An artificial intelligence system based on multi-modal endoscopic images for the diagnosis of gastric neoplasms (with video)
10.3760/cma.j.cn321463-20230227-00647
- VernacularTitle:基于多模态内镜图像的人工智能模型诊断胃肿瘤性病变研究（含视频）
- Author:
Xiao TAO
1
;
Lianlian WU
;
Hongliu DU
;
Zehua DONG
;
Honggang YU
Author Information
1. 武汉大学人民医院消化内科 消化系统疾病湖北省重点实验室，武汉 430060
- Keywords:
Artificial intelligence;
Gastric neoplasms;
Weak magnifying endoscopy;
Multi-modal
- From:
Chinese Journal of Digestive Endoscopy
2024;41(9):690-696
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To develop an artificial intelligence model based on multi-modal endoscopic images for identifying gastric neoplasms and to compare its diagnostic efficacy with traditional models and endoscopists.Methods:A total of 3 267 images of gastric neoplasms and non-neoplastic lesions under white light (WL) endoscopy and weak magnification (WM) endoscopy from 463 patients at the Digestive Endoscopy Center of Renmin Hospital of Wuhan University from March 2018 to December 2019 were utilized. Two single-modal models (WL model and WM model) were constructed based on WL and WM images separately. WL and WM images of corresponding lesions were combined into image pairs for creating a multi-modal (MM) characteristics integration model. A test set consisting of 696 images of 102 lesions from 97 patients from March 2020 to March 2021 was used to compare the diagnostic efficacy of the single-modal models and a multi-modal model for gastric neoplastic lesions at both the image and the lesion levels. Additionally, video clips of 80 lesions from 80 patients from January 2022 to June 2022 were employed to compare diagnostic efficacy of the WM model, the MM model and 7 endoscopists at the lesion level for gastric neoplasms.Results:In the image test set, the sensitivity and accuracy of MM model were 84.96% (576/678), and 86.89% (1 220/1 289), respectively, for diagnosing gastric neoplasms at the image level, which were superior to 63.13% (113/179) and 80.59% (353/438) of WM model ( χ2=42.81, P<0.001; χ2=10.33, P=0.001), and also better than those of WL model [70.47% (74/105), χ2=13.52, P<0.001; 67.82% (175/258), χ2=57.27, P<0.001]. The MM model showed a sensitivity of 87.50% (28/32), a specificity of 88.57% (62/70), and an accuracy of 88.24% (90/102) at the lesion level. The specificity ( χ2=22.99, P<0.001) and accuracy ( χ2=19.06, P<0.001) were significantly higher than those of WL model; however, there was no significant difference compared with those of the WM model ( P>0.05). In the video test, the sensitivity, specificity and accuracy of the MM model at the lesion level were 95.00% (19/20), 93.33% (56/60) and 93.75% (75/80). These results were significantly better than those of endoscopists, who had a sensitivity of 77.14% (108/140), a specificity of 79.29% (333/420), and an accuracy of 78.75% (441/560), with significant differences ( χ2=18.62, P<0.001; χ2=35.07, P<0.001; χ2=53.12, P<0.001), and was higher than the sensitivity of advanced endoscopists [83.33% (50/60)] with significant difference ( χ2=4.23, P=0.040). Conclusion:The artificial intelligence model based on multi-modal endoscopic images for the diagnosis of gastric neoplasms shows high efficacy in both image and video test sets, outperforming the average diagnostic performance of endoscopists in the video test.
