Stanford DRO Toolkit: Digital Reference Objects for Standardization of Radiomic Features.

Area of Science:

• Medical Imaging
• Radiomics
• Computational Anatomy

Background:

• Radiomics analysis relies on quantitative descriptors extracted from medical images.
• Standardization and replication of radiomics findings across different software implementations are crucial for clinical adoption.
• Existing software for image feature extraction lacks standardization, hindering reproducibility.

Purpose of the Study:

• To develop a software toolkit for creating standardized 3D digital reference objects.
• To enable reproducible radiomics analyses by providing customizable and mathematically defined objects.
• To facilitate the validation and comparison of radiomics software.

Main Methods:

• Development of a software toolkit to generate 3D digital reference objects.
• Mathematical definition of objects using continuous functions with customizable parameters (size, shape, intensity, texture, margin sharpness).
• Discretization of objects and saving them as DICOM files for compatibility.

Main Results:

• Successful creation of a software toolkit for generating 3D digital reference objects.
• Parameterized derivations of radiomics values for the created objects are presented.
• A user-downloadable sample collection of generated objects is provided for example use cases.

Conclusions:

• The developed software toolkit provides a method for standardizing radiomics analysis.
• The digital reference objects facilitate reproducible extraction of quantitative image descriptors.
• This approach supports the validation and inter-software comparison of radiomics findings.