Man Against Machine in Real-Time Videos: Performance of a Deep Learning System in Diagnosing Early Gastric Cancer and Predicting Invasion Depth and Differentiation Status in Comparison to 46 Endoscopists

2021 ◽  
Author(s):  
Qi Wu ◽  
Lianlian Wu ◽  
Jing Wang ◽  
Xinqi He ◽  
Yijie Zhu ◽  
...  
Keyword(s):  
Gastric Cancer ◽  
Deep Learning ◽  
Early Gastric Cancer ◽  
Real Time ◽  
Learning System ◽  
Invasion Depth ◽  
Differentiation Status

A deep learning-based system for identifying differentiation status and delineating the margins of early gastric cancer in magnifying narrow-band imaging endoscopy

Endoscopy ◽  
2020 ◽  
Author(s):  
Tingsheng Ling ◽  
Lianlian Wu ◽  
Yiwei Fu ◽  
Qinwei Xu ◽  
Ping An ◽  
...  
Keyword(s):  
Gastric Cancer ◽  
Deep Learning ◽  
Early Gastric Cancer ◽  
Real Time ◽  
Narrow Band ◽  
Narrow Band Imaging ◽  
Superior Performance ◽  
Accurate Identification ◽  
Real Time System ◽  
Differentiation Status

Abstract Background Accurate identification of the differentiation status and margins for early gastric cancer (EGC) is critical for determining the surgical strategy and achieving curative resection in EGC patients. The aim of this study was to develop a real-time system to accurately identify differentiation status and delineate the margins of EGC on magnifying narrow-band imaging (ME-NBI) endoscopy. Methods 2217 images from 145 EGC patients and 1870 images from 139 EGC patients were retrospectively collected to train and test the first convolutional neural network (CNN1) to identify EGC differentiation status. The performance of CNN1 was then compared with that of experts using 882 images from 58 EGC patients. Finally, 928 images from 132 EGC patients and 742 images from 87 EGC patients were used to train and test CNN2 to delineate the EGC margins. Results The system correctly predicted the differentiation status of EGCs with an accuracy of 83.3 % (95 % confidence interval [CI] 81.5 % – 84.9 %) in the testing dataset. In the man – machine contest, CNN1 performed significantly better than the five experts (86.2 %, 95 %CI 75.1 % – 92.8 % vs. 69.7 %, 95 %CI 64.1 % – 74.7 %). For delineating EGC margins, the system achieved an accuracy of 82.7 % (95 %CI 78.6 % – 86.1 %) in differentiated EGC and 88.1 % (95 %CI 84.2 % – 91.1 %) in undifferentiated EGC under an overlap ratio of 0.80. In unprocessed EGC videos, the system achieved real-time diagnosis of EGC differentiation status and EGC margin delineation in ME-NBI endoscopy. Conclusion We developed a deep learning-based system to accurately identify differentiation status and delineate the margins of EGC in ME-NBI endoscopy. This system achieved superior performance when compared with experts and was successfully tested in real EGC videos.

Blue laser imaging-bright improves the real-time detection rate of early gastric cancer: a randomized controlled study

2019 ◽  
Vol 89 (1) ◽  
pp. 47-57 ◽  
Author(s):  
Osamu Dohi ◽  
Nobuaki Yagi ◽  
Yuji Naito ◽  
Akifumi Fukui ◽  
Yasuyuki Gen ◽  
...  
Keyword(s):  
Gastric Cancer ◽  
Early Gastric Cancer ◽  
Real Time ◽  
Detection Rate ◽  
Blue Laser ◽  
Controlled Study ◽  
Laser Imaging ◽  
The Real ◽  
Randomized Controlled ◽  
Blue Laser Imaging

scholarly journals Factors Affecting the Accuracy of Endoscopic Ultrasonography in the Diagnosis of Early Gastric Cancer Invasion Depth: A Meta-analysis

2019 ◽  
Vol 2019 ◽  
pp. 1-11
Author(s):  
Ding Shi ◽  
Xiao-xia Xi
Keyword(s):  
Gastric Cancer ◽  
Early Gastric Cancer ◽  
Likelihood Ratio ◽  
Endoscopic Ultrasonography ◽  
Positive Likelihood Ratio ◽  
Negative Likelihood Ratio ◽  
Diagnostic Value ◽  
Depth Of Invasion ◽  
Invasion Depth ◽  
Sensitivity Specificity

Background. Endoscopic ultrasonography (EUS) is the first imaging modality for investigating the depth of invasion in early gastric cancer (EGC). However, there is presently no consensus on the accuracy of EUS in diagnosing the invasion depth of EGC. Aim. This study is aimed at systematically evaluating the accuracy of EUS in diagnosing the invasion depth of EGC and its affecting factors. Methods. The literatures were identified by searching PubMed, SpringerLink, Cochrane Library, Web of Science, Nature, and Karger knowledge databases. Two researchers extracted the data from the literature and reconstructed these in 2×2 tables. The Meta-DiSc software was used to evaluate the overall sensitivity, specificity, positive likelihood ratio, negative likelihood ratio, diagnostic advantage ratio, and 95% confidence interval (CI). The SROC was drawn, and the area under the curve (AUC) was calculated to evaluate the diagnostic value. Results. A total of 17 articles were selected, which included 4525 cases of lesions. The sensitivity, specificity, positive likelihood ratio, negative likelihood ratio, diagnostic dominance ratio, and 95% CI of EUS for diagnosing EGC was 0.87 (95% CI: 0.86-0.88), 0.67 (95% CI: 0.65-0.70), 2.90 (95% CI: 2.25-3.75), 0.17 (95% CI: 0.13-0.23), and 18.25 (95% CI: 12.61-26.39), respectively. The overall overstaging rate of mucosa/submucosa 1 (M/SM1) and SM by EUS was 13.31% and 32.8%, respectively, while the overall understaging rate of SM was 29.7%. The total misdiagnosis rates for EUS were as follows: 30.4% for lesions≥2 cm and 20.9% for lesions<2 cm, 27.7% for ulcerative lesions and 21.4% for nonulcerative lesions, and 22% for differentiated lesions and 26.9% for undifferentiated lesions. Conclusion. EUS has a moderate diagnostic value for the depth of invasion of EGC. The shape, size, and differentiation of lesions might be the main factors that affect the accuracy of EUS in diagnosing EGC.

scholarly journals A Deep Learning System for Recognizing Facial Expression in Real-Time

2019 ◽  
Vol 15 (2) ◽  
pp. 1-20 ◽  
Author(s):  
Yu Miao ◽  
Haiwei Dong ◽  
Jihad Mohamad Al Jaam ◽  
Abdulmotaleb El Saddik
Keyword(s):  
Deep Learning ◽  
Facial Expression ◽  
Real Time ◽  
Learning System
Sign in / Sign up

Export Citation Format

Share Document