DSA radiomics and clinical features for hemorrhagic risk assessment in bAVM embolization

Abstract

Background: Hemorrhagic complications (HCs) remain a major cause of morbidity in endovascular treatment (EVT) for brain arteriovenous malformation (bAVM), while reliable EVT-specific risk assessment remains challenging. This study aimed to develop and internally evaluate an exploratory model integrating digital subtraction angiography (DSA) radiomics and clinical features to assess the risk of EVT-related HCs.

Methods: This retrospective study included 620 consecutive bAVM patients who underwent first-time EVT between 2011 and 2024. Radiomics features were extracted from anteroposterior and lateral DSA images of the nidus. Clinical-only, radiomics-only, and combined models were developed using four algorithms with stratified five-fold cross-validation on the training set and evaluated on the test set. Model interpretability was performed using SHapley Additive exPlanations (SHAP).

Results: HCs occurred in 100 patients (16.1%). On the test set, the XGBoost-based fusion model achieved the best overall performance (AUC 0.81), showing a modest improvement over the radiomics-only XGBoost model while outperforming all clinical-only models. Radiomics features contributed substantially to model performance, with clinical variables providing complementary information.

Conclusions: A model integrating DSA radiomics and clinical features provides an exploratory approach to assessing the risk of EVT-related HCs in patients with bAVM. These findings suggest the potential value of DSA-derived radiomics for individualized peri-procedural risk assessment, although external validation is required before clinical application.

Supplementary Information: The online version contains supplementary material available at https://doi.org/10.1186/s12880-026-02587-7.

Fusion of Deep Learning Radiomics and Clinical Features for Individualized Prediction of Neoadjuvant Chemotherapy Response in Osteosarcoma
Jianhua Mu, Yi Luo, Han Liu +7
Diagnostics (Basel, Switzerland)1 Oct 2026Peer-reviewedCC BYFull textGeneral medicineEurope PMC
Pre-stenting Bronchial Artery Embolization for Bronchial Hemorrhagic Tumor
Keisuke Shiraha, Ken Sato, Yuki Takigawa +1
Internal Medicine1 Aug 2026Peer-reviewedCC BY-NC-NDFull textHuman geneticsJ-STAGE
Habitat radiomics combined with clinical features for predicting tertiary lymphoid structure status in pancreatic ductal adenocarcinoma: development and validation of a preoperative CECT-based model.
Ke Li, Rui Guo, Xianqi Wang +6
Gland surgery28 Sep 2026Peer-reviewedCC BY-NC-NDFull textMedical imaging & radiologyEurope PMC
Balloon-occluded ethylene vinyl alcohol copolymer embolization of a hemorrhagic renal mass in an octogenarian initially unfit for nephrectomy.
Anthony D Mohabir, Leili Mirsadraei, Anthony Corcoran +1
Radiology case reports20 Oct 2026Peer-reviewedCC BY-NC-NDFull textGeneral medicineEurope PMC
Multiparametric MRI radiomics and clinical features for preoperative prediction of high-grade meningioma: a two-centre retrospective study
Lulu Gai, Yudong Yan, Wenzheng Lyu +3
Frontiers in Oncology2 Nov 2026Peer-reviewedCC BY 4.0OncologyFrontiersCrossref

Cite

BibTeX RIS

BibTeX

@article{zeng2026radiomics,
  title = {{DSA radiomics and clinical features for hemorrhagic risk assessment in bAVM embolization}},
  author = {Ni Zeng and Caihong Li and Jianbin Zhu and Qiuxian Wang and Zhibo Wen},
  journal = {BMC Medical Imaging},
  year = {2026},
  month = jul,
  doi = {10.1186/s12880-026-02587-7},
  url = {https://doi.org/10.1186/s12880-026-02587-7},
  note = {CC BY-NC-ND 4.0},
}

RIS

TY  - JOUR
TI  - DSA radiomics and clinical features for hemorrhagic risk assessment in bAVM embolization
AU  - Ni Zeng
AU  - Caihong Li
AU  - Jianbin Zhu
AU  - Qiuxian Wang
AU  - Zhibo Wen
PY  - 2026
DA  - 2026/07/29
JO  - BMC Medical Imaging
DO  - 10.1186/s12880-026-02587-7
UR  - https://doi.org/10.1186/s12880-026-02587-7
AB  - Background: Hemorrhagic complications (HCs) remain a major cause of morbidity in endovascular treatment (EVT) for brain arteriovenous malformation (bAVM), while reliable EVT-specific risk assessment remains challenging. This study aimed to develop and internally evaluate an exploratory model integrating digital subtraction angiography (DSA) radiomics and clinical features to assess the risk of EVT-related HCs.  Methods: This retrospective study included 620 consecutive bAVM patients who underwent first-time EVT between 2011 and 2024. Radiomics features were extracted from anteroposterior and lateral DSA images of the nidus. Clinical-only, radiomics-only, and combined models were developed using four algorithms with stratified five-fold cross-validation on the training set and evaluated on the test set. Model interpretability was performed using SHapley Additive exPlanations (SHAP).  Results: HCs occurred in 100 patients (16.1%). On the test set, the XGBoost-based fusion model achieved the best overall performance (AUC 0.81), showing a modest improvement over the radiomics-only XGBoost model while outperforming all clinical-only models. Radiomics features contributed substantially to model performance, with clinical variables providing complementary information.  Conclusions: A model integrating DSA radiomics and clinical features provides an exploratory approach to assessing the risk of EVT-related HCs in patients with bAVM. These findings suggest the potential value of DSA-derived radiomics for individualized peri-procedural risk assessment, although external validation is required before clinical application.  Supplementary Information: The online version contains supplementary material available at https://doi.org/10.1186/s12880-026-02587-7.
SN  - 1471-2342
ER  -

Abstract

Related

Cite