Enhanced detection of headache presentation in unruptured brain arteriovenous malformation through combined radiologic features: A cross-sectional study

Background: Although determining angioarchitecture provide qualitative insights into headache-susceptible brain arteriovenous malformation (BAVM), the potential of quantitative radiomics to detect headache in unruptured BAVM remains unclear. We developed classification models that integrate radiomic...

Full description

Saved in:
Bibliographic Details
Main Authors: Chia-Yu Liu, Chia-Feng Lu, Jr-Wei Wu, Yong-Sin Hu, Jih-Yuan Lin, Huai-Che Yang, Jing-Kai Loo, Feng-Chi Chang, Kang-Du Liu, Chung-Jung Lin
Format: Article
Language:English
Published: Elsevier 2025-06-01
Series:Neuroscience Informatics
Subjects:
Online Access:http://www.sciencedirect.com/science/article/pii/S2772528625000159
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850270263702192128
author Chia-Yu Liu
Chia-Feng Lu
Jr-Wei Wu
Yong-Sin Hu
Jih-Yuan Lin
Huai-Che Yang
Jing-Kai Loo
Feng-Chi Chang
Kang-Du Liu
Chung-Jung Lin
author_facet Chia-Yu Liu
Chia-Feng Lu
Jr-Wei Wu
Yong-Sin Hu
Jih-Yuan Lin
Huai-Che Yang
Jing-Kai Loo
Feng-Chi Chang
Kang-Du Liu
Chung-Jung Lin
author_sort Chia-Yu Liu
collection DOAJ
description Background: Although determining angioarchitecture provide qualitative insights into headache-susceptible brain arteriovenous malformation (BAVM), the potential of quantitative radiomics to detect headache in unruptured BAVM remains unclear. We developed classification models that integrate radiomic features and angioarchitecture to assist unruptured BAVM headache treatment decision-making. Methods: We considered patients with unruptured BAVM who underwent magnetic resonance imaging between 2010 and 2023. 146 radiomic features were assessed. Radiomic features were delineated, and angioarchitecture was analyzed. Statistical analyses, including least absolute shrinkage and selection operator regression and logistic regression, were used to select features and develop models. Receiver operating characteristic and decision curve analyses were performed to evaluate performance. Results: The clinical model based on age, sex, and parieto-occipital lesion location achieved an area under the curve (AUC) of 0.741. Adding two significant radiomic features and one angioarchitecture feature enhanced the models. The radiomic and angioarchitecture models achieved an AUC of 0.763. The combined model, with an AUC of 0.799, significantly outperformed the clinical model (P=0.046). Decision curve analysis indicated that the combined model performed best at threshold probabilities between 15% and 40%. Conclusion: Integrating radiomic features and angioarchitecture enhances the identification of unruptured BAVM headache.
format Article
id doaj-art-62e4e24dc9974b2fb9d4114b451ad8a3
institution OA Journals
issn 2772-5286
language English
publishDate 2025-06-01
publisher Elsevier
record_format Article
series Neuroscience Informatics
spelling doaj-art-62e4e24dc9974b2fb9d4114b451ad8a32025-08-20T01:52:42ZengElsevierNeuroscience Informatics2772-52862025-06-015210020010.1016/j.neuri.2025.100200Enhanced detection of headache presentation in unruptured brain arteriovenous malformation through combined radiologic features: A cross-sectional studyChia-Yu Liu0Chia-Feng Lu1Jr-Wei Wu2Yong-Sin Hu3Jih-Yuan Lin4Huai-Che Yang5Jing-Kai Loo6Feng-Chi Chang7Kang-Du Liu8Chung-Jung Lin9Department of Radiology, Taipei Veterans General Hospital, No. 201, Sec. 2, Shipai Rd., Beitou District, 11217, Taipei City, TaiwanDepartment of Biomedical Imaging and Radiological Sciences, National Yang Ming Chiao Tung University, No. 155, Sec. 2, Linong St., Beitou District, Taipei City, 112, TaiwanDepartment of Neurology, Neurological Institute, Taipei Veterans General Hospital, No. 201, Sec. 2, Shipai Rd., Beitou District, 11217, Taipei City, TaiwanDepartment of Radiology, Taipei Veterans General Hospital, No. 201, Sec. 2, Shipai Rd., Beitou District, 11217, Taipei City, Taiwan; Department of Radiology, Taipei Hospital, Ministry of Health and Welfare, No. 127, Su-Yuan Rd., Hsin-Chuang Dist., New Taipei City, 24213, TaiwanDepartment of Radiology, Taipei Veterans General Hospital, No. 201, Sec. 2, Shipai Rd., Beitou District, 11217, Taipei City, TaiwanDepartment of Neurosurgery, Neurological Institute, Taipei Veterans General Hospital, No. 201, Sec. 2, Shipai Rd., Beitou District, 11217, Taipei City, TaiwanDepartment of Radiology, Taipei Veterans General Hospital, No. 201, Sec. 2, Shipai Rd., Beitou District, 11217, Taipei City, TaiwanDepartment of Radiology, Taipei Veterans General Hospital, No. 201, Sec. 2, Shipai Rd., Beitou District, 11217, Taipei City, TaiwanDepartment of Neurology, Neurological Institute, Taipei Veterans General Hospital, No. 201, Sec. 2, Shipai Rd., Beitou District, 11217, Taipei City, TaiwanDepartment of Radiology, Taipei Veterans General Hospital, No. 201, Sec. 2, Shipai Rd., Beitou District, 11217, Taipei City, Taiwan; Corresponding author.Background: Although determining angioarchitecture provide qualitative insights into headache-susceptible brain arteriovenous malformation (BAVM), the potential of quantitative radiomics to detect headache in unruptured BAVM remains unclear. We developed classification models that integrate radiomic features and angioarchitecture to assist unruptured BAVM headache treatment decision-making. Methods: We considered patients with unruptured BAVM who underwent magnetic resonance imaging between 2010 and 2023. 