Initial Experiences with Artificial Neural Networks in the Detection of Computed Tomography Perfusion Deficits

Background: Head computed tomography (CT) with perfusion imaging has become crucial in the selection of patients for mechanical thrombectomy. In recent years, machine learning has rapidly evolved and found applications in a wide variety of health care tasks. We report our initial experiences with training a neural network to predict the presence and laterality of a perfusion deficit in patients with acute ischemic stroke. Methods: CT perfusion images of patients with suspicion for acute ischemic stroke were obtained. The data were split into training and validation sets. A long-term, recurrent convolutional network was constructed, which consisted of a convolutional neural network stacked on top of a long short-term memory layer. Results: Of the 396 patients, 139 (35.1%) had a right-sided perfusion deficit, 199 (50.3%) had a left-sided deficit, and 58 (14.6%) had no evidence of a deficit. The best model was able to achieve an accuracy of 85.8% on validation data. Receiver operating characteristic curves were generated for each class, and an area under the curve (AUC) was calculated for each class. For right-sided deficits, the AUC was 0.90, for left-sided deficits, the AUC was 0.96, and for no deficit, the AUC was 0.93. Conclusions: The field of machine learning, powered by convolutional neural networks for the task of image recognition and processing, has quickly developed in recent years. We constructed an artificial neural network that can identify and classify the presence and laterality of a perfusion deficit on CT perfusion imaging.