Click here to

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

STRATIS: A Cloud-Enabled Software Toolbox for RAdioTherapy and Imaging Analysis

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

Presentations

PO-GePV-M-61 (Sunday, 7/10/2022)   [Eastern Time (GMT-4)]

ePoster Forums

Purpose: STRATIS-forge organization on GitHub aims to provide repositories of codebases and cloud-enabled workflows for radiation therapy and image analysis.

Methods: STRATIS-forge provides a collection of repositories and workflows in the form of reusable Jupyter notebooks with Python as the base kernel. This allows the use of Python packages along with software written in commonly used programming languages such as R and MATLAB/GNU-Octave. Bridge libraries such as rpy2 for R and oct2py for GNU-Octave are used for connecting with software written in respective languages. STRATIS leverages medical image informatics tools such as 3DSlicer, CERR, Plastimatch, ITK, VTK for application specific computations along with analysis libraries in Python and R. STRATIS provides tools to transfer data between various software platforms. For example, sliCERR module allows data to flow between CERR and 3DSlicer, wrappers for image registration libraries and data transformations to use step-wise image registration. STRATIS workflows can be run on a cloud platform such as Imaging Data Commons as well as a local Jupyter notebook environment. Public datasets can be accessed, for example, from TCIA via Google’s BigQuery. Imaging Datasets can also be pulled from XNAT using REST API from the pyxnat package.

Results: The repositories containing codebases and notebooks are open source, GPL copyright software distributed at the STRATIS-forge GitHub organization https://www.github.com/stratis-forge. The workflows include graph-based clustering of imaging and dosimetric features, deformable image registration for mapping radiation therapy dose distributions to a reference geometry, image segmentation using deep learning models, normal tissue complication and tumor control models for radiation therapy, novel semi-quantitative features from dynamic MRI scans, IBSI-compatible radiomics and texture features.

Conclusion: STRATIS-forge allows researchers to deploy and reuse workflows for radiotherapy and image analysis. This leads to reproducible analysis while leveraging the computational resources from popular cloud platforms such as Google, AWS, FireCloud.

Keywords

Not Applicable / None Entered.

Taxonomy

Not Applicable / None Entered.

Contact Email

Share: