Skip to content
DraconiticResearch
Search Sources API
182,460 studies31 sourcesUpdated 53 min ago
Peer-reviewedCC BY 4.0Frontiers in Neurology

Generative enhancement of suboptimally opacified cerebral CT angiography for large-vessel occlusion detection: a multireader study

  • Yifan Tu1
  • Mengfei Wang1,2
  • Zheng Sun1
  • Yao Zeng1,2
  • Yunyi Li1,2
  • Yixiao Tang1
  • Yulu Yang1
  • Zhongzheng Cao1
  • Shundong Hu1
  • Dan Wang1
  • Yuehua Li1,3
  1. 1Department of Radiology, Shanghai Sixth People's Hospital Affiliated to Shanghai Jiao Tong University School of Medicine
  2. 2Donghua University
  3. 3Faculty of Medical Imaging Technology, College of Health Science and Technology, Shanghai Jiao Tong University School of Medicine
Read full textPDFDOICite

Abstract

Background: Rapid identification of proximal anterior circulation large-vessel occlusion (LVO) on CT angiography (CTA) is essential for thrombectomy triage. Suboptimal arterial phase timing can obscure the intracranial arteries, prompting repeat CTA, additional contrast exposure, and treatment delay. We evaluated whether deep-learning-generated vascular images from a single suboptimally opacified acquisition improved LVO detection.

Methods: This retrospective study included 626 patients. RescueCTA, a paired 2.5D conditional generative model, was developed in 391 patients using last-frame CT perfusion (CTP) as the simulated acquisition-failure input and same-session CTA as the target. Evaluation included 194 primary cases with simulated inputs and 41 exploratory cases with genuinely underopacified CTA. RescueCTA was compared with three generative baselines. Two radiologists assessed image quality. Four readers compared source images alone with source images plus RescueCTA for target LVO detection in the primary cohort; two also evaluated the exploratory cohort. The primary endpoint was the multireader multicase (MRMC) area under the receiver operating characteristic curve (AUC); exploratory analyses were descriptive.

Results: In the primary cohort, RescueCTA image quality was higher than that of all three baselines (3.90 ± 0.81). Combined reading improved AUC from 0.850 to 0.933 (ΔAUC = 0.083 [95% confidence interval (CI), 0.025–0.141]; p = 0.007), while paired CTA reached 0.948 ( p = 0.594 vs. combined reading). Strict AUC improved from 0.756 to 0.880 ( p < 0.001) but remained lower than that of paired CTA (0.935; p = 0.019). In the exploratory cohort, combined reading yielded a higher AUC than underopacified CTA alone (0.981 vs. 0.889).

Conclusion: Deep-learning-generated vascular images improved LVO detection under simulated underopacification and showed preliminary feasibility in genuinely underopacified CTA. Larger real-world cohorts are needed to establish diagnostic performance after suboptimal arterial phase CTA.

Related

  1. Test-bolus peak attenuation for predicting aortic enhancement in coronary computed tomography angiography: an exploratory subgroup analysis
    Quantitative imaging in medicine and surgeryPeer-reviewedCC BY-NC-NDFull textMedical imaging & radiologyEurope PMC
  2. Combined non-contrast enhancement 3D bright-blood and 2D black-blood coronary magnetic resonance angiography for the assessment of coronary artery stenosis
    Qiu-Ju Hu, Di Tian, Dong-Sheng Jin +7
    Quantitative imaging in medicine and surgeryPeer-reviewedCC BY-NC-NDFull textMedical imaging & radiologyEurope PMC
  3. Fully automated diaphragm segmentation on routine 2D-cine and 3D thoracic MRI: multireader validation
    Samia Boussouar, Alexiane Pasquier, Damien Vaurs +16
    European radiology experimentalPeer-reviewedCC BYFull textMedical imaging & radiologyEurope PMC
  4. Diagnostic Accuracy and Renal Safety of Carbon Dioxide Angiography in Lower Limb Angiography: A Systematic Review and Meta-Analysis
    Diagnostics (Basel, Switzerland)Peer-reviewedCC BYFull textGeneral medicineEurope PMC
  5. Cardiac allograft vasculopathy demonstrated by coronary angiography and histopathologic correlates
    Eric Jonas, Nathan O. Stitziel, Sariah Hyacinth +1
    CirculationPeer-reviewedLicence not statedFull textSurgery & anaesthesiaPMC

