Diagnostic performance of an algorithm for automated large vessel occlusion detection on CT angiography

Luijten, Sven P.R.; Wolff, Lennard; Duvekot, Martijne H C; Doormaal, Pieter‐Jan van; Moudrous, Walid; Kerkhoff, Henk; Nijeholt, Geert J. Lycklama à; Bokkers, Reinoud P H; Yo, Lonneke S.F.; Hofmeijer, Jeannette; Zwam, Wim H. van; Es, Adriaan C.G.M. van; Dippel, Diederik W.J.; Roozenbeek, Bob; Lugt, Aad van der

doi:10.1136/neurintsurg-2021-017842

Cited by 34 publications

(16 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…20 The algorithm is based on a previously developed quantitative method for collateral grading in patients with a large vessel occlusion. 21 The software first locates the ischemic hemisphere using an occlusion detection algorithm 22 and segments the vessels (arteries) distal to it. This is achieved by convolutional neural networks designed and trained specifically for these tasks.…”

Section: Methodsmentioning

confidence: 99%

Time Since Stroke Onset, Quantitative Collateral Score, and Functional Outcome After Endovascular Treatment for Acute Ischemic Stroke

et al. 2022

Self Cite

View full text Add to dashboard Cite

Background:In patients with ischemic stroke undergoing endovascular treatment (EVT), time to treatment and collateral status are important prognostic factors and may be correlated. We aimed to assess the relation between time to CT angiography (CTA) and a quantitatively determined collateral score and to assess whether the collateral score modified the relation between time to recanalization and functional outcome.Methods:We analysed data from patients with acute ischemic stroke included in the MR CLEAN Registry between 2014 and 2017, who had a carotid terminus or M1 occlusion and were treated with EVT within 6.5 hours of symptom onset. A quantitative collateral score (qCS) was determined from baseline CTA using a validated automated image analysis algorithm. We also determined a 4-point visual collateral score (vCS). Multivariable regression models were used to assess the relations between time to imaging and the qCS, and between the time to recanalization and functional outcome (90-day modified Rankin Scale score). An interaction term (time to recanalization * qCS) was entered in the latter model to test whether the qCS modifies this relation. Sensitivity analyses were performed using the vCS.Results:We analyzed 1813 patients. Median time from symptom onset to CTA was 91 minutes [IQR:65-150] and the median qCS was 49% [IQR:25-78]. Longer time to CTA was not associated with the log-transformed qCS (adjusted β per 30 minutes, 0.002, 95%CI: -0.006 – 0.011). Both a higher qCS (acOR per 10% increase: 1.06, 95%CI: 1.03-1.09) and shorter time to recanalization (acOR per 30 minutes: 1.17, 95%CI: 1.13-1.22) were independently associated with a shift towards better functional outcome. The qCS did not modify the relation between time to recanalization and functional outcome (p for interaction: 0.28). Results from sensitivity analyses using the vCS were similar.Conclusions:In the first 6.5 hours of ischemic stroke caused by carotid-terminus or M1 occlusion, the collateral status is unaffected by time to imaging and the benefit of a shorter time to recanalization is independent of baseline collateral status.

show abstract

Section: Methodsmentioning

confidence: 99%

Time Since Stroke Onset, Quantitative Collateral Score, and Functional Outcome After Endovascular Treatment for Acute Ischemic Stroke

et al. 2022

Self Cite

View full text Add to dashboard Cite

show abstract

“…We intend to use this tree to e.g. find lesions (occlusions in more distal arteries, such as M2, which is still a challenging task Luijten et al, 2021), better quantification of collateral status, and possibly also linking/registering the 3D vascular information from CTA and CTP to interventional DSA images for improving image guidance during endovascular treatment. The tree formation method performed reasonably well in the proximal and middle segments.…”

Section: Discussionmentioning

confidence: 99%

Deep reinforcement learning for cerebral anterior vessel tree extraction from 3D CTA images

Wolff

et al. 2023

Medical Image Analysis

View full text Add to dashboard Cite

“…We focused on locating vessel occlusions instead of detecting vessel occlusions. Studies that automatically detected vessel occlusions on CTA did not evaluate their performance in locating vessel occlusions ( 19 – 24 ). Instead, these studies were designed to distinguish a group of patients with a large vessel occlusion from a group of patients without a detectable vessel occlusion.…”

Section: Discussionmentioning

confidence: 99%

Spatial CT perfusion data helpful in automatically locating vessel occlusions for acute ischemic stroke patients

et al. 2023

View full text Add to dashboard Cite

IntroductionLocating a vessel occlusion is important for clinical decision support in stroke healthcare. The advent of endovascular thrombectomy beyond proximal large vessel occlusions spurs alternative approaches to locate vessel occlusions. We explore whether CT perfusion (CTP) data can help to automatically locate vessel occlusions.MethodsWe composed an atlas with the downstream regions of particular vessel segments. Occlusion of these segments should result in the hypoperfusion of the corresponding downstream region. We differentiated between seven-vessel occlusion locations (ICA, proximal M1, distal M1, M2, M3, ACA, and posterior circulation). We included 596 patients from the DUtch acute STroke (DUST) multicenter study. Each patient CTP data set was processed with perfusion software to determine the hypoperfused region. The downstream region with the highest overlap with the hypoperfused region was considered to indicate the vessel occlusion location. We assessed the indications from CTP against expert annotations from CTA.ResultsOur atlas-based model had a mean accuracy of 86% and could achieve substantial agreement with the annotations from CTA according to Cohen's kappa coefficient (up to 0.68). In particular, anterior large vessel occlusions and occlusions in the posterior circulation could be located with an accuracy of 80 and 92%, respectively.ConclusionThe spatial layout of the hypoperfused region can help to automatically indicate the vessel occlusion location for acute ischemic stroke patients. However, variations in vessel architecture between patients seemed to limit the capacity of CTP data to distinguish between vessel occlusion locations more accurately.

show abstract

Diagnostic performance of an algorithm for automated large vessel occlusion detection on CT angiography

Cited by 34 publications

References 30 publications

Time Since Stroke Onset, Quantitative Collateral Score, and Functional Outcome After Endovascular Treatment for Acute Ischemic Stroke

Time Since Stroke Onset, Quantitative Collateral Score, and Functional Outcome After Endovascular Treatment for Acute Ischemic Stroke

Deep reinforcement learning for cerebral anterior vessel tree extraction from 3D CTA images

Spatial CT perfusion data helpful in automatically locating vessel occlusions for acute ischemic stroke patients

Contact Info

Product

Resources

About