Cascaded-FCN
This repository contains the pre-trained models for a Cascaded-FCN in caffe and tensorflow that segments the liver and its lesions out of axial CT images and a python wrapper for dense 3D Conditional Random Fields 3D CRFs.
This work was published in MICCAI 2016 paper (arXiv link) titled :
Automatic Liver and Lesion Segmentation in CT Using Cascaded Fully Convolutional Neural Networks and 3D Conditional Random Fields
If you want to use our code we offer an docker image, which runs our code and has all dependencies installed including the correct caffe version. After having installed docker and nvidia docker:
sudo GPU=0 nvidia-docker run -v $(pwd):/data -P --net=host --workdir=/Cascaded-FCN -ti --privileged patrickchrist/cascadedfcn bash
And than start jupyter notebook and browse to localhost:8888
jupyter notebook
Please look at Readme and Documentation at https://github.com/FelixGruen/tensorflow-u-net
If you have used these models in your research please use the following BibTeX for citation :
@Inbook{Christ2016, title="Automatic Liver and Lesion Segmentation in CT Using Cascaded Fully Convolutional Neural Networks and 3D Conditional Random Fields", author="Christ, Patrick Ferdinand and Elshaer, Mohamed Ezzeldin A. and Ettlinger, Florian and Tatavarty, Sunil and Bickel, Marc and Bilic, Patrick and Rempfler, Markus and Armbruster, Marco and Hofmann, Felix and D'Anastasi, Melvin and Sommer, Wieland H. and Ahmadi, Seyed-Ahmad and Menze, Bjoern H.", editor="Ourselin, Sebastien and Joskowicz, Leo and Sabuncu, Mert R. and Unal, Gozde and Wells, William", bookTitle="Medical Image Computing and Computer-Assisted Intervention -- MICCAI 2016: 19th International Conference, Athens, Greece, October 17-21, 2016, Proceedings, Part II", year="2016", publisher="Springer International Publishing", address="Cham", pages="415--423", isbn="978-3-319-46723-8", doi="10.1007/978-3-319-46723-8_48", url="http://dx.doi.org/10.1007/978-3-319-46723-8_48" }
@ARTICLE{2017arXiv170205970C, author = {{Christ}, P.~F. and {Ettlinger}, F. and {Gr{"u}n}, F. and {Elshaera}, M.~E.~A. and {Lipkova}, J. and {Schlecht}, S. and {Ahmaddy}, F. and {Tatavarty}, S. and {Bickel}, M. and {Bilic}, P. and {Rempfler}, M. and {Hofmann}, F. and {Anastasi}, M.~D and {Ahmadi}, S.-A. and {Kaissis}, G. and {Holch}, J. and {Sommer}, W. and {Braren}, R. and {Heinemann}, V. and {Menze}, B.}, title = "{Automatic Liver and Tumor Segmentation of CT and MRI Volumes using Cascaded Fully Convolutional Neural Networks}", journal = {ArXiv e-prints}, archivePrefix = "arXiv", eprint = {1702.05970}, primaryClass = "cs.CV", keywords = {Computer Science - Computer Vision and Pattern Recognition, Computer Science - Artificial Intelligence}, year = 2017, }
@inproceedings{Christ2017SurvivalNetPP, title={SurvivalNet: Predicting patient survival from diffusion weighted magnetic resonance images using cascaded fully convolutional and 3D convolutional neural networks}, author={Patrick Ferdinand Christ and Florian Ettlinger and Georgios Kaissis and Sebastian Schlecht and Freba Ahmaddy and Felix Gr{"{u}n and Alexander Valentinitsch and Seyed-Ahmad Ahmadi and Rickmer Braren and Bjoern H. Menze}, booktitle={ISBI}, year={2017} }
This work uses 2 cascaded UNETs,
In step1, a UNET segments the liver from an axial abdominal CT slice. The segmentation output is a binary mask with bright pixels denoting the segmented object. By segmenting all slices in a volume we obtain a 3D segmentation.
(Optional) We enhance the liver segmentation using 3D dense CRF (conditional random field). The resulting enhanced liver segmentation is then used further for step2.
In step2 another UNET takes an enlarged liver slice and segments its lesions.
The input to both networks is 572x572 generated by applying reflection mirroring at all 4 sides of a 388x388 slice. The boundary 92 pixels are reflecting, resulting in (92+388+92)x(92+388+92) = 572x572.
An illustration of the pipeline is shown below :
For detailed Information have a look in our presentation
You can find the 3D CRF at 3DCRF-python. Please follow the installation description in the Readme.
These models are published with unrestricted use for research and educational purposes. For commercial use, please refer to the paper authors.
上一篇:LSTM-FCN
下一篇:matconvnet-fcn
还没有评论,说两句吧!
热门资源
seetafaceJNI
项目介绍 基于中科院seetaface2进行封装的JAVA...
spark-corenlp
This package wraps Stanford CoreNLP annotators ...
Keras-ResNeXt
Keras ResNeXt Implementation of ResNeXt models...
capsnet-with-caps...
CapsNet with capsule-wise convolution Project ...
inferno-boilerplate
This is a very basic boilerplate example for pe...
智能在线
400-630-6780
聆听.建议反馈
E-mail: support@tusaishared.com