Cerebellum CNN

<meta name="title" content="Cerebellum Parcellation with Convolutional Neural Networks"/>

Cerebellum Parcellation with Convolutional Neural Networks

This work originally appeared at Proceedings of SPIE Medical Imaging (SPIE-MI 2019), San Diego, CA, February 16 - 21, 2019. It can be downloaded as a Singularity image from Cerebellar CNN Segmentation Code. The download is 1.7GB. If you use this work please cite:

• S. Han, Y. He, A. Carass, S.H. Ying, and J.L. Prince, "Cerebellum parcellation with convolutional neural networks", Proceedings of SPIE Medical Imaging (SPIE-MI 2019), San Diego, CA, February 16 - 21, 2019. (doi)

What the Singularity image can do

The Singularity performs the following steps:

• ROBEX is used to estimate a brain mask. This mask is then smoothed to generate a brain weight image.
• N4 from ANTs is used to perform the bias field correction. The bias field is estimated using the weight image calculated above.
• The images are rigidly registered to the ICBM2009c nonlinear symmetric template using the ANTs package. The first MPRAGE image from the longitudinal series of a subject is registered to this template and other images with different contrasts/from the following sessions are registered to the first MPRAGE image.
• The cerebellum of an MNI-registered MPRAGE image is parcellated using the method described in "Shuo Han, et al., Cerebellum parcellation with convolutional neural networks, SPIE 2019 Medical Imaging Image Processing".

• Removing the neck should improve the results, such as using `robustfov` from fsl, but it is not done in this singularity image.

• (OPTIONAL) Generate a HTML report file. To generate the report, the images are converted to .png files which could occupy a large hard disk space.

• GNU Parallel is used to run jobs in parallel.

Installation

• Install singularity 2.6

File structure (important)

• The images should be organized in the BIDS format. For this singularity to run, the file structure is (be sure to create a subfolder `data/` under the processing folder):

<data_dir>/
   data/
      sub-<subject-label>/
         ses-<session-label>/
            anat/
               sub-<sub-label>_ses-<ses-label>_acq-mprage_[run-<run-label>]_T1w.nii.gz 

Note the "acq" label is required and run label is optional.

The processing will create subfolders "n4", "mni", "parc", and "post" under the folder "<data_dir>". The final parcellation results are in "<data_dir>/post".

"*n4.nii.gz" is the bias field corrected image
"*n4_mni.nii.gz" is the image in the MNI space
"*n4_mni_seg.nii.gz" is the parcellation from the deep networks
"*n4_mni_seg_post.nii.gz" is the final post-processed result.

ICV is calculated from the ROBEX mask. A ".tsv" file of the ICV is stored in "<data_dir>/n4" as "*_T1w_mask_icv.tsv".

The volumes of the cerebellar regions are calculated from the "*post.nii.gz" and stored in "<data_dir>/post" as "*post_volumes.tsv".

An optional report is stored as "<data_dir>/report.html".

Using a bash/python/perl etc. script to organize your data using soft (symbolic) link is recommended.

Example Usage

• Assuming the singularity image is ~/parc.simg

singularity run ~/parc.simg -d ~/processing

• Generate a report html

singularity run ~/parc.simg -d ~/processing -r

• Run multiple 16 jobs in parallel

singularity run ~/parc.simg -d ~/processing -r -j 16

• Print help

singularity run ~/parc.simg -h

Cerebellum labels

   "12":  "Corpus_Medullare",
   "33":  "Left_Lobules_I-III",
   "36":  "Right_Lobules_I-III",
   "43":  "Left_Lobule_IV",
   "46":  "Right_Lobule_IV",
   "53":  "Left_Lobule_V",
   "56":  "Right_Lobule_V",
   "63":  "Left_Lobule_VI",
   "60":  "Vermis_VI",
   "66":  "Right_Lobule_VI",
   "73":  "Left_Lobule_VIIAf",
   "76":  "Right_Lobule_VIIAf",
   "74":  "Left_Lobule_VIIAt",
   "70":  "Vermis_VII",
   "77":  "Right_Lobule_VIIAt",
   "75":  "Left_Lobule_VIIB",
   "78":  "Right_Lobule_VIIB",
   "83":  "Left_Lobule_VIIIA",
   "80":  "Vermis_VIII",
   "86":  "Right_Lobule_VIIIA",
   "84":  "Left_Lobule_VIIIB",
   "87":  "Right_Lobule_VIIIB",
   "93":  "Left_Lobule_IX",
   "90":  "Vermis_IX",
   "96":  "Right_Lobule_IX",
   "103": "Left_Lobule_X",
   "100": "Vermis_X",
   "106": "Right_Lobule_X"