Tags

Type your tag names separated by a space and hit enter

Real-time automated diagnosis of precancerous lesion and early esophageal squamous cell carcinoma using a deep learning model (with videos).

Abstract

BACKGROUND AND AIMS

We developed a computer-assisted diagnostic (CAD) system for real-time automated diagnosis of precancerous lesions and early esophageal squamous cell carcinoma (ESCC) to assist the diagnosis of esophageal cancer.

METHODS

A total of 6,473 narrow-band imaging (NBI) images, including precancerous lesions, early ESCC, and noncancerous lesions were used to train the CAD system. We validated the CAD system using both endoscopic images and video datasets. The receiver operating characteristic (ROC) curve of the CAD system was generated based on image datasets. An artificial intelligence (AI) probability heat map was generated for each input of endoscopic images. The red color indicated high possibility of cancerous lesion, whereas the blue color indicated noncancerous lesions on the probability heat map. When the CAD system detected any precancerous lesion or early ESCC, the lesion of interest was masked with color.

RESULTS

The image datasets contained 1,480 malignant NBI images from 59 consecutive cancerous cases (sensitivity, 98.04%) and 5,191 noncancerous NBI images from 2,004 cases (specificity, 95.03%). The area under curve (AUC) was 0.989. The video datasets of precancerous lesions or early ESCC included 27 nonmagnifying videos (per-frame-sensitivity 60.8%, per-lesion-sensitivity, 100%) and 20 magnifying videos (per-frame-sensitivity 96.1%, per-lesion-sensitivity, 100%). Unaltered full-range normal esophagus videos included 33 videos (per-frame-specificity 99.9%, per-case specificity, 90.9%).

CONCLUSIONS

A deep learning model demonstrated high sensitivity and specificity in both endoscopic images and video datasets. The real-time CAD system has a promising use in the near future to assist endoscopists in diagnosing precancerous lesions and ESCC.

Authors+Show Affiliations

Department of Gastroenterology, West China Hospital, Sichuan University, China.Shanghai Wision AI Co., Ltd. Shanghai, China.Department of Gastroenterology, West China Hospital, Sichuan University, China.Department of Gastroenterology, West China Hospital, Sichuan University, China.Department of Gastroenterology, West China Hospital, Sichuan University, China.Department of Gastroenterology, West China Hospital, Sichuan University, China.Department of Gastroenterology, West China Hospital, Sichuan University, China.Department of Gastroenterology, West China Hospital, Sichuan University, China.Shanghai Wision AI Co., Ltd. Shanghai, China.Shanghai Wision AI Co., Ltd. Shanghai, China.Shanghai Wision AI Co., Ltd. Shanghai, China.San Bernardino Gastroenterology Associates, Inc. & ACE Endoscopy and Surgery Center, Rialto CA, USA.Jaswant Rai Speciality Hospital, Meerut, India.Shanghai Wision AI Co., Ltd. Shanghai, China.Department of Gastroenterology, West China Hospital, Sichuan University, China. Electronic address: hubingnj@163.com.

Pub Type(s)

Journal Article

Language

eng

PubMed ID

31445040

Citation

Guo, LinJie, et al. "Real-time Automated Diagnosis of Precancerous Lesion and Early Esophageal Squamous Cell Carcinoma Using a Deep Learning Model (with Videos)." Gastrointestinal Endoscopy, 2019.
Guo L, Xiao X, Wu C, et al. Real-time automated diagnosis of precancerous lesion and early esophageal squamous cell carcinoma using a deep learning model (with videos). Gastrointest Endosc. 2019.
Guo, L., Xiao, X., Wu, C., Zeng, X., Zhang, Y., Du, J., ... Hu, B. (2019). Real-time automated diagnosis of precancerous lesion and early esophageal squamous cell carcinoma using a deep learning model (with videos). Gastrointestinal Endoscopy, doi:10.1016/j.gie.2019.08.018.
Guo L, et al. Real-time Automated Diagnosis of Precancerous Lesion and Early Esophageal Squamous Cell Carcinoma Using a Deep Learning Model (with Videos). Gastrointest Endosc. 2019 Aug 21; PubMed PMID: 31445040.
* Article titles in AMA citation format should be in sentence-case
TY - JOUR T1 - Real-time automated diagnosis of precancerous lesion and early esophageal squamous cell carcinoma using a deep learning model (with videos). AU - Guo,LinJie, AU - Xiao,Xiao, AU - Wu,ChunCheng, AU - Zeng,Xianhui, AU - Zhang,Yuhang, AU - Du,Jiang, AU - Bai,Shuai, AU - Xie,Jia, AU - Zhang,Zhiwei, AU - Li,Yuhong, AU - Wang,Xuedan, AU - Cheung,Onpan, AU - Sharma,Malay, AU - Liu,Jingjia, AU - Hu,Bing, Y1 - 2019/08/21/ PY - 2019/04/15/received PY - 2019/08/08/accepted PY - 2019/8/25/entrez PY - 2019/8/25/pubmed PY - 2019/8/25/medline JF - Gastrointestinal endoscopy JO - Gastrointest. Endosc. N2 - BACKGROUND AND AIMS: We developed a computer-assisted diagnostic (CAD) system for real-time automated diagnosis of precancerous lesions and early esophageal squamous cell carcinoma (ESCC) to assist the diagnosis of esophageal cancer. METHODS: A total of 6,473 narrow-band imaging (NBI) images, including precancerous lesions, early ESCC, and noncancerous lesions were used to train the CAD system. We validated the CAD system using both endoscopic images and video datasets. The receiver operating characteristic (ROC) curve of the CAD system was generated based on image datasets. An artificial intelligence (AI) probability heat map was generated for each input of endoscopic images. The red color indicated high possibility of cancerous lesion, whereas the blue color indicated noncancerous lesions on the probability heat map. When the CAD system detected any precancerous lesion or early ESCC, the lesion of interest was masked with color. RESULTS: The image datasets contained 1,480 malignant NBI images from 59 consecutive cancerous cases (sensitivity, 98.04%) and 5,191 noncancerous NBI images from 2,004 cases (specificity, 95.03%). The area under curve (AUC) was 0.989. The video datasets of precancerous lesions or early ESCC included 27 nonmagnifying videos (per-frame-sensitivity 60.8%, per-lesion-sensitivity, 100%) and 20 magnifying videos (per-frame-sensitivity 96.1%, per-lesion-sensitivity, 100%). Unaltered full-range normal esophagus videos included 33 videos (per-frame-specificity 99.9%, per-case specificity, 90.9%). CONCLUSIONS: A deep learning model demonstrated high sensitivity and specificity in both endoscopic images and video datasets. The real-time CAD system has a promising use in the near future to assist endoscopists in diagnosing precancerous lesions and ESCC. SN - 1097-6779 UR - https://www.unboundmedicine.com/medline/citation/31445040/Real-time_automated_diagnosis_of_precancerous_lesion_and_early_esophageal_squamous_cell_carcinoma_using_a_deep_learning_model_(with_videos) L2 - https://linkinghub.elsevier.com/retrieve/pii/S0016-5107(19)32187-X DB - PRIME DP - Unbound Medicine ER -