全脑分割方法Fastsurfer复现

首先使用较多的脑部分析工具包括spm,FSL以及使用较多的Freesurfer.

Freesurfer是被使用最多的，可以很好的处理T1影像，做VBM,SBM等分析。就是有一个缺点就是费时。一个病例的处理时间可能长达8小时左右。2020年提出的Fastsurfer将深度学习的方法，改进了皮层重建的方法使这一任务所需时间大大减少。

Fastsurfer:《FastSurfer - A fast and accurate deep learning based neuroimaging pipeline》

相关介绍：FastSurfer深度学习脑部成像工具：1分钟脑部分割，1小时皮质重建！

在开始之前确保你已经安装好freesurfer并下载好Fastsurfer的代码并配置好环境。

1.single_seg#仅做单个全脑分割

终端打开cd path/to/fastsurfer/FastsurferCNN

export FREESURFER_HOME=/usr/local/freesurfer
source $FREESURFER_HOME/SetUpFreeSurfer.sh
export FASTSURFER_HOME=/home/vision/xtt/FastSurfer

python3 eval.py --i_dir /home/vision/xtt/FastSurfer \ ##输入T1路径
--o_dir /home/vision/xtt/FastSurfer \ ##输出分割结果路径
--t subject2 \  ##T1前一级文件名
--log temp_Competitive.log \
--network_sagittal_path ../checkpoints/Sagittal_Weights_FastSurferCNN/ckpts/Epoch_30_training_state.pkl \
--network_coronal_path ../checkpoints/Coronal_Weights_FastSurferCNN/ckpts/Epoch_30_training_state.pkl \
--network_axial_path ../checkpoints/Axial_Weights_FastSurferCNN/ckpts/Epoch_30_training_state.pkl 

全脑分割结果

2.multiple seg

终端打开cd path/to/fastsurfer/FastsurferCNN

python3 eval.py --i_dir /home/vision/xxx/xxx/GroupA/fixdata \
--o_dir /home/vision/xxx/xxx/GroupA/fixdata \
--in_name T1.nii.gz \
--log temp_Competitive.log \
--network_sagittal_path ../checkpoints/Sagittal_Weights_FastSurferCNN/ckpts/Epoch_30_training_state.pkl \
--network_coronal_path ../checkpoints/Coronal_Weights_FastSurferCNN/ckpts/Epoch_30_training_state.pkl \
--network_axial_path ../checkpoints/Axial_Weights_FastSurferCNN/ckpts/Epoch_30_training_state.pkl 

3.皮质分区与皮层重建

将分割结果重建一个文件夹MRI放进去

#single surfer###############
export FREESURFER_HOME=/usr/local/freesurfer
source $FREESURFER_HOME/SetUpFreeSurfer.sh
export FASTSURFER_HOME=/home/vision/xtt/FastSurfer
datadir=/home/vision/xtt/seg
segdir=/home/vision/xtt/seg
targetdir=/home/vision/xtt/seg 
./recon-surf.sh --sid subject13 \
                --sd $targetdir \
                --t1 $datadir/subject13/T1.nii.gz \
                --parallel \ 
                --py python3.6 

##多个
python3 eval.py --i_dir /home/vision/xtt/ALS/GroupA/fixdata \
--o_dir /home/vision/xtt/ALS/GroupA/fixdata \
--in_name T1.nii.gz \
--log temp_Competitive.log \
--network_sagittal_path ../checkpoints/Sagittal_Weights_FastSurferCNN/ckpts/Epoch_30_training_state.pkl \
--network_coronal_path ../checkpoints/Coronal_Weights_FastSurferCNN/ckpts/Epoch_30_training_state.pkl \
--network_axial_path ../checkpoints/Axial_Weights_FastSurferCNN/ckpts/Epoch_30_training_state.pkl 

结果查看：

皮层重建结果

freeview -v \
    mri/T1.mgz \
    mri/wm.mgz \
    mri/brainmask.mgz \
    mri/aseg.mgz:colormap=lut:opacity=0.2 \
    -f \
    surf/lh.white:edgecolor=blue \
    surf/lh.pial:edgecolor=red \
    surf/rh.white:edgecolor=blue \
    surf/rh.pial:edgecolor=red
#surfer view-	3D###########
freeview -f bert/surf/lh.pial:annot=aparc.annot:name=pial_aparc:visible=0 \
bert/surf/lh.pial:annot=aparc.a2009s.annot:name=pial_aparc_des:visible=0 \
bert/surf/lh.inflated:overlay=lh.thickness:overlay_threshold=0.1,3::name=inflated_thickness:visible=0 \
bert/surf/lh.inflated:visible=0 \
bert/surf/lh.white:visible=0 \
bert/surf/lh.pial \
--viewport 3d

全脑分割统计表，可以后续做SBM VBM