Cite

BibTeXRIS
BibTeX
@article{tu2026generative,
  title = {{Generative enhancement of suboptimally opacified cerebral CT angiography for large-vessel occlusion detection: a multireader study}},
  author = {Yifan Tu and Mengfei Wang and Zheng Sun and Yao Zeng and Yunyi Li and Yixiao Tang and Yulu Yang and Zhongzheng Cao and Shundong Hu and Dan Wang and Yuehua Li},
  journal = {Frontiers in Neurology},
  year = {2026},
  month = oct,
  doi = {10.3389/fneur.2026.1955596},
  url = {https://www.frontiersin.org/articles/10.3389/fneur.2026.1955596/full},
  note = {CC BY 4.0},
}
RIS
TY  - JOUR
TI  - Generative enhancement of suboptimally opacified cerebral CT angiography for large-vessel occlusion detection: a multireader study
AU  - Yifan Tu
AU  - Mengfei Wang
AU  - Zheng Sun
AU  - Yao Zeng
AU  - Yunyi Li
AU  - Yixiao Tang
AU  - Yulu Yang
AU  - Zhongzheng Cao
AU  - Shundong Hu
AU  - Dan Wang
AU  - Yuehua Li
PY  - 2026
DA  - 2026/10/13
JO  - Frontiers in Neurology
DO  - 10.3389/fneur.2026.1955596
UR  - https://www.frontiersin.org/articles/10.3389/fneur.2026.1955596/full
AB  - Background: Rapid identification of proximal anterior circulation large-vessel occlusion (LVO) on CT angiography (CTA) is essential for thrombectomy triage. Suboptimal arterial phase timing can obscure the intracranial arteries, prompting repeat CTA, additional contrast exposure, and treatment delay. We evaluated whether deep-learning-generated vascular images from a single suboptimally opacified acquisition improved LVO detection.  Methods: This retrospective study included 626 patients. RescueCTA, a paired 2.5D conditional generative model, was developed in 391 patients using last-frame CT perfusion (CTP) as the simulated acquisition-failure input and same-session CTA as the target. Evaluation included 194 primary cases with simulated inputs and 41 exploratory cases with genuinely underopacified CTA. RescueCTA was compared with three generative baselines. Two radiologists assessed image quality. Four readers compared source images alone with source images plus RescueCTA for target LVO detection in the primary cohort; two also evaluated the exploratory cohort. The primary endpoint was the multireader multicase (MRMC) area under the receiver operating characteristic curve (AUC); exploratory analyses were descriptive.  Results: In the primary cohort, RescueCTA image quality was higher than that of all three baselines (3.90 ± 0.81). Combined reading improved AUC from 0.850 to 0.933 (ΔAUC = 0.083 [95% confidence interval (CI), 0.025–0.141]; p = 0.007), while paired CTA reached 0.948 ( p = 0.594 vs. combined reading). Strict AUC improved from 0.756 to 0.880 ( p < 0.001) but remained lower than that of paired CTA (0.935; p = 0.019). In the exploratory cohort, combined reading yielded a higher AUC than underopacified CTA alone (0.981 vs. 0.889).  Conclusion: Deep-learning-generated vascular images improved LVO detection under simulated underopacification and showed preliminary feasibility in genuinely underopacified CTA. Larger real-world cohorts are needed to establish diagnostic performance after suboptimal arterial phase CTA.
SN  - 1664-2295
ER  -