Puybareau, Zhou Zhao, Younes Khoudli, Edwin Carlinet, Yongchao Xu, Thierry Géraud |

| title = Left Atrial Segmentation In a Few Seconds Using Fully Convolutional Network and Transfer Learning | | title = Left Atrial Segmentation In a Few Seconds Using Fully Convolutional Network and Transfer Learning | ||

| booktitle = Proceedings of the Workshop on Statistical Atlases and Computational Modelling of the Heart (STACOM), in conjunction with MICCAI | | booktitle = Proceedings of the Workshop on Statistical Atlases and Computational Modelling of the Heart (STACOM), in conjunction with MICCAI | ||

| series = Lecture Notes in Computer Science | | series = Lecture Notes in Computer Science | ||

| publisher = Springer | | publisher = Springer | ||

− | | abstract = In this paper, we propose a fast automatic method that segments left atrial cavity from 3D GE-MRIs without any manual assistance, using a fully convolutional network (FCN) and transfer learning. | + | | abstract = In this paper, we propose a fast automatic method that segments left atrial cavity from 3D GE-MRIs without any manual assistance, using a fully convolutional network (FCN) and transfer learning. This FCN is the base network of VGG-16, pre-trained on ImageNet for natural image classification, and fine tuned with the training dataset of the MICCAI 2018 Atrial Segmentation 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 <math>n^{\text{th}}</math> slice of the volume to segment, we consider three imagescorresponding to the <math>(n-1)^{\text{th}}</math>, <math>n^{\text{th}}</math>and <math>(n+1)^{\text{th}}</math> 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 <math>n^{\text{th}}</math> slice. We process all slices, then stack the results to form the 3D output segmentation. With such a technique, the segmentation of the left atrial cavity on a 3D volume takes only a few seconds. We obtain a Dice score of 0.92 both on the training set in our experiments before the challenge, and on the test set of the challenge. |

+ | | lrdepaper = http://www.lrde.epita.fr/dload/papers/puybareau.18.stacom.pdf | ||

| lrdeprojects = Olena | | lrdeprojects = Olena | ||

| lrdekeywords = Image | | lrdekeywords = Image | ||

Line 16: | Line 17: | ||

@InProceedings<nowiki>{</nowiki> puybareau.18.stacom, | @InProceedings<nowiki>{</nowiki> puybareau.18.stacom, | ||

author = <nowiki>{</nowiki>\'Elodie Puybareau and Zhou Zhao and Younes Khoudli and | author = <nowiki>{</nowiki>\'Elodie Puybareau and Zhou Zhao and Younes Khoudli and | ||

− | Yongchao Xu and Thierry G\'eraud<nowiki>}</nowiki>, | + | Edwin Carlinet and Yongchao Xu and Thierry G\'eraud<nowiki>}</nowiki>, |

title = <nowiki>{</nowiki>Left Atrial Segmentation In a Few Seconds Using Fully | title = <nowiki>{</nowiki>Left Atrial Segmentation In a Few Seconds Using Fully | ||

Convolutional Network and Transfer Learning<nowiki>}</nowiki>, | Convolutional Network and Transfer Learning<nowiki>}</nowiki>, |

## Revision as of 18:38, 25 October 2018

- Authors
- Élodie Puybareau, Zhou Zhao, Younes Khoudli, Edwin Carlinet, Yongchao Xu, Thierry Géraud
- Where
- Proceedings of the Workshop on Statistical Atlases and Computational Modelling of the Heart (STACOM), in conjunction with MICCAI
- Type
- inproceedings
- Publisher
- Springer
- Projects
- Olena
- Keywords
- Image
- Date
- 2018-10-25

## Abstract

In this paper, we propose a fast automatic method that segments left atrial cavity from 3D GE-MRIs without any manual assistance, using a fully convolutional network (FCN) and transfer learning. This FCN is the base network of VGG-16, pre-trained on ImageNet for natural image classification, and fine tuned with the training dataset of the MICCAI 2018 Atrial Segmentation 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 slice of the volume to segment, we consider three imagescorresponding to the , and 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 slice. We process all slices, then stack the results to form the 3D output segmentation. With such a technique, the segmentation of the left atrial cavity on a 3D volume takes only a few seconds. We obtain a Dice score of 0.92 both on the training set in our experiments before the challenge, and on the test set of the challenge.

## Documents

## Bibtex (lrde.bib)

@InProceedings{ puybareau.18.stacom, author = {\'Elodie Puybareau and Zhou Zhao and Younes Khoudli and Edwin Carlinet and Yongchao Xu and Thierry G\'eraud}, title = {Left Atrial Segmentation In a Few Seconds Using Fully Convolutional Network and Transfer Learning}, booktitle = {Proceedings of the Workshop on Statistical Atlases and Computational Modelling of the Heart (STACOM), in conjunction with MICCAI}, year = 2018, series = {Lecture Notes in Computer Science}, publisher = {Springer}, abstract = {In this paper, we propose a fast automatic method that segments left atrial cavity from 3D GE-MRIs without any manual assistance, using a fully convolutional network (FCN) and transfer learning. This FCN is the base network of VGG-16, pre-trained on ImageNet for natural image classification, and fine tuned with the training dataset of the MICCAI 2018 Atrial Segmentation 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 $n^{\text{th}}$ slice of the volume to segment, we consider three images, corresponding to the $(n-1)^{\text{th}}$, $n^{\text{th}}$, and $(n+1)^{\text{th}}$ 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 $n^{\text{th}}$ slice. We process all slices, then stack the results to form the 3D output segmentation. With such a technique, the segmentation of the left atrial cavity on a 3D volume takes only a few seconds. We obtain a Dice score of 0.92 both on the training set in our experiments before the challenge, and on the test set of the challenge.} }