Thrombus imaging characteristics are associated with treatment success and functional outcomes in stroke patients. However, assessing these characteristics based on manual annotations is labor intensive and subject to observer bias. Therefore, we aimed to create an automated pipeline for consistent and fast full thrombus segmentation. We used multi-center, multi-scanner datasets of anterior circulation stroke patients with baseline NCCT and CTA for training (n = 228) and testing (n = 100). We first found the occlusion location using StrokeViewer LVO and created a bounding box around it. Subsequently, we trained dual modality U-Net based convolutional neural networks (CNNs) to segment the thrombus inside this bounding box. We experimented with: (1) U-Net with two input channels for NCCT and CTA, and U-Nets with two encoders where (2) concatenate, (3) add, and (4) weighted-sum operators were used for feature fusion. Furthermore, we proposed a dynamic bounding box algorithm to adjust the bounding box. The dynamic bounding box algorithm reduces the missed cases but does not improve Dice. The two-encoder U-Net with a weighted-sum feature fusion shows the best performance (surface Dice 0.78, Dice 0.62, and 4% missed cases). Final segmentation results have high spatial accuracies and can therefore be used to determine thrombus characteristics and potentially benefit radiologists in clinical practice.
Background: It remains unclear if deep-learning-based white matter lesion (DL-WML) volume can predict outcome after ischemic stroke. Purpose: We aimed to develop, validate, and evaluate DL-WML volume in NCCT as a risk factor and IVT effect modifier compared to the Fazekas scale (WML-Faz) in patients also receiving EVT in an EVT-capable center. Methods: A deep-learning model for WML segmentation in NCCT was developed and validated internally and externally. The volumetric correspondence of DL-WML volume per mL was reported relative to expert annotation with the intraclass correlation coefficient (ICC) and a Bland-Altman analysis reporting bias and limits of agreement (LoA). In a post-hoc analysis of the MR CLEAN No-IV trial, univariable and multivariable regression models were used to report (un)adjusted common odds ratios ([a]cOR) to associate DL-WML volume and WML-Faz with the occurrence of symptomatic-intracerebral hemorrhage (sICH) and 90-day functional outcome with the modified Rankin Scale (mRS). Results: DL-WML volumes were comparable with those of the ground truth for both the internal test set (10/20(50%) male, age median:72[IQR:67-85], ICC mean:0.91 95%CI:[0.87;0.94];bias:-3mL LoA:[-12mL;7mL]) and the external test set (36/101(36%) male, age median:59[IQR:42-73], ICC mean:0.87 95%CI:[0.71;0.95];bias:-2mL LoA:[-11mL;7mL]). 516 patients from the MR CLEAN No-IV trial (291/516(56%) male, age median:71 IQR:[62-79],) were analyzed. Both DL-WML volume and WML-Faz were associated with sICH (DL-WML volume acOR:1.31 95%CI[1.08;1.60], WML-Faz acOR:1.53 95%CI[1.02;2.31]) and mRS (DL-WML volume acOR:0.84 95%CI[0.76;0.94], WML-Faz acOR:0.73 95%CI[0.60;0.88]). Only for the unadjusted analysis, WML-Faz was an IVT effect modifier (p=0.046), DL-WML was not (p=0.274). Conclusion: DL-WML volume and WML-Faz had a similar relationship with functional outcome and sICH.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.