Improving B-mode ultrasound diagnostic performance for focal liver lesions using deep learning: A multicentre study
Background: The diagnosis performance of B-mode ultrasound (US) for focal liver lesions (FLLs) is relatively limited. We aimed to develop a deep convolutional neural network of US (DCNN-US) for aiding radiologists in classification of malignant from benign FLLs. Materials and methods: This study was...
| Main Authors: | , , , , , , , , , , , , , , , , , , , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2020-06-01
|
| Series: | EBioMedicine |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2352396420301523 |
| _version_ | 1818921818866581504 |
|---|---|
| author | Qi Yang Jingwei Wei Xiaohan Hao Dexing Kong Xiaoling Yu Tianan Jiang Junqing Xi Wenjia Cai Yanchun Luo Xiang Jing Yilin Yang Zhigang Cheng Jinyu Wu Huiping Zhang Jintang Liao Pei Zhou Yu Song Yao Zhang Zhiyu Han Wen Cheng Lina Tang Fangyi Liu Jianping Dou Rongqin Zheng Jie Yu Jie Tian Ping Liang |
| author_facet | Qi Yang Jingwei Wei Xiaohan Hao Dexing Kong Xiaoling Yu Tianan Jiang Junqing Xi Wenjia Cai Yanchun Luo Xiang Jing Yilin Yang Zhigang Cheng Jinyu Wu Huiping Zhang Jintang Liao Pei Zhou Yu Song Yao Zhang Zhiyu Han Wen Cheng Lina Tang Fangyi Liu Jianping Dou Rongqin Zheng Jie Yu Jie Tian Ping Liang |
| author_sort | Qi Yang |
| collection | DOAJ |
| description | Background: The diagnosis performance of B-mode ultrasound (US) for focal liver lesions (FLLs) is relatively limited. We aimed to develop a deep convolutional neural network of US (DCNN-US) for aiding radiologists in classification of malignant from benign FLLs. Materials and methods: This study was conducted in 13 hospitals and finally 2143 patients with 24,343 US images were enrolled. Patients who had non-cystic FLLs with pathological results were enrolled. The FLLs from 11 hospitals were randomly divided into training and internal validations (IV) cohorts with a 4:1 ratio for developing and evaluating DCNN-US. Diagnostic performance of the model was verified using external validation (EV) cohort from another two hospitals. The diagnosis value of DCNN-US was compared with that of contrast enhanced computed tomography (CT)/magnetic resonance image (MRI) and 236 radiologists, respectively. Findings: The AUC of ModelLBC for FLLs was 0.924 (95% CI: 0.889–0.959) in the EV cohort. The diagnostic sensitivity and specificity of ModelLBC were superior to 15-year skilled radiologists (86.5% vs 76.1%, p = 0.0084 and 85.5% vs 76.9%, p = 0.0051, respectively). Accuracy of ModelLBC was comparable to that of contrast enhanced CT (both 84.7%) but inferior to contrast enhanced MRI (87.9%) for lesions detected by US. Interpretation: DCNN-US with high sensitivity and specificity in diagnosing FLLs shows its potential to assist less-experienced radiologists in improving their performance and lowering their dependence on sectional imaging in liver cancer diagnosis. |
| first_indexed | 2024-12-20T01:43:42Z |
| format | Article |
| id | doaj.art-47107e92e01a409e9fea1a6ebc4f149f |
| institution | Directory Open Access Journal |
| issn | 2352-3964 |
| language | English |
| last_indexed | 2024-12-20T01:43:42Z |
| publishDate | 2020-06-01 |
| publisher | Elsevier |
| record_format | Article |
| series | EBioMedicine |
