Click here to

Session: Multi-Disciplinary General ePoster Viewing [Return to Session]

CERRx: The Extended CERR Platform Integrated with Open Source Image Informatics Ecosystem

A Apte*, A Iyer, E LoCastro, J Deasy, Memorial Sloan-Kettering Cancer Center, Maywood, NJ

Presentations

PO-GePV-M-43 (Sunday, 7/25/2021)   [Eastern Time (GMT-4)]

Purpose: CERR, a widely used computational platform for dosimetry and image analysis including a research friendly data structure, was developed in Matlab considering the ease of development and wide use by physicists and computer scientists. This work aims to make CERR independent of Matlab license and GNU Octave-compatible. Python has recently emerged as the preferred choice for building scientific computing tools due to its open source nature, user-friendly syntax and ability to serve as a “glue language”. Commonly used image informatics tools such as ITK, VTK, 3DSlicer provide Python APIs to allow simplified access to their C++ library. This work enables CERR functions to be accessible in Python, thus enhancing its utility in image informatics.

Methods: CERR was made compatible with GNU Octave to support batch computation as well as graphical user interface. Octave community-maintained packages for image and statistical analysis were used as replacements for Matlab’s built-in functions and toolboxes. Oct2py package in Python was used to call CERR functions from Python. This allows CERR to be used in JupyterLab with Octave-kernel and exchanging the results of CERR computations with a variety of programming languages.

Results: Octave-compatible CERR codebase including JupyterLab notebooks demonstrating deep learning image segmentation and feature computation are distributed as open source, GPL copyright software at https://www.github.com/cerr/CERR. This work has resulted in expanding CERR’s utility for medical image informatics including (1) accessing CERR data structure from XNAT via JupyterLab notebooks, (2) Using CERR functionality in 3DSlicer extensions, (3) Containerizing CERR’s pre and post processing for deep learning image segmentation pipeline in a license-free environment, (4) Using novel features for dosimetry, radiomics and outcomes prediction models in other open-source software.

Conclusion: This work makes CERR a truly open-source software platform that can be easily plugged into other components of the image informatics ecosystem.

Funding Support, Disclosures, and Conflict of Interest: Research was supported by the following grants: (1) P30 CA008748/CA/NCI NIH HHS/United States (2) R01 CA198121/CA/NCI NIH HHS/United States

ePosters

    Keywords

    Not Applicable / None Entered.

    Taxonomy

    Not Applicable / None Entered.

    Contact Email

    Share: