Graph feature selection for enhancing radiomic stability and reproducibility across multiple institutions in head and neck cancer

Abstract Radiomic biomarkers offer promise for precision oncology. However, their clinical utility is limited by variability from differing imaging protocols and the high dimensionality of radiomics data. Feature selection is key for better interpretability, accuracy, and efficiency, yet traditional...

Full description

Saved in:
Bibliographic Details
Main Authors: Hajar Moradmand, Jason Molitoris, Xiao Ling, Lisa Schumaker, Erin Allor, Hannah Thomas, Danielle Arons, Matthew Ferris, Rebecca Krc, William Silva Mendes, Phuoc Tran, Amit Sawant, Ranee Mehra, Daria A. Gaykalova, Lei Ren
Format: Article
Language:English
Published: Nature Portfolio 2025-07-01
Series:Scientific Reports
Online Access:https://doi.org/10.1038/s41598-025-12161-w
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849332674804580352
author Hajar Moradmand
Jason Molitoris
Xiao Ling
Lisa Schumaker
Erin Allor
Hannah Thomas
Danielle Arons
Matthew Ferris
Rebecca Krc
William Silva Mendes
Phuoc Tran
Amit Sawant
Ranee Mehra
Daria A. Gaykalova
Lei Ren
author_facet Hajar Moradmand
Jason Molitoris
Xiao Ling
Lisa Schumaker
Erin Allor
Hannah Thomas
Danielle Arons
Matthew Ferris
Rebecca Krc
William Silva Mendes
Phuoc Tran
Amit Sawant
Ranee Mehra
Daria A. Gaykalova
Lei Ren
author_sort Hajar Moradmand
collection DOAJ
description Abstract Radiomic biomarkers offer promise for precision oncology. However, their clinical utility is limited by variability from differing imaging protocols and the high dimensionality of radiomics data. Feature selection is key for better interpretability, accuracy, and efficiency, yet traditional methods lack stability and reproducibility. We investigate a Graph-Based Feature Selection (Graph-FS) approach that models feature interdependencies to identify stable radiomic signatures for head and neck squamous cell carcinoma (HNSCC) across institutions. We retrospectively analyzed 1,648 radiomic features extracted from the gross tumor volumes of 752 HNSCC patients from three institutions. After standard preprocessing and applying 36 radiomics parameter configurations to simulate variability, we compared Graph-FS with established methods: Boruta, Lasso, Recursive Feature Elimination (RFE), and Minimum Redundancy Maximum Relevance (mRMR). We evaluated feature selection stability and reproducibility using Pearson correlation, the Jaccard Index (JI), and the Dice-Sorensen Index (DSI) and assessed ranking consistency with Kendall’s Coefficient of Concordance (W). Graph-FS achieved higher stability (JI = 0.46, DSI = 0.62, OP = 45.8%) versus baseline methods with JI of 0.005 (Boruta), 0.010 (Lasso), 0.006 (RFE) and 0.014 (mRMR). These results demonstrate that Graph-FS enhances feature stability, reproducibility, and predictive performance. This method could facilitate integration into AI-driven radiomics workflows for reliable, multi-center biomarker discovery.
format Article
id doaj-art-2803fb99229f49758300bbce4b437c2a
institution Kabale University
issn 2045-2322
language English
publishDate 2025-07-01
publisher Nature Portfolio
record_format Article
series Scientific Reports
spelling doaj-art-2803fb99229f49758300bbce4b437c2a2025-08-20T03:46:08ZengNature PortfolioScientific Reports2045-23222025-07-0115111610.1038/s41598-025-12161-wGraph feature selection for enhancing radiomic stability and reproducibility across multiple institutions in head and neck cancerHajar Moradmand0Jason Molitoris1Xiao Ling2Lisa Schumaker3Erin Allor4Hannah Thomas5Danielle Arons6Matthew Ferris7Rebecca Krc8William Silva Mendes9Phuoc Tran10Amit Sawant11Ranee Mehra12Daria A. Gaykalova13Lei Ren14Department of Radiation Oncology, University of Maryland School of MedicineDepartment of Radiation Oncology, University of Maryland School of MedicineDepartment of Mathematics, Auburn University at MontgomeryMarlene and Stewart Greenebaum Comprehensive Cancer Center, University of Maryland School of MedicineInstitute for Genome Sciences, University of Maryland School of MedicineInstitute for Genome Sciences, University of Maryland School of MedicineInstitute for Genome Sciences, University of Maryland School of MedicineDepartment of Radiation Oncology, University of Maryland School of MedicineDepartment of Radiation Oncology, University of Maryland School of MedicineDepartment of Radiation Oncology, University of Maryland School of MedicineDepartment of Radiation Oncology, University of Maryland School of MedicineDepartment of Radiation Oncology, University of Maryland School of MedicineDepartment of Mathematics, Auburn University at MontgomeryMarlene and Stewart Greenebaum Comprehensive Cancer Center, University of Maryland School of MedicineDepartment of Radiation Oncology, University of Maryland School of