Skip to content

tommy-qichang/ATMI

Repository files navigation

ATMI - Annotation Tool for Medical Image


CircleCI codecov License: GPL v3

The ATMI (Annotation Tool for Medical Image) provides a user friendly and comprehensive tool for medical image annotation especially for machine learning purpose.

The following features will be implemented in this system:

  1. Easy to install Install: Install ATMI just one click without install database, config environments etc.
  2. Multi-tasks capability: Create multiple instance(annotation tasks) under the same system. So, you don’t need to install multiple times.
  3. Audit process embeded: Audit process embedded. Different annotators could contour one case multiple times and arbitrate by the auditors.
  4. Support comprehensive annotation methods: Provide multiple annotation methods to meet the requirements of any medical annotation tasks, especially for machine learning settings.
  5. Support medical operations: Provide medical professional operations to help annotators view medical images correctly and efficiently. For instance, modify window width/ window level. Could extract multiple series/phases.
  6. Support iPad and finger gestures : Able to annotate using iPad and pencils. Finger gestures friendly.
  7. Easy to navigate between different slices/series/cases/tasks: Able to play medical images frame by frame and paste annotations from the last frame to the current frame.
  8. Natually integrate with Machine learning tasks: Easy to import the masks predicted by any machine learning algorithms. Easy to export the masks as Hdf5 files for future use.

Demo

ATMI Demo

How to install ATMI.

You need to install Python3 and Nodejs(v10) first.

  1. Setup an virtual environment for python 3 like [conda](https://docs.conda.io/en/latest/miniconda.html), and activate.
  2. Unzip/clone the ATMI code folder.
  3. pip install -r requirements.txt #under ATMI folder.
  4. cd atmi; npm install;npx webpack #build JS files
  5. sh run_flask.sh #start flask server.

Steps to setup development environment.

  1. Install Python3 and Nodejs.

  2. In China, please install cnpm for a better performance : npm install -g cnpm --registry=https://registry.npm.taobao.org

  3. CD root folder of ATMI project, run pip install tox

  4. run tox -r to install virtual environment and libraries. (include Python,Flask lib, python test related lib etc)

    4.1 If you are using IDE, please use the python interpreter under tox path: ./.tox/py37/bin/python.

    4.2 If you are using Command line, please direct activate py3 environment: cd ./.tox/py37/bin/ && source ./activate && cd -

  5. CD atmi folder: cd atmi and install packages, run npm install (or cnpm install ).

  6. RUN the project in debug model:

    6.1 open another terminal, and run npm run watch to launch webpack.

    6.2 run python __init__.py