KNIME exploration, image analysis point of view



Introduction/potential interest

What is Knime?

From Knime wikipedia article:

The Konstanz Information Miner, is an open source data analytics, reporting and integration platform. KNIME integrates various components for machine learning and data mining through its modular data pipelining concept. A graphical user interface allows assembly of nodes for data preprocessing (ETL: Extraction, Transformation, Loading), for modeling and data analysis and visualization.

What is the interest of Knime for image analysis?

  • Possibility to build Heterogeneous pipeline combining python/R/matlab/ImageJ(Java)
  • Automation of the batch processing and its parallelization
  • Access to reporting and data analyzis function

What is the interest of KNIME for a service?

  • Unify complex pipeline in a single workflow that can be proposed in service portfolio and can be enhanced for more versatility
  • ease reuse of basic pipeline while avoiding to rebuild the infrastructure part of the analysis (reporting, batch processing, experimental condition handling for instance)

One see a view of Knime interface in Figure \ref{fig:GeneralView}

\label{fig:GeneralView} General view of the Knime interface

General Resources (B, HK)

Knime can be downloaded from knime website. Current version ( as of 2016-07-06) is 3.1.2. The software can be updated from the menu File>Update Knime. However major version update cannot be obtained through that mechanism and require a complete reinstall (at least from version 2 to version 3 of the software).

Knime website provides many resources to get started with the software as well as solving more advanced issues. The software also benefits from a dynamic community creating nodes collections for specific fields and answering questions on the forum.

Many examples of workflow are also available from Knime Examples server. From within Knime, right click on the Examples entry and select Login in the contextual menu. Many examples organized in a topical way can then be accessed.