146 radiomic features were assessed. Radiomic features were delineated, and angioarchitecture was analyzed. Statistical analyses, including least absolute shrinkage and selection operator regression and logistic regression, were used to select features and develop models. Receiver operating characteristic and decision curve analyses were performed to evaluate performance. Results: The clinical model based on age, sex, and parieto-occipital lesion location achieved an area under the curve (AUC) of 0.741. Adding two significant radiomic features and one angioarchitecture feature enhanced the models. The radiomic and angioarchitecture models achieved an AUC of 0.763. The combined model, with an AUC of 0.799, significantly outperformed the clinical model (P=0.046). Decision curve analysis indicated that the combined model performed best at threshold probabilities between 15% and 40%. Conclusion: Integrating radiomic features and angioarchitecture enhances the identification of unruptured BAVM headache.http://www.sciencedirect.com/science/article/pii/S2772528625000159AngioarchitectureBrain arteriovenous malformationDecision curve analysisHeadacheRadiomics
spellingShingle Chia-Yu Liu
Chia-Feng Lu
Jr-Wei Wu
Yong-Sin Hu
Jih-Yuan Lin
Huai-Che Yang
Jing-Kai Loo
Feng-Chi Chang
Kang-Du Liu
Chung-Jung Lin
Enhanced detection of headache presentation in unruptured brain arteriovenous malformation through combined radiologic features: A cross-sectional study
Neuroscience Informatics
Angioarchitecture
Brain arteriovenous malformation
Decision curve analysis
Headache
Radiomics
title Enhanced detection of headache presentation in unruptured brain arteriovenous malformation through combined radiologic features: A cross-sectional study
title_full Enhanced detection of headache presentation in unruptured brain arteriovenous malformation through combined radiologic features: A cross-sectional study
title_fullStr Enhanced detection of headache presentation in unruptured brain arteriovenous malformation through combined radiologic features: A cross-sectional study
title_full_unstemmed Enhanced detection of headache presentation in unruptured brain arteriovenous malformation through combined radiologic features: A cross-sectional study
title_short Enhanced detection of headache presentation in unruptured brain arteriovenous malformation through combined radiologic features: A cross-sectional study
title_sort enhanced detection of headache presentation in unruptured brain arteriovenous malformation through combined radiologic features a cross sectional study
topic Angioarchitecture
Brain arteriovenous malformation
Decision curve analysis
Headache
Radiomics
url http://www.sciencedirect.com/science/article/pii/S2772528625000159
work_keys_str_mv AT chiayuliu enhanceddetectionofheadachepresentationinunrupturedbrainarteriovenousmalformationthroughcombinedradiologicfeaturesacrosssectionalstudy
AT chiafenglu enhanceddetectionofheadachepresentationinunrupturedbrainarteriovenousmalformationthroughcombinedradiologicfeaturesacrosssectionalstudy
AT jrweiwu enhanceddetectionofheadachepresentationinunrupturedbrainarteriovenousmalformationthroughcombinedradiologicfeaturesacrosssectionalstudy
AT yongsinhu enhanceddetectionofheadachepresentationinunrupturedbrainarteriovenousmalformationthroughcombinedradiologicfeaturesacrosssectionalstudy
AT jihyuanlin enhanceddetectionofheadachepresentationinunrupturedbrainarteriovenousmalformationthroughcombinedradiologicfeaturesacrosssectionalstudy
AT huaicheyang enhanceddetectionofheadachepresentationinunrupturedbrainarteriovenousmalformationthroughcombinedradiologicfeaturesacrosssectionalstudy
AT jingkailoo enhanceddetectionofheadachepresentationinunrupturedbrainarteriovenousmalformationthroughcombinedradiologicfeaturesacrosssectionalstudy
AT fengchichang enhanceddetectionofheadachepresentationinunrupturedbrainarteriovenousmalformationthroughcombinedradiologicfeaturesacrosssectionalstudy
AT kangduliu enhanceddetectionofheadachepresentationinunrupturedbrainarteriovenousmalformationthroughcombinedradiologicfeaturesacrosssectionalstudy
AT chungjunglin enhanceddetectionofheadachepresentationinunrupturedbrainarteriovenousmalformationthroughcombinedradiologicfeaturesacrosssectionalstudy