Deep learning-assisted identification and quantification of aneurysmal subarachnoid hemorrhage in non-contrast CT scans: Development and external validation of Hybrid 2D/3D UNet

Hu, Ping; Zhou, Haizhu; Yan, Tengfeng; Miu, Hongping; Xiao, Feng; Zhu, Xinyi; Shu, Lei; Yang, Shuang; Jin, Ruiyun; Dou, Wenlei; Ren, Baoyu; Zhu, Lizhen; Liu, Wanrong; Zhang, Yihan; Zeng, Kaisheng; Ye, Minhua; Lv, Shigang; Wu, Miaojing; Deng, Gang; Hu, Rong; Zhan, Renya; Chen, Qianxue; Zhang, Dong; Zhu, Xingen

doi:10.1016/j.neuroimage.2023.120321

Cited by 6 publications

(6 citation statements)

References 39 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Quality assessment for individual studies can be found in Supplementary Table 2 (Supplemental Digital Content 4, http://links.lww.com/JS9/C84 ). In summary, nine studies 28 , 32 , 33 , 39 , 40 , 46 , 52 , 56 , 61 had low risk of bias and applicability in all domains; while, the rest studies had risk of bias in at least one domain.…”

Section: Resultsmentioning

confidence: 85%

“…USA ( n =14) and China ( n =12) were two most countries with the largest number of publications. Nine included studies 27 , 28 , 31 , 35 , 40 , 41 , 55 , 60 run internal validation via cross-validation method, 14 included studies 29 , 30 , 32 – 34 , 37 – 39 , 46 , 47 , 56 , 57 , 59 , 61 via random split method, 3 included studies 42 , 48 , 50 via hold-out method, and 10 included studies 15 , 36 , 43 , 45 , 49 , 51 – 54 , 58 not report or not application. Twelve included studies 30 , 32 , 33 , 37 , 39 , 41 , 46 , 47 , 50 , 51 , 55 , 56 , 61 performed a further external validation.…”

Section: Resultsmentioning

confidence: 99%

“…Nine included studies 27 , 28 , 31 , 35 , 40 , 41 , 55 , 60 run internal validation via cross-validation method, 14 included studies 29 , 30 , 32 – 34 , 37 – 39 , 46 , 47 , 56 , 57 , 59 , 61 via random split method, 3 included studies 42 , 48 , 50 via hold-out method, and 10 included studies 15 , 36 , 43 , 45 , 49 , 51 – 54 , 58 not report or not application. Twelve included studies 30 , 32 , 33 , 37 , 39 , 41 , 46 , 47 , 50 , 51 , 55 , 56 , 61 performed a further external validation. Regarding the target tasks for processing the NCCT imaging, 30 studies 15 , 28 , 30 – 40 , 43 – 47 , 50 – 61 focused on detection, 27 studies 15 , 27 , 29 , 31 , 33 – 35 , 38 – 45 , 48 – 53 , 56 – 61 focused on segmentation, and 18 studies…”

Section: Resultsmentioning

confidence: 99%

“…Sensitivity, specificity, AUROC, PPV, NPV, and precision were calculated to assess DL models detection performance. twenty-five studies 28 , 30 – 34 , 36 – 38 , 43 , 45 – 48 , 51 – 61 recorded sensitivity (Supplementary Figure 1, Supplemental Digital Content 4, http://links.lww.com/JS9/C84 , pooled sensitivity 0.89, 95% CI: 0.88–0.90), 18 studies 28 , 30 , 32 – 34 , 36 – 38 , 43 , 46 – 48 , 51 – 56 recorded specificity (Supplementary Figure 2, Supplemental Digital Content 4, http://links.lww.com/JS9/C84 , pooled specificity 0.91, 95% CI: 0.89–0.93), six studies 28 , 30 , 32 , 37 , 46 , 52 recorded AUROC (Supplementary Figure 3, Supplemental Digital Content 4, http://links.lww.com/JS9/C84 , pooled AUROC 0.94, 95% CI: 0.93–0.95), 10 studies 31 , 33 , 36 – 38 , 51 , 54 , 57 – 59 reported PPV (Supplementary Figure 4, Supplemental Digital Content 4, http://links.lww.com/JS9/C84 , pooled PPV 0.92, 95% CI: 0.91–0.93), six studies 33 , 36 – 38 , 51 , 54 reported NPV (Supplementary Figure 5, Supplemental Digital Content 4, http://links.lww.com/JS9/C84 , pooled NPV 0.94, 95% CI: 0.91–0.96), nine studies 28 , 29 , 31 , 35 , 39 , 53 , 55 , 56 , 61 ...…”

