Real-time semantic segmentation of gastric intestinal metaplasia using a deep learning approach

다국어 초록 (Multilingual Abstract)

Background/Aims: Previous artificial intelligence (AI) models attempting to segment gastric intestinal metaplasia (GIM) areas havefailed to be deployed in real-time endoscopy due to their slow inference speeds. Here, we propose a new GIM segmentation AI modelwith inference speeds faster than 25 frames per second that maintains a high level of accuracy.
Methods: Investigators from Chulalongkorn University obtained 802 histological-proven GIM images for AI model training. Fourstrategies were proposed to improve the model accuracy. First, transfer learning was employed to the public colon datasets. Second, animage preprocessing technique contrast-limited adaptive histogram equalization was employed to produce clearer GIM areas. Third,data augmentation was applied for a more robust model. Lastly, the bilateral segmentation network model was applied to segment GIMareas in real time. The results were analyzed using different validity values.
Results: From the internal test, our AI model achieved an inference speed of 31.53 frames per second. GIM detection showed sensitivity,specificity, positive predictive, negative predictive, accuracy, and mean intersection over union in GIM segmentation values of 93%,80%, 82%, 92%, 87%, and 57%, respectively.
Conclusions: The bilateral segmentation network combined with transfer learning, contrast-limited adaptive histogram equalization,and data augmentation can provide high sensitivity and good accuracy for GIM detection and segmentation.

참고문헌 (Reference)

1 Bernal J, "WM-DOVA maps for accurate polyp highlighting in colonoscopy : validation vs. saliency maps from physicians" 43 : 99-111, 2015

2 Simonyan K, "Very deep convolutional networks for large-scale image recognition" 2015

3 Ronneberger O, "U-Net: convolutional networks for biomedical image segmentation" 234-241, 2015

4 Pittayanon R, "The learning curve of gastric intestinal metaplasia interpretation on the images obtained by probe-based confocal laser endomicroscopy" 2012 : 278045-, 2012

5 Read P, "Restoration of motion picture film" Butterworth-Heinemann 2000

6 Rodriguez-Diaz E, "Real-time artificial intelligence-based histologic classification of colorectal polyps with augmented visualization" 93 : 662-670, 2021

7 Savarino E, "Narrow-band imaging with magnifying endoscopy is accurate for detecting gastric intestinal metaplasia" 19 : 2668-2675, 2013

8 Wang C, "Localizing and identifying intestinal metaplasia based on deep learning in oesophagoscope" 1-4, 2019

9 Russell BC, "LabelMe : a database and web-based tool for image annotation" 77 : 157-173, 2008

10 Jha D, "Kvasir-SEG: a segmented polyp dataset" 451-462, 2020

11 Fox JG, "Inflammation, atrophy, and gastric cancer" 117 : 60-69, 2007

12 Yosinski J, "How transferable are features in deep neural networks?" 3320-3328, 2014

13 Zuiderveld K, "Graphics gems IV" Academic Press 474-485, 1994

14 Lecun Y, "Gradient-based learning applied to document recognition" 86 : 2278-2324, 1998

15 Li Y, "GT-Net: a deep learning network for gastric tumor diagnosis" 20-24, 2018

16 Chen LC, "Encoder-decoder with atrous separable convolution for semantic image segmentation" 833-851, 2018

17 Tan M, "EfficientNet: rethinking model scaling for convolutional neural networks" 6105-6114, 2019

18 Panteris V, "Diagnostic capabilities of high-definition white light endoscopy for the diagnosis of gastric intestinal metaplasia and correlation with histologic and clinical data" 26 : 594-601, 2014

19 Zhang L, "Diagnosis of gastric lesions through a deep convolutional neural network" 33 : 788-796, 2021

20 Huang G, "Densely connected convolutional networks" 2261-2269, 2017

21 He K, "Deep residual learning for image recognition" 770-778, 2016

22 Goncalves W, "Deep learning in gastric tissue diseases : a systematic review" 7 : e000371-, 2020

23 임지환 ; 김나영 ; 이혜승 ; 최기영 ; 조소영 ; 전일영 ; 최치언 ; 윤혁 ; 신철민 ; 박영수 ; 이동호 ; 정현채, "Correlation between Endoscopic and Histological Diagnoses of Gastric Intestinal Metaplasia" 거트앤리버 소화기연관학회협의회 7 (7): 41-45, 2013

24 Li L, "Convolutional neural network for the diagnosis of early gastric cancer based on magnifying narrow band imaging" 23 : 126-132, 2020

25 Sun X, "Colorectal polyp segmentation by U-Net with dilation convolution" 851-858, 2019

26 Dixon MF, "Classification and grading of gastritis. The updated Sydney System. International Workshop on the Histopathology of Gastritis, Houston 1994" 20 : 1161-1181, 1996

27 Banks M, "British Society of Gastroenterology guidelines on the diagnosis and management of patients at risk of gastric adenocarcinoma" 68 : 1545-1575, 2019

28 Yu C, "BiSeNet: bilateral segmentation network for real-time semantic segmentation" 334-349, 2018

29 Xu M, "Artificial intelligence in the diagnosis of gastric precancerous conditions by image-enhanced endoscopy : a multicenter, diagnostic study(with video)" 94 : 540-548, 2021

30 Mori Y, "Artificial intelligence in colonoscopy : now on the market. What’s next?" 36 : 7-11, 2021

31 Suzuki H, "Artificial intelligence for cancer detection of the upper gastrointestinal tract" 33 : 254-262, 2021

32 Hirasawa T, "Application of artificial intelligence using a convolutional neural network for detecting gastric cancer in endoscopic images" 21 : 653-660, 2018

33 Sun M, "Accurate gastric cancer segmentation in digital pathology images using deformable convolution and multi-scale embedding networks" 7 : 75530-75541, 2019

34 ASGE Technology Committee, "ASGE Technology Committee systematic review and meta-analysis assessing the ASGE PIVI thresholds for adopting real-time endoscopic assessment of the histology of diminutive colorectal polyps" 81 : 502-, 2015

35 Wu C, "A prospective randomized tandem gastroscopy pilot study of linked color imaging versus white light imaging for detection of upper gastrointestinal lesions" 36 : 2562-2567, 2021

36 Ang TL, "A multicenter randomized comparison between high-definition white light endoscopy and narrow band imaging for detection of gastric lesions" 27 : 1473-1478, 2015