| spelling | doaj.art-47107e92e01a409e9fea1a6ebc4f149f2022-12-21T19:57:50ZengElsevierEBioMedicine2352-39642020-06-0156102777Improving B-mode ultrasound diagnostic performance for focal liver lesions using deep learning: A multicentre studyQi Yang0Jingwei Wei1Xiaohan Hao2Dexing Kong3Xiaoling Yu4Tianan Jiang5Junqing Xi6Wenjia Cai7Yanchun Luo8Xiang Jing9Yilin Yang10Zhigang Cheng11Jinyu Wu12Huiping Zhang13Jintang Liao14Pei Zhou15Yu Song16Yao Zhang17Zhiyu Han18Wen Cheng19Lina Tang20Fangyi Liu21Jianping Dou22Rongqin Zheng23Jie Yu24Jie Tian25Ping Liang26Department of Interventional Ultrasound, Chinese PLA General Hospital, 28 Fuxing Road, Beijing 100853, ChinaKey Laboratory of Molecular Imaging, Institute of Automation, Chinese Academy of Sciences, Beijing, China; University of Chinese Academy of Sciences, Beijing, ChinaKey Laboratory of Molecular Imaging, Institute of Automation, Chinese Academy of Sciences, Beijing, China; University of Chinese Academy of Sciences, Beijing, China; Centers for Biomedical Engineering, University of Science and Technology of China, University of Science and Technology of China, Hefei, ChinaSchool of Mathematical Sciences, Zhejiang University, Hangzhou, ChinaDepartment of Interventional Ultrasound, Chinese PLA General Hospital, 28 Fuxing Road, Beijing 100853, ChinaDepartment of Ultrasound, the First Affiliated hospital, College of Medicine, Zhejiang University, Hangzhou, Jiangsu, ChinaDepartment of Interventional Ultrasound, Chinese PLA General Hospital, 28 Fuxing Road, Beijing 100853, ChinaDepartment of Interventional Ultrasound, Chinese PLA General Hospital, 28 Fuxing Road, Beijing 100853, ChinaDepartment of Interventional Ultrasound, Chinese PLA General Hospital, 28 Fuxing Road, Beijing 100853, ChinaDepartment of Ultrasound, Tianjin Third Central Hospital, Tianjin, ChinaDepartment of Ultrasound Diagnosis, Tangdu Hospital, Fourth Military Medical University, Xi'an, ChinaDepartment of Interventional Ultrasound, Chinese PLA General Hospital, 28 Fuxing Road, Beijing 100853, ChinaDepartment of Ultrasound, Harbin The First Hospital, Harbin, ChinaDepartment of Medical Ultrasound, Ma'anshan People's Hospital, Ma'anshan, ChinaDepartment of Diagnostic Ultrasound, Xiangya Hospital, Changsha, ChinaDepartment of Ultrasound, Central Theater Command General Hospital, Chinese People's Liberation Army, Wuhan, ChinaDepartment of Diagnostic Ultrasound, The Second Affiliated Hospital of Dalian Medical University, Dalian, ChinaDepartment of Ultrasound, Beijing Ditan Hospital, Capital Medical University, Beijing, ChinaDepartment of Interventional Ultrasound, Chinese PLA General Hospital, 28 Fuxing Road, Beijing 100853, ChinaDepartment of Ultrasound, Harbin Medical University Cancer Hospital, Harbin, ChinaDepartment of Ultrasound, Fujian Cancer Hospital&Fujian Medical University Cancer Hospita, Fuzhou, ChinaDepartment of Interventional Ultrasound, Chinese PLA General Hospital, 28 Fuxing Road, Beijing 100853, ChinaDepartment of Interventional Ultrasound, Chinese PLA General Hospital, 28 Fuxing Road, Beijing 100853, ChinaGuangdong Key Laboratory of Liver Disease Research, Department of Medical Ultrasound, The Third Affiliated Hospital of Sun Yat-sen University, Guangzhou, China; Corresponding author at: Guangdong Key Laboratory of Liver Disease Research, Department of Medical Ultrasound, The Third Affiliated Hospital of Sun Yat-sen University, Guangzhou, ChinaDepartment of Interventional Ultrasound, Chinese PLA General Hospital, 28 Fuxing Road, Beijing 100853, China; Corresponding author at: Department of Interventional