Abstract
In stroke imaging, CT angiography (CTA) is used for detecting arterial occlusions. These images could also provide information on the extent of ischemia. The study aim was to develop and evaluate a convolutional neural network (CNN)–based algorithm for detecting and segmenting acute ischemic lesions from CTA images of patients with suspected middle cerebral artery stroke. These results were compared to volumes reported by widely used CT perfusion–based RAPID software (IschemaView). A 42-layer-deep CNN was trained on 50 CTA volumes with manually delineated targets. The lower bound for predicted lesion size to reliably discern stroke from false positives was estimated. The severity of false positives and false negatives was reviewed visually to assess the clinical applicability and to further guide the method development. The CNN model corresponded to the manual segmentations with voxel-wise sensitivity 0.54 (95% confidence interval: 0.44–0.63), precision 0.69 (0.60–0.76), and Sørensen–Dice coefficient 0.61 (0.52–0.67). Stroke/nonstroke differentiation accuracy 0.88 (0.81–0.94) was achieved when only considering the predicted lesion size (i.e., regardless of location). By visual estimation, 46% of cases showed some false findings, such as CNN highlighting chronic periventricular white matter changes or beam hardening artifacts, but only in 9% the errors were severe, translating to 0.91 accuracy. The CNN model had a moderately strong correlation to RAPID-reported Tmax > 10 s volumes (Pearson’s r = 0.76 (0.58–0.86)). The results suggest that detecting anterior circulation ischemic strokes from CTA using a CNN-based algorithm can be feasible when accompanied with physiological knowledge to rule out false positives.
Highlights
Applying artificial intelligence in medical research has experienced an exponential growth in interest over the past decades [1]
As shown in a preceding paper, a convolutional neural network (CNN) model is feasible in detecting ischemic parenchymal regions associated with acute thrombosis of the middle cerebral artery (MCA) in computed tomography (CT) angiography (CTA) images [17]
The proposed deep CNN model does not process any CT perfusion (CTP) or magnetic resonance (MR) diffusion data, providing an alternative with higher resolution and smaller ionizing radiation dose compared to CTP
Summary
Applying artificial intelligence in medical research has experienced an exponential growth in interest over the past decades [1]. As shown in a preceding paper, a CNN model is feasible in detecting ischemic parenchymal regions associated with acute thrombosis of the middle cerebral artery (MCA) in CTA images [17]. In addition to the published studies, several commercial solutions have been made available based on similar methods [18]. A CTA-based, hemispheric asymmetry aware, 42-layer-deep CNN model was developed, and its performance was compared to a commercial CTP-based software RAPID (iSchemaView, Menlo Park, California, USA). The performance of RAPID has been evaluated in earlier studies by comparing RAPID CTP analysis to diffusion–perfusion mismatch and prediction of final infarct volume from MR images [20]. CTA is routinely acquired to detect large vessel occlusions, and the ischemic regions are not typically evaluated from these images. The proposed method could complement perfusion analysis (e.g., in cases where the perfusion study is non-diagnostic) without changes to the existing imaging protocols
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
