Performance of an artificial intelligence-based diagnostic support tool for early gastric cancers: Retrospective study

被引：15

作者：

Ishioka, Mitsuaki ^{[1
]}

Osawa, Hiroyuki ^{[6
]}

Hirasawa, Toshiaki ^{[1
,9
]}

Kawachi, Hiroshi ^{[2
]}

Nakano, Kaoru ^{[2
]}

Fukushima, Noriyoshi ^{[7
]}

Sakaguchi, Mio ^{[7
]}

Tada, Tomohiro ^{[3
,4
,8
]}

Kato, Yusuke ^{[3
]}

Shibata, Junichi ^{[3
]}

Ozawa, Tsuyoshi ^{[3
]}

Tajiri, Hisao ^{[5
]}

Fujisaki, Junko ^{[1
]}

机构：

[1] Japanese Fdn Canc Res, Canc Inst Hosp, Dept Gastroenterol, Tokyo, Japan

[2] Japanese Fdn Canc Res, Canc Inst Hosp, Dept Pathol, Tokyo, Japan

[3] AI Med Serv Inc, Tokyo, Japan

[4] Univ Tokyo, Grad Sch Med, Dept Surg Oncol, Tokyo, Japan

[5] Jikei Univ, Sch Med, Tokyo, Japan

[6] Jichi Med Univ, Dept Med, Div Gastroenterol, Tochigi, Japan

[7] Jichi Med Univ, Dept Pathol, Tochigi, Japan

[8] Tada Tomohiro Inst Gastroenterol & Proctol, Saitama, Japan

[9] Japanese Fdn Canc Res, Canc Inst Hosp, Dept Gastroenterol, 3-8-31 Ariake,Koto, Tokyo 1358550, Japan

来源：

DIGESTIVE ENDOSCOPY | 2023年 / 35卷 / 04期

关键词：

artificial intelligence; diagnosis; endoscopy; gastric cancer; Helicobacter pylori; CONVOLUTIONAL NEURAL-NETWORK; MAGNIFYING ENDOSCOPY; LIMITATIONS; SURFACE; CAGA;

D O I：

10.1111/den.14455

中图分类号：

R57 [消化系及腹部疾病];

学科分类号：

摘要：

ObjectivesEndoscopists' abilities to diagnose early gastric cancers (EGCs) vary, especially between specialists and nonspecialists. We developed an artificial intelligence (AI)-based diagnostic support tool "Tango" to differentiate EGCs and compared its performance with that of endoscopists. MethodsThe diagnostic performances of Tango and endoscopists (34 specialists, 42 nonspecialists) were compared using still images of 150 neoplastic and 165 non-neoplastic lesions. Neoplastic lesions included EGCs and adenomas. The primary outcome was to show the noninferiority of Tango (based on sensitivity) over specialists. The secondary outcomes were the noninferiority of Tango (based on accuracy) over specialists and the superiority of Tango (based on sensitivity and accuracy) over nonspecialists. The lower limit of the 95% confidence interval (CI) of the difference between Tango and the specialists for sensitivity was calculated, with >-10% defined as noninferiority and >0% defined as superiority in the primary outcome. The comparable differences between Tango and the endoscopists for each performance were calculated, with >10% defined as superiority and >0% defined as noninferiority in the secondary outcomes. ResultsTango achieved superiority over the specialists based on sensitivity (84.7% vs. 65.8%, difference 18.9%, 95% CI 12.3-25.3%) and demonstrated noninferiority based on accuracy (70.8% vs. 67.4%). Tango achieved superiority over the nonspecialists based on sensitivity (84.7% vs. 51.0%) and accuracy (70.8% vs. 58.4%). ConclusionsThe AI-based diagnostic support tool for EGCs demonstrated a robust performance and may be useful to reduce misdiagnosis.

引用

页码：483 / 491

页数：9

共 21 条

[21] Differences in upper gastrointestinal neoplasm detection rates based on inspection time and esophagogastroduodenoscopy training
Yoshimizu, Shoichi
Hirasawa, Toshiaki
Horiuchi, Yusuke
Omae, Masami
Ishiyama, Akiyoshi
Yoshio, Toshiyuki
Tsuchida, Tomohiro
Fujisaki, Junko
[J]. ENDOSCOPY INTERNATIONAL OPEN, 2018, 6 (10) : E1190 - E1197

← 1 2 3 →