Yunbao Cao1, Jing Yu2, Hu Zhang1, Jian Xiong1, Zhonghua Luo1. 1. Department of Interventional Radiology, Tangdu Hospital, Fourth Military Medical University, Xi'an, China. 2. Department of Radiology, Xijing Hospital, Fourth Military Medical University, Xi'an, China.
Abstract
Background: Computed tomography (CT) is a common imaging technique for diagnosis of liver tumors. However, the intensity similarity on non-contrast CT images is small, making it difficult for radiologists to visually identify hepatic cavernous hemangioma (HCH) and hepatocellular carcinoma (HCC). Recently, convolutional neural networks (CNN) have been widely used in the study of medical image classification because more discriminative image features can be extracted than the human eye. Therefore, this study focused on developing a CNN model for identifying HCH and HCC. Methods: This study is a retrospective study. A dataset consisting of 774 non-contrast CT images was collected from 50 patients with HCC or HCH, and the ground truth was given by three radiologists based on contrast-enhanced CT. Firstly, the non-contrast CT images dataset were randomly divided into a training set (n=559) and a test set (n=215). Then, we performed preprocessing of the non-contrast CT images using pseudo-color conversion, and the proposed CNN model developed using training set. Finally, the following indicators (accuracy, precision, recall) were used to quantitatively analyze the results. Results: In the test set, the proposed CNN model achieved a high classification accuracy of 84.25%, precision of 81.36%, and recall of 82.18%. Conclusions: The CNN model for identifying HCH and HCC improves the accuracy of diagnosis on non-contrast CT images. 2022 Journal of Gastrointestinal Oncology. All rights reserved.
Background: Computed tomography (CT) is a common imaging technique for diagnosis of liver tumors. However, the intensity similarity on non-contrast CT images is small, making it difficult for radiologists to visually identify hepatic cavernous hemangioma (HCH) and hepatocellular carcinoma (HCC). Recently, convolutional neural networks (CNN) have been widely used in the study of medical image classification because more discriminative image features can be extracted than the human eye. Therefore, this study focused on developing a CNN model for identifying HCH and HCC. Methods: This study is a retrospective study. A dataset consisting of 774 non-contrast CT images was collected from 50 patients with HCC or HCH, and the ground truth was given by three radiologists based on contrast-enhanced CT. Firstly, the non-contrast CT images dataset were randomly divided into a training set (n=559) and a test set (n=215). Then, we performed preprocessing of the non-contrast CT images using pseudo-color conversion, and the proposed CNN model developed using training set. Finally, the following indicators (accuracy, precision, recall) were used to quantitatively analyze the results. Results: In the test set, the proposed CNN model achieved a high classification accuracy of 84.25%, precision of 81.36%, and recall of 82.18%. Conclusions: The CNN model for identifying HCH and HCC improves the accuracy of diagnosis on non-contrast CT images. 2022 Journal of Gastrointestinal Oncology. All rights reserved.
Authors: Clinton J Wang; Charlie A Hamm; Lynn J Savic; Marc Ferrante; Isabel Schobert; Todd Schlachter; MingDe Lin; Jeffrey C Weinreb; James S Duncan; Julius Chapiro; Brian Letzen Journal: Eur Radiol Date: 2019-05-15 Impact factor: 5.315
Authors: Michael R Rudnick; Amanda K Leonberg-Yoo; Harold I Litt; Raphael M Cohen; Susan Hilton; Peter P Reese Journal: Am J Kidney Dis Date: 2019-08-28 Impact factor: 8.860