MedicineAbstract Radiomic biomarkers offer promise for precision oncology. However, their clinical utility is limited by variability from differing imaging protocols and the high dimensionality of radiomics data. Feature selection is key for better interpretability, accuracy, and efficiency, yet traditional methods lack stability and reproducibility. We investigate a Graph-Based Feature Selection (Graph-FS) approach that models feature interdependencies to identify stable radiomic signatures for head and neck squamous cell carcinoma (HNSCC) across institutions. We retrospectively analyzed 1,648 radiomic features extracted from the gross tumor volumes of 752 HNSCC patients from three institutions. After standard preprocessing and applying 36 radiomics parameter configurations to simulate variability, we compared Graph-FS with established methods: Boruta, Lasso, Recursive Feature Elimination (RFE), and Minimum Redundancy Maximum Relevance (mRMR). We evaluated feature selection stability and reproducibility using Pearson correlation, the Jaccard Index (JI), and the Dice-Sorensen Index (DSI) and assessed ranking consistency with Kendall’s Coefficient of Concordance (W). Graph-FS achieved higher stability (JI = 0.46, DSI = 0.62, OP = 45.8%) versus baseline methods with JI of 0.005 (Boruta), 0.010 (Lasso), 0.006 (RFE) and 0.014 (mRMR). These results demonstrate that Graph-FS enhances feature stability, reproducibility, and predictive performance. This method could facilitate integration into AI-driven radiomics workflows for reliable, multi-center biomarker discovery.https://doi.org/10.1038/s41598-025-12161-w
spellingShingle Hajar Moradmand
Jason Molitoris
Xiao Ling
Lisa Schumaker
Erin Allor
Hannah Thomas
Danielle Arons
Matthew Ferris
Rebecca Krc
William Silva Mendes
Phuoc Tran
Amit Sawant
Ranee Mehra
Daria A. Gaykalova
Lei Ren
Graph feature selection for enhancing radiomic stability and reproducibility across multiple institutions in head and neck cancer
Scientific Reports
title Graph feature selection for enhancing radiomic stability and reproducibility across multiple institutions in head and neck cancer
title_full Graph feature selection for enhancing radiomic stability and reproducibility across multiple institutions in head and neck cancer
title_fullStr Graph feature selection for enhancing radiomic stability and reproducibility across multiple institutions in head and neck cancer
title_full_unstemmed Graph feature selection for enhancing radiomic stability and reproducibility across multiple institutions in head and neck cancer
title_short Graph feature selection for enhancing radiomic stability and reproducibility across multiple institutions in head and neck cancer
title_sort graph feature selection for enhancing radiomic stability and reproducibility across multiple institutions in head and neck cancer
url https://doi.org/10.1038/s41598-025-12161-w
work_keys_str_mv AT hajarmoradmand graphfeatureselectionforenhancingradiomicstabilityandreproducibilityacrossmultipleinstitutionsinheadandneckcancer
AT jasonmolitoris graphfeatureselectionforenhancingradiomicstabilityandreproducibilityacrossmultipleinstitutionsinheadandneckcancer
AT xiaoling graphfeatureselectionforenhancingradiomicstabilityandreproducibilityacrossmultipleinstitutionsinheadandneckcancer
AT lisaschumaker graphfeatureselectionforenhancingradiomicstabilityandreproducibilityacrossmultipleinstitutionsinheadandneckcancer
AT erinallor graphfeatureselectionforenhancingradiomicstabilityandreproducibilityacrossmultipleinstitutionsinheadandneckcancer
AT hannahthomas graphfeatureselectionforenhancingradiomicstabilityandreproducibilityacrossmultipleinstitutionsinheadandneckcancer
AT daniellearons graphfeatureselectionforenhancingradiomicstabilityandreproducibilityacrossmultipleinstitutionsinheadandneckcancer
AT matthewferris graphfeatureselectionforenhancingradiomicstabilityandreproducibilityacrossmultipleinstitutionsinheadandneckcancer
AT rebeccakrc graphfeatureselectionforenhancingradiomicstabilityandreproducibilityacrossmultipleinstitutionsinheadandneckcancer
AT williamsilvamendes graphfeatureselectionforenhancingradiomicstabilityandreproducibilityacrossmultipleinstitutionsinheadandneckcancer
AT phuoctran graphfeatureselectionforenhancingradiomicstabilityandreproducibilityacrossmultipleinstitutionsinheadandneckcancer
AT amitsawant graphfeatureselectionforenhancingradiomicstabilityandreproducibilityacrossmultipleinstitutionsinheadandneckcancer
AT raneemehra graphfeatureselectionforenhancingradiomicstabilityandreproducibilityacrossmultipleinstitutionsinheadandneckcancer
AT dariaagaykalova graphfeatureselectionforenhancingradiomicstabilityandreproducibilityacrossmultipleinstitutionsinheadandneckcancer
AT leiren graphfeatureselectionforenhancingradiomicstabilityandreproducibilityacrossmultipleinstitutionsinheadandneckcancer