Ultrasound, Chinese PLA General Hospital, 28 Fuxing Road, Beijing 100853, ChinaKey Laboratory of Molecular Imaging, Institute of Automation, Chinese Academy of Sciences, Beijing, China; University of Chinese Academy of Sciences, Beijing, China; Corresponding author at: Key Laboratory of Molecular Imaging, Institute of Automation, Chinese Academy of Sciences, Beijing, ChinaDepartment of Interventional Ultrasound, Chinese PLA General Hospital, 28 Fuxing Road, Beijing 100853, China; Corresponding author at: Department of Interventional Ultrasound, Chinese PLA General Hospital, 28 Fuxing Road, Beijing 100853, ChinaBackground: The diagnosis performance of B-mode ultrasound (US) for focal liver lesions (FLLs) is relatively limited. We aimed to develop a deep convolutional neural network of US (DCNN-US) for aiding radiologists in classification of malignant from benign FLLs. Materials and methods: This study was conducted in 13 hospitals and finally 2143 patients with 24,343 US images were enrolled. Patients who had non-cystic FLLs with pathological results were enrolled. The FLLs from 11 hospitals were randomly divided into training and internal validations (IV) cohorts with a 4:1 ratio for developing and evaluating DCNN-US. Diagnostic performance of the model was verified using external validation (EV) cohort from another two hospitals. The diagnosis value of DCNN-US was compared with that of contrast enhanced computed tomography (CT)/magnetic resonance image (MRI) and 236 radiologists, respectively. Findings: The AUC of ModelLBC for FLLs was 0.924 (95% CI: 0.889–0.959) in the EV cohort. The diagnostic sensitivity and specificity of ModelLBC were superior to 15-year skilled radiologists (86.5% vs 76.1%, p = 0.0084 and 85.5% vs 76.9%, p = 0.0051, respectively). Accuracy of ModelLBC was comparable to that of contrast enhanced CT (both 84.7%) but inferior to contrast enhanced MRI (87.9%) for lesions detected by US. Interpretation: DCNN-US with high sensitivity and specificity in diagnosing FLLs shows its potential to assist less-experienced radiologists in improving their performance and lowering their dependence on sectional imaging in liver cancer diagnosis.http://www.sciencedirect.com/science/article/pii/S2352396420301523UltrasoundConvolutional neural networkFocal liver lesionsDiagnosis |
| spellingShingle | Qi Yang Jingwei Wei Xiaohan Hao Dexing Kong Xiaoling Yu Tianan Jiang Junqing Xi Wenjia Cai Yanchun Luo Xiang Jing Yilin Yang Zhigang Cheng Jinyu Wu Huiping Zhang Jintang Liao Pei Zhou Yu Song Yao Zhang Zhiyu Han Wen Cheng Lina Tang Fangyi Liu Jianping Dou Rongqin Zheng Jie Yu Jie Tian Ping Liang Improving B-mode ultrasound diagnostic performance for focal liver lesions using deep learning: A multicentre study EBioMedicine Ultrasound Convolutional neural network Focal liver lesions Diagnosis |
| title | Improving B-mode ultrasound diagnostic performance for focal liver lesions using deep learning: A multicentre study |
| title_full | Improving B-mode ultrasound diagnostic performance for focal liver lesions using deep learning: A multicentre study |
| title_fullStr | Improving B-mode ultrasound diagnostic performance for focal liver lesions using deep learning: A multicentre study |
| title_full_unstemmed | Improving B-mode ultrasound diagnostic performance for focal liver lesions using deep learning: A multicentre study |
| title_short | Improving B-mode ultrasound diagnostic performance for focal liver lesions using deep learning: A multicentre study |
| title_sort | improving b mode ultrasound diagnostic performance for focal liver lesions using deep learning a multicentre study |
| topic | Ultrasound Convolutional neural network Focal liver lesions Diagnosis |
| url | http://www.sciencedirect.com/science/article/pii/S2352396420301523 |
| work_keys_str_mv | AT qiyang improvingbmodeultrasounddiagnosticperformanceforfocalliverlesionsusingdeeplearningamulticentrestudy AT jingweiwei improvingbmodeultrasounddiagnosticperformanceforfocalliverlesionsusingdeeplearningamulticentrestudy AT xiaohanhao improvingbmodeultrasounddiagnosticperformanceforfocalliverlesionsusingdeeplearningamulticentrestudy AT dexingkong improvingbmodeultrasounddiagnosticperformanceforfocalliverlesionsusingdeeplearningamulticentrestudy AT xiaolingyu improvingbmodeultrasounddiagnosticperformanceforfocalliverlesionsusingdeeplearningamulticentrestudy AT tiananjiang improvingbmodeultrasounddiagnosticperformanceforfocalliverlesionsusingdeeplearningamulticentrestudy AT junqingxi improvingbmodeultrasounddiagnosticperformanceforfocalliverlesionsusingdeeplearningamulticentrestudy AT wenjiacai improvingbmodeultrasounddiagnosticperformanceforfocalliverlesionsusingdeeplearningamulticentrestudy AT yanchunluo improvingbmodeultrasounddiagnosticperformanceforfocalliverlesionsusingdeeplearningamulticentrestudy AT xiangjing improvingbmodeultrasounddiagnosticperformanceforfocalliverlesionsusingdeeplearningamulticentrestudy AT yilinyang improvingbmodeultrasounddiagnosticperformanceforfocalliverlesionsusingdeeplearningamulticentrestudy AT zhigangcheng improvingbmodeultrasounddiagnosticperformanceforfocalliverlesionsusingdeeplearningamulticentrestudy AT jinyuwu improvingbmodeultrasounddiagnosticperformanceforfocalliverlesionsusingdeeplearningamulticentrestudy AT huipingzhang improvingbmodeultrasounddiagnosticperformanceforfocalliverlesionsusingdeeplearningamulticentrestudy AT jintangliao improvingbmodeultrasounddiagnosticperformanceforfocalliverlesionsusingdeeplearningamulticentrestudy AT peizhou improvingbmodeultrasounddiagnosticperformanceforfocalliverlesionsusingdeeplearningamulticentrestudy AT yusong improvingbmodeultrasounddiagnosticperformanceforfocalliverlesionsusingdeeplearningamulticentrestudy AT yaozhang improvingbmodeultrasounddiagnosticperformanceforfocalliverlesionsusingdeeplearningamulticentrestudy AT zhiyuhan improvingbmodeultrasounddiagnosticperformanceforfocalliverlesionsusingdeeplearningamulticentrestudy AT wencheng improvingbmodeultrasounddiagnosticperformanceforfocalliverlesionsusingdeeplearningamulticentrestudy AT linatang improvingbmodeultrasounddiagnosticperformanceforfocalliverlesionsusingdeeplearningamulticentrestudy AT fangyiliu improvingbmodeultrasounddiagnosticperformanceforfocalliverlesionsusingdeeplearningamulticentrestudy AT jianpingdou improvingbmodeultrasounddiagnosticperformanceforfocalliverlesionsusingdeeplearningamulticentrestudy AT rongqinzheng improvingbmodeultrasounddiagnosticperformanceforfocalliverlesionsusingdeeplearningamulticentrestudy AT jieyu improvingbmodeultrasounddiagnosticperformanceforfocalliverlesionsusingdeeplearningamulticentrestudy AT jietian improvingbmodeultrasounddiagnosticperformanceforfocalliverlesionsusingdeeplearningamulticentrestudy AT pingliang improvingbmodeultrasounddiagnosticperformanceforfocalliverlesionsusingdeeplearningamulticentrestudy |