Section: Resultsmentioning

confidence: 99%

See 3 more Smart Citations

Deep learning-assisted detection and segmentation of intracranial hemorrhage in noncontrast computed tomography scans of acute stroke patients: a systematic review and meta-analysis

Hu,

Yan,

Xiao

et al. 2024

International Journal of Surgery

View full text Add to dashboard Cite

Background: Deep learning (DL)-assisted detection and segmentation of intracranial hemorrhage stroke in noncontrast computed tomography (NCCT) scans are well-established, but evidence on this topic is lacking. Materials and methods: PubMed and Embase databases were searched from their inception to November 2023 to identify related studies. The primary outcomes included sensitivity, specificity, and the Dice Similarity Coefficient (DSC); while the secondary outcomes were positive predictive value (PPV), negative predictive value (NPV), precision, area under the receiver operating characteristic curve (AUROC), processing time, and volume of bleeding. Random-effect model and bivariate model were used to pooled independent effect size and diagnostic meta-analysis data, respectively. Results: A total of 36 original studies were included in this meta-analysis. Pooled results indicated that DL technologies have a comparable performance in intracranial hemorrhage detection and segmentation with high values of sensitivity (0.89, 95% CI: 0.88–0.90), specificity (0.91, 95% CI: 0.89–0.93), AUROC (0.94, 95% CI: 0.93–0.95), PPV (0.92, 95% CI: 0.91–0.93), NPV (0.94, 95% CI: 0.91–0.96), precision (0.83, 95% CI: 0.77–0.90), DSC (0.84, 95% CI: 0.82–0.87). There is no significant difference between manual labeling and DL technologies in hemorrhage quantification (MD 0.08, 95% CI: −5.45–5.60, P=0.98), but the latter takes less process time than manual labeling (WMD 2.26, 95% CI: 1.96–2.56, P=0.001). Conclusion: This systematic review has identified a range of DL algorithms that the performance was comparable to experienced clinicians in hemorrhage lesions identification, segmentation, and quantification but with greater efficiency and reduced cost. It is highly emphasized that multicenter randomized controlled clinical trials will be needed to validate the performance of these tools in the future, paving the way for fast and efficient decision-making during clinical procedure in patients with acute hemorrhagic stroke.

show abstract

Section: Resultsmentioning

confidence: 85%

Section: Resultsmentioning

confidence: 99%

Section: Resultsmentioning

confidence: 99%

Section: Resultsmentioning

confidence: 99%

See 2 more Smart Citations

Deep learning-assisted detection and segmentation of intracranial hemorrhage in noncontrast computed tomography scans of acute stroke patients: a systematic review and meta-analysis

Hu,

Yan,

Xiao

et al. 2024

International Journal of Surgery

View full text Add to dashboard Cite

show abstract

“…Особливо це стосується мікрохірургічних ме тодів. Д ля прогнозування наслідків аневризматичного субарахноїдального крововиливу (САК) традиційно використовують клініко-неврологічний стан, тип, поширення та об'єм вну трішньочерепного крововиливу, виявленого під час нейровізуалізаційного обстеження [5][6][7][8]. Ус т анов ле но, що вп лив т ак их пр е дик т орів є обов'язковою ланкою патогенезу аневризматичних САК, що визначає тактику ведення хворих [9][10][11].…”

Section: вступunclassified

Clinical and neuroimaging predictors of the outcome of microsurgical intervention against the background of cerebral aneurysms rupture

Kazantseva,

Zorin

2024

Ukr Neurosurg J

View full text Add to dashboard Cite

Objective: minimization of the risks of complications of microsurgical intervention (MI) for cerebral aneurysm (CA) rupture due to established clinical and neuroimaging predictors. Material and methods. Evaluation of microsurgical treatment of 418 patients in the acute period of cerebral aneurysms (CA) rupture in the 2013-2018 period, that were operated at the Vascular Neurosurgery Center of the Dnipropetrovsk Regional Clinical Hospital named after I.I. Mechnikov, taking into account factors that may have an influence on favorable or unfavorable functional consequences. Age, gender, as well as clinical-neurological and neuroimaging assessment of the condition of patients according to the severity of aneurysmal subarachnoid hemorrhage (SAH) according to the Hunt-Hess classification, WFNS, level of consciousness impairment according to the Glasgow Coma Scale (GCS) were studied. The severity of aneurysmal subarachnoid hemorrhage was evaluated according to the Fisher prognostic scale according to the performed spiral computed tomography (SCT). Cerebral angiospasm (CAS) was evaluated according to cerebral angiography (CAG), and its severity according to transcranial dopplerography (TCD) on the day of hospitalization and surgery. The performed microsurgical intervention (MI) was evaluated according to the Glasgow Outcome Scale (GOS). The study was both retrospective and prospective. Results. Among 77 (18.4%) patients with adverse consequences of MI according to 1-3 points of GOS, a dependence was reliably found on the level of consciousness disturbance according to GCS, the severity of meningeal and focal symptoms, the severity of parenchymal hemorrhage, CAS manifestations on the day of surgery according to TCD, the duration of surgery after the rupture of CA (p<0.001). There was no statistically significant relationship between age, sex, localization, form and side of CA rupture. Also, a strong inverse correlation was found between GCS and classification of SAH according to WFNS rs=‒0.96 (95% CI 0.96-0.97) and Hunt-Hess rs=‒0.81 (95% CI 0.77-0.84) during hospitalization. Conclusions. The effectiveness of MI in the case of CA rupture depends on the severity of prognostic criteria of clinical and neurological examination - evaluation by the GCS, neurological disorders, severity of parenchymal hemorrhage, manifestations of CAS on the day of surgery for TCD, the duration of surgery after CA rupture. Revealed relationship between the score according to the GOS, classifications of SAH according to the WFNS and Hunt-Hess significantly simplifies diagnostic measures during the examination of patients in the acute period of CA rupture.

show abstract

Composite acoustic hole segmentation by semi-supervised learning for robotic multi-spindle drilling of aero-engine nacelle acoustic liners

Dong,

Mei,

et al. 2024

Composites Part A: Applied Science and Manufacturing

View full text Add to dashboard Cite

Deep learning-assisted identification and quantification of aneurysmal subarachnoid hemorrhage in non-contrast CT scans: Development and external validation of Hybrid 2D/3D UNet

Cited by 6 publications

References 39 publications

Deep learning-assisted detection and segmentation of intracranial hemorrhage in noncontrast computed tomography scans of acute stroke patients: a systematic review and meta-analysis

Deep learning-assisted detection and segmentation of intracranial hemorrhage in noncontrast computed tomography scans of acute stroke patients: a systematic review and meta-analysis

Clinical and neuroimaging predictors of the outcome of microsurgical intervention against the background of cerebral aneurysms rupture

Composite acoustic hole segmentation by semi-supervised learning for robotic multi-spindle drilling of aero-engine nacelle acoustic liners

Contact Info

Product

Resources

About