Skip to content

Latest commit

 

History

History
116 lines (67 loc) · 4.21 KB

README.md

File metadata and controls

116 lines (67 loc) · 4.21 KB
Raw

AbART - ANTs based Atlas Registration Tool

Overview

This is a web based interactive tool to register MRI volumes to the Marmoset Atlas.

Users can load and visualize their MRI volumes (NIfTI-1 format), interactively reorient them (to allow correct registration), then submit for registration & transformation, and finally retrieve their image volume transformed in the Atlas space.

It is implemented following a multi-container architecture, since actual registration process are intended to run remotely, but it can also be used in "Desktop" mode where all components run on the same machine.


Run in "Desktop" mode

Prerequisites, that must be installed and working

  • For linux Users:
    git and docker

  • For macOS Users:
    git and Docker Desktop

  • For Windows Users:
    wsl2 and Docker Desktop

Important note:

Be sure that you and your organization comply with the licensing terms of Docker, especially if you are intending to use Docker Desktop.

see:

Extra steps for MS-Windows user only

AbART is run under WSL (Windows Subsystem for Linux), allowing to use the same commands as the Linux and macOS ones (described in next section).
Therefore, a few extra steps are necessary to properly setup wsl and Docker Desktop, as follow:

  • Install wsl (see https://docs.microsoft.com/en-us/windows/wsl/install), then set default version to 2 (wsl --set-default-version 2).

  • Then install a linux distro from Microsoft store (https://aka.ms/wslstore). Note: Since Docker WSL integration relies on glibc, we recommend to use Ubuntu 22.04LTS over the lighter Alpine distro that won't work by default for our purpose.

  • Set the linux distro including glibc as the default one (e.g. wsl --set-default Ubuntu-22.04), and ensure that Docker integration WSL with default distro is enabled (see https://docs.docker.com/desktop/windows/wsl/#enabling-docker-support-in-wsl-2-distros).

  • Note: Several reboots might be necessary after installation of the different components.

  • Start a new terminal running the default distro (shortcut [Windows]+R, then type wsl),

  • ( Depending on the distro used, git needs to be installed if not shipped by default. )

  • Execute the following commands in at the wsl terminal prompt.

Steps to follow

  1. clone this repo to get the scripts :
#shallow clone this repo
git clone --depth 1 https://github.com/cau-riken/abart.git

# pull published images before starting 
sudo docker pull rikencau/abart-worker:latest
sudo docker pull rikencau/abart-manager:latest
sudo docker pull rikencau/abart-ui:latest
  1. Start the Docker stack :
#go to the cloned repo directory
cd abart
./start.sh

Notes :

  • the UI should automatically open in your web-browser, at this url: http://localhost:9090/index.html
  • The containers can be stopped by pressing [Ctrl]+[C] key in the terminal window.
  • Manager process will produce logs on the terminal console while it is running.

Tested configurations:

  • Linux Ubuntu 22.04lts, Docker Engine v20.10.8 / Firefox 100 & Chromium 101.
  • macOs Big Sur 11.6.6, Docker Desktop v4.8.2 (Docker Engine v20.10.14) / Firefox 91.9.1esr & Chrome 101 & Safari 15.5, but for performance reasons, the use of Safari is not recommended.
  • Windows 10.0 (build 19043), Docker Desktop v4.9.0 (WSL2) / Firefox v101.0.1 & Edge v102.0.1245.39

Usage

Basics: load a NIfTI file, and switch view modes

AbART_loadAndViewModes-sub.mp4

2D slices view mode

AbART_2DSlices_Navigation-sub.mp4

3D slices view mode and brain model

AbART_3DSlices_Navigation-sub.mp4

Registration to Brain Atlas

AbART_registration_ext-sub.mp4

Guided alignment using Landmarks

AbART_autoAlignmentUsingPOI-sub.mp4