Loading…
The effect of incorporating domain knowledge with deep learning in identifying benign and malignant gastric whitish lesions: A retrospective study
Background and Aim Early whitish gastric neoplasms can be easily misdiagnosed; differential diagnosis of gastric whitish lesions remains a challenge. We aim to build a deep learning (DL) model to diagnose whitish gastric neoplasms and explore the effect of adding domain knowledge in model constructi...
Saved in:
| Published in: | Journal of gastroenterology and hepatology 2024-07, Vol.39 (7), p.1343-1351 |
|---|---|
| Main Authors: | , , , , , , , , , , , , , , , , , , |
| Format: | Article |
| Language: | English |
| Subjects: | |
| Citations: | Items that this one cites |
| Online Access: | Get full text |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| cited_by | |
|---|---|
| cites | cdi_FETCH-LOGICAL-c3135-72cd84863f2ae0082033070911457e08a22e039cb28ff2b5918a23b4e1198ae53 |
| container_end_page | 1351 |
| container_issue | 7 |
| container_start_page | 1343 |
| container_title | Journal of gastroenterology and hepatology |
| container_volume | 39 |
| creator | Zeng, Xiaoquan Yang, Lang Dong, Zehua Gong, Dexin Li, Yanxia Deng, Yunchao Du, Hongliu Li, Xun Xu, Youming Luo, Chaijie Wang, Junxiao Tao, Xiao Zhang, Chenxia Zhu, Yijie Jiang, Ruiqing Yao, Liwen Wu, Lianlian Jin, Peng Yu, Honggang |
| description | Background and Aim
Early whitish gastric neoplasms can be easily misdiagnosed; differential diagnosis of gastric whitish lesions remains a challenge. We aim to build a deep learning (DL) model to diagnose whitish gastric neoplasms and explore the effect of adding domain knowledge in model construction.
Methods
We collected 4558 images from two institutions to train and test models. We first developed two sole DL models (1 and 2) using supervised and semi‐supervised algorithms. Then we selected diagnosis‐related features through literature research and developed feature‐extraction models to determine features including boundary, surface, roundness, depression, and location. Then predictions of the five feature‐extraction models and sole DL model were combined and inputted into seven machine‐learning (ML) based fitting‐diagnosis models. The optimal model was selected as ENDOANGEL‐WD (whitish‐diagnosis) and compared with endoscopists.
Results
Sole DL 2 had higher sensitivity (83.12% vs 68.67%, Bonferroni adjusted P = 0.024) than sole DL 1. Adding domain knowledge, the decision tree performed best among the seven ML models, achieving higher specificity than DL 1 (84.38% vs 72.27%, Bonferroni adjusted P |
| doi_str_mv | 10.1111/jgh.16525 |
| format | article |
| fullrecord | <record><control><sourceid>proquest_cross</sourceid><recordid>TN_cdi_proquest_miscellaneous_2932937720</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2932937720</sourcerecordid><originalsourceid>FETCH-LOGICAL-c3135-72cd84863f2ae0082033070911457e08a22e039cb28ff2b5918a23b4e1198ae53</originalsourceid><addsrcrecordid>eNp1kctu1DAUhi1ERYcpC14AWWJDF2l9nSTsqoq2oErdtGvLSY4TD4kdbKejeQ2eGA9TWCBhWfLt83ds_Qi9p-SC5na57YcLupFMvkIrKgQpaCk2r9GKVFQWNaf1KXob45YQIkgp36BTXgkqOJEr9PNxAAzGQJuwN9i61ofZB52s63HnJ20d_u78boSuB7yzacAdwIxH0MEdmHxuO3DJmv1h2YCzvcPadXjSY55ql3CvYwq2xbvBJhuHfDla7-JnfIUDpODjnMvbZ8AxLd3-DJ0YPUZ49zKu0dPNl8fru-L-4fbr9dV90XLKZVGytqtEteGGaSCkYoRzUpKaUiFLIJVmDAiv24ZVxrBG1jRv8UYApXWlQfI1-nT0zsH_WCAmNdnYwjhqB36JitU897LM4jX6-A-69Utw-XUq16xqlskyU-dHqs1figGMmoOddNgrStQhKJWDUr-DyuyHF-PSTND9Jf8kk4HLI7CzI-z_b1Lfbu-Oyl-H5Z2p</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>3078923297</pqid></control><display><type>article</type><title>The effect of incorporating domain knowledge with deep learning in identifying benign and malignant gastric whitish lesions: A retrospective study</title><source>Wiley</source><creator>Zeng, Xiaoquan ; Yang, Lang ; Dong, Zehua ; Gong, Dexin ; Li, Yanxia ; Deng, Yunchao ; Du, Hongliu ; Li, Xun ; Xu, Youming ; Luo, Chaijie ; Wang, Junxiao ; Tao, Xiao ; Zhang, Chenxia ; Zhu, Yijie ; Jiang, Ruiqing ; Yao, Liwen ; Wu, Lianlian ; Jin, Peng ; Yu, Honggang</creator><creatorcontrib>Zeng, Xiaoquan ; Yang, Lang ; Dong, Zehua ; Gong, Dexin ; Li, Yanxia ; Deng, Yunchao ; Du, Hongliu ; Li, Xun ; Xu, Youming ; Luo, Chaijie ; Wang, Junxiao ; Tao, Xiao ; Zhang, Chenxia ; Zhu, Yijie ; Jiang, Ruiqing ; Yao, Liwen ; Wu, Lianlian ; Jin, Peng ; Yu, Honggang</creatorcontrib><description>Background and Aim
Early whitish gastric neoplasms can be easily misdiagnosed; differential diagnosis of gastric whitish lesions remains a challenge. We aim to build a deep learning (DL) model to diagnose whitish gastric neoplasms and explore the effect of adding domain knowledge in model construction.
Methods
We collected 4558 images from two institutions to train and test models. We first developed two sole DL models (1 and 2) using supervised and semi‐supervised algorithms. Then we selected diagnosis‐related features through literature research and developed feature‐extraction models to determine features including boundary, surface, roundness, depression, and location. Then predictions of the five feature‐extraction models and sole DL model were combined and inputted into seven machine‐learning (ML) based fitting‐diagnosis models. The optimal model was selected as ENDOANGEL‐WD (whitish‐diagnosis) and compared with endoscopists.
Results
Sole DL 2 had higher sensitivity (83.12% vs 68.67%, Bonferroni adjusted P = 0.024) than sole DL 1. Adding domain knowledge, the decision tree performed best among the seven ML models, achieving higher specificity than DL 1 (84.38% vs 72.27%, Bonferroni adjusted P < 0.05) and higher accuracy than DL 2 (80.47%, Bonferroni adjusted P < 0.001) and was selected as ENDOANGEL‐WD. ENDOANGEL‐WD showed better accuracy compared with 10 endoscopists (75.70%, P < 0.001).
Conclusions
We developed a novel system ENDOANGEL‐WD combining domain knowledge and traditional DL to detect gastric whitish neoplasms. Adding domain knowledge improved the performance of traditional DL, which provided a novel solution for establishing diagnostic models for other rare diseases potentially.</description><identifier>ISSN: 0815-9319</identifier><identifier>ISSN: 1440-1746</identifier><identifier>EISSN: 1440-1746</identifier><identifier>DOI: 10.1111/jgh.16525</identifier><identifier>PMID: 38414305</identifier><language>eng</language><publisher>Australia: Wiley Subscription Services, Inc</publisher><subject>Deep learning ; Differential diagnosis ; domain knowledge ; feature extraction ; Tumors ; whitish gastric lesions</subject><ispartof>Journal of gastroenterology and hepatology, 2024-07, Vol.39 (7), p.1343-1351</ispartof><rights>2024 Journal of Gastroenterology and Hepatology Foundation and John Wiley & Sons Australia, Ltd.</rights><rights>2024 Journal of Gastroenterology and Hepatology Foundation and John Wiley & Sons Australia, Ltd</rights><lds50>peer_reviewed</lds50><woscitedreferencessubscribed>false</woscitedreferencessubscribed><cites>FETCH-LOGICAL-c3135-72cd84863f2ae0082033070911457e08a22e039cb28ff2b5918a23b4e1198ae53</cites><orcidid>0000-0002-1882-7060 ; 0000-0002-5234-7394</orcidid></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>314,780,784,27924,27925</link.rule.ids><backlink>$$Uhttps://www.ncbi.nlm.nih.gov/pubmed/38414305$$D View this record in MEDLINE/PubMed$$Hfree_for_read</backlink></links><search><creatorcontrib>Zeng, Xiaoquan</creatorcontrib><creatorcontrib>Yang, Lang</creatorcontrib><creatorcontrib>Dong, Zehua</creatorcontrib><creatorcontrib>Gong, Dexin</creatorcontrib><creatorcontrib>Li, Yanxia</creatorcontrib><creatorcontrib>Deng, Yunchao</creatorcontrib><creatorcontrib>Du, Hongliu</creatorcontrib><creatorcontrib>Li, Xun</creatorcontrib><creatorcontrib>Xu, Youming</creatorcontrib><creatorcontrib>Luo, Chaijie</creatorcontrib><creatorcontrib>Wang, Junxiao</creatorcontrib><creatorcontrib>Tao, Xiao</creatorcontrib><creatorcontrib>Zhang, Chenxia</creatorcontrib><creatorcontrib>Zhu, Yijie</creatorcontrib><creatorcontrib>Jiang, Ruiqing</creatorcontrib><creatorcontrib>Yao, Liwen</creatorcontrib><creatorcontrib>Wu, Lianlian</creatorcontrib><creatorcontrib>Jin, Peng</creatorcontrib><creatorcontrib>Yu, Honggang</creatorcontrib><title>The effect of incorporating domain knowledge with deep learning in identifying benign and malignant gastric whitish lesions: A retrospective study</title><title>Journal of gastroenterology and hepatology</title><addtitle>J Gastroenterol Hepatol</addtitle><description>Background and Aim
Early whitish gastric neoplasms can be easily misdiagnosed; differential diagnosis of gastric whitish lesions remains a challenge. We aim to build a deep learning (DL) model to diagnose whitish gastric neoplasms and explore the effect of adding domain knowledge in model construction.
Methods
We collected 4558 images from two institutions to train and test models. We first developed two sole DL models (1 and 2) using supervised and semi‐supervised algorithms. Then we selected diagnosis‐related features through literature research and developed feature‐extraction models to determine features including boundary, surface, roundness, depression, and location. Then predictions of the five feature‐extraction models and sole DL model were combined and inputted into seven machine‐learning (ML) based fitting‐diagnosis models. The optimal model was selected as ENDOANGEL‐WD (whitish‐diagnosis) and compared with endoscopists.
Results
Sole DL 2 had higher sensitivity (83.12% vs 68.67%, Bonferroni adjusted P = 0.024) than sole DL 1. Adding domain knowledge, the decision tree performed best among the seven ML models, achieving higher specificity than DL 1 (84.38% vs 72.27%, Bonferroni adjusted P < 0.05) and higher accuracy than DL 2 (80.47%, Bonferroni adjusted P < 0.001) and was selected as ENDOANGEL‐WD. ENDOANGEL‐WD showed better accuracy compared with 10 endoscopists (75.70%, P < 0.001).
Conclusions
We developed a novel system ENDOANGEL‐WD combining domain knowledge and traditional DL to detect gastric whitish neoplasms. Adding domain knowledge improved the performance of traditional DL, which provided a novel solution for establishing diagnostic models for other rare diseases potentially.</description><subject>Deep learning</subject><subject>Differential diagnosis</subject><subject>domain knowledge</subject><subject>feature extraction</subject><subject>Tumors</subject><subject>whitish gastric lesions</subject><issn>0815-9319</issn><issn>1440-1746</issn><issn>1440-1746</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2024</creationdate><recordtype>article</recordtype><recordid>eNp1kctu1DAUhi1ERYcpC14AWWJDF2l9nSTsqoq2oErdtGvLSY4TD4kdbKejeQ2eGA9TWCBhWfLt83ds_Qi9p-SC5na57YcLupFMvkIrKgQpaCk2r9GKVFQWNaf1KXob45YQIkgp36BTXgkqOJEr9PNxAAzGQJuwN9i61ofZB52s63HnJ20d_u78boSuB7yzacAdwIxH0MEdmHxuO3DJmv1h2YCzvcPadXjSY55ql3CvYwq2xbvBJhuHfDla7-JnfIUDpODjnMvbZ8AxLd3-DJ0YPUZ49zKu0dPNl8fru-L-4fbr9dV90XLKZVGytqtEteGGaSCkYoRzUpKaUiFLIJVmDAiv24ZVxrBG1jRv8UYApXWlQfI1-nT0zsH_WCAmNdnYwjhqB36JitU897LM4jX6-A-69Utw-XUq16xqlskyU-dHqs1figGMmoOddNgrStQhKJWDUr-DyuyHF-PSTND9Jf8kk4HLI7CzI-z_b1Lfbu-Oyl-H5Z2p</recordid><startdate>202407</startdate><enddate>202407</enddate><creator>Zeng, Xiaoquan</creator><creator>Yang, Lang</creator><creator>Dong, Zehua</creator><creator>Gong, Dexin</creator><creator>Li, Yanxia</creator><creator>Deng, Yunchao</creator><creator>Du, Hongliu</creator><creator>Li, Xun</creator><creator>Xu, Youming</creator><creator>Luo, Chaijie</creator><creator>Wang, Junxiao</creator><creator>Tao, Xiao</creator><creator>Zhang, Chenxia</creator><creator>Zhu, Yijie</creator><creator>Jiang, Ruiqing</creator><creator>Yao, Liwen</creator><creator>Wu, Lianlian</creator><creator>Jin, Peng</creator><creator>Yu, Honggang</creator><general>Wiley Subscription Services, Inc</general><scope>NPM</scope><scope>AAYXX</scope><scope>CITATION</scope><scope>7T5</scope><scope>7U9</scope><scope>H94</scope><scope>7X8</scope><orcidid>https://orcid.org/0000-0002-1882-7060</orcidid><orcidid>https://orcid.org/0000-0002-5234-7394</orcidid></search><sort><creationdate>202407</creationdate><title>The effect of incorporating domain knowledge with deep learning in identifying benign and malignant gastric whitish lesions: A retrospective study</title><author>Zeng, Xiaoquan ; Yang, Lang ; Dong, Zehua ; Gong, Dexin ; Li, Yanxia ; Deng, Yunchