MENU

Click here to

Ă—

Are you sure ?

Yes, do it No, cancel

ROdiomX: A New Validated Software for the Radiomics Analysis of Medical Images in Radiation Oncology

H Bagher-Ebadian1*, M Lu1, F Siddiqui1, A Ghanem1, N Wen1, Q Wu1, B Movsas1, I Chetty1, (1) Henry Ford Health System, Detroit, MI

Presentations

(Sunday, 7/12/2020)   [Eastern Time (GMT-4)]

Room: AAPM ePoster Library

Purpose: Here we introduce an in-house-designed software platform (ROdiomX) for the radiomics analysis of medical images in radiation oncology. ROdiomX is a MATLAB–based framework for the computation of a large number of radiomic features and feature aggregation techniques in compliance with the Image-Biomarker-Standardization-Initiative (IBSI) guidelines, which includes post-processing protocols and quantitative benchmark results for analysis of computational phantom images.


Methods: ROdiomiX is a com computation cores were implemented on the basis of the guidelines proposed by the IBSI. The ROdimoX software is capable of computing the following 10 different feature categories: Local-Intensity features, Intensity-Histogram features, Intensity-Based-Statistical features, Intensity-Volume-Histogram features, Gray-Level-Co-occurrence based features, Gray-Level-Run-Length based features, Gray-Level-Size-Zone based features, Gray-Level-Distance-Zone based features, Neighborhood-Grey-Tone-Difference based features, and Neighboring-Grey-Level-Dependence based features. ROdiomX software was validated using the following two different 3D digital phantoms: The IBSI and the HFHS gradient digital phantoms. Intraclass correlations (ICC with A-k method) for the degree of absolute agreement for the measurements that are averages of k independent measurements on randomly selected objects with confidence level of 0.95 were calculated and compared for each feature category.


Results: Among the 10 feature categories with 143 total features for 10 different feature aggregation methods (in compliance with IBSI guidelines), the agreement between quantitative radiomics values computed with the ROdiomiX software versus the IBSI benchmark data was: ICC>0.9 with CL=0.95 (Ranked: Excellent). Agreement between ROdiomiX and the HFHS-designed computational platform benchmark data was: ICC>0.9 with CL=0.95 (Ranked: Excellent).


Conclusion: The authors successfully developed a platform for computation of quantitative radiomics features. The image preprocessing and computational software cores were designed following the procedures specific by the IBSI. Benchmarking testing was in excellent agreement against the IBSI and HFHS-designed computational phantoms.

Funding Support, Disclosures, and Conflict of Interest: This work was supported in part by a grant from Varian Medical Systems (Palo Alto, CA).

Keywords

Texture Analysis, Numerical Analysis, Feature Extraction

Taxonomy

IM/TH- Informatics: Informatics in Imaging (general)

Contact Email