Publications-Periodical Articles
Article View/Open
Publication Export
-
題名 Rapid Assessment of Acute Ischemic Stroke by Computed Tomography using Deep Convolutional Neural Networks 作者 羅崇銘
Lo, Chung-Ming
Hung, Peng-Hsiang
Lin, Daw-Tung貢獻者 圖檔所 關鍵詞 Acute ischemic stroke;Computed tomography;Convolutional neural networks 日期 2021-05 上傳時間 28-Mar-2022 15:59:04 (UTC+8) 摘要 Acute stroke is one of the leading causes of disability and death worldwide. Regarding clinical diagnoses, a rapid and accurate procedure is necessary for patients suffering from acute stroke. This study proposes an automatic identification scheme for acute ischemic stroke using deep convolutional neural networks (DCNNs) based on non-contrast computed tomographic (NCCT) images. Our image database for the classification model was composed of 1254 grayscale NCCT images from 96 patients (573 images) with acute ischemic stroke and 121 normal controls (681 images). According to the consensus of critical stroke findings by two neuroradiologists, a gold standard was established and used to train the proposed DCNN using machine-generated image features. Including the earliest DCNN, AlexNet, the popular Inception-v3, and ResNet-101 were proposed. To train the limited data size, transfer learning with ImageNet parameters was also used. The established models were evaluated by tenfold cross-validation and tested on an independent dataset containing 50 patients with acute ischemic stroke (108 images) and 58 normal controls (117 images) from another institution. AlexNet without pretrained parameters achieved an accuracy of 97.12%, a sensitivity of 98.11%, a specificity of 96.08%, and an area under the receiver operating characteristic curve (AUC) of 0.9927. Using transfer learning, transferred AlexNet, transferred Inception-v3, and transferred ResNet-101 achieved accuracies between 90.49 and 95.49%. Tested with a dataset from another institution, AlexNet showed an accuracy of 60.89%, a sensitivity of 18.52%, and a specificity of 100%. Transferred AlexNet, Inception-v3, and ResNet-101 achieved accuracies of 81.77%, 85.78%, and 80.89%, respectively. The proposed DCNN architecture as a computer-aided diagnosis system showed that training from scratch can generate a customized model for a specific scanner, and transfer learning can generate a more generalized model to provide diagnostic suggestions of acute ischemic stroke to radiologists. 關聯 Journal of Digital Imaging, Vol.34, pp.637–646 資料類型 article DOI https://doi.org/10.1007/s10278-021-00457-y dc.contributor 圖檔所 dc.creator (作者) 羅崇銘 dc.creator (作者) Lo, Chung-Ming dc.creator (作者) Hung, Peng-Hsiang dc.creator (作者) Lin, Daw-Tung dc.date (日期) 2021-05 dc.date.accessioned 28-Mar-2022 15:59:04 (UTC+8) - dc.date.available 28-Mar-2022 15:59:04 (UTC+8) - dc.date.issued (上傳時間) 28-Mar-2022 15:59:04 (UTC+8) - dc.identifier.uri (URI) http://nccur.lib.nccu.edu.tw/handle/140.119/139474 - dc.description.abstract (摘要) Acute stroke is one of the leading causes of disability and death worldwide. Regarding clinical diagnoses, a rapid and accurate procedure is necessary for patients suffering from acute stroke. This study proposes an automatic identification scheme for acute ischemic stroke using deep convolutional neural networks (DCNNs) based on non-contrast computed tomographic (NCCT) images. Our image database for the classification model was composed of 1254 grayscale NCCT images from 96 patients (573 images) with acute ischemic stroke and 121 normal controls (681 images). According to the consensus of critical stroke findings by two neuroradiologists, a gold standard was established and used to train the proposed DCNN using machine-generated image features. Including the earliest DCNN, AlexNet, the popular Inception-v3, and ResNet-101 were proposed. To train the limited data size, transfer learning with ImageNet parameters was also used. The established models were evaluated by tenfold cross-validation and tested on an independent dataset containing 50 patients with acute ischemic stroke (108 images) and 58 normal controls (117 images) from another institution. AlexNet without pretrained parameters achieved an accuracy of 97.12%, a sensitivity of 98.11%, a specificity of 96.08%, and an area under the receiver operating characteristic curve (AUC) of 0.9927. Using transfer learning, transferred AlexNet, transferred Inception-v3, and transferred ResNet-101 achieved accuracies between 90.49 and 95.49%. Tested with a dataset from another institution, AlexNet showed an accuracy of 60.89%, a sensitivity of 18.52%, and a specificity of 100%. Transferred AlexNet, Inception-v3, and ResNet-101 achieved accuracies of 81.77%, 85.78%, and 80.89%, respectively. The proposed DCNN architecture as a computer-aided diagnosis system showed that training from scratch can generate a customized model for a specific scanner, and transfer learning can generate a more generalized model to provide diagnostic suggestions of acute ischemic stroke to radiologists. dc.format.extent 1561763 bytes - dc.format.mimetype application/pdf - dc.relation (關聯) Journal of Digital Imaging, Vol.34, pp.637–646 dc.subject (關鍵詞) Acute ischemic stroke;Computed tomography;Convolutional neural networks dc.title (題名) Rapid Assessment of Acute Ischemic Stroke by Computed Tomography using Deep Convolutional Neural Networks dc.type (資料類型) article dc.identifier.doi (DOI) 10.1007/s10278-021-00457-y dc.doi.uri (DOI) https://doi.org/10.1007/s10278-021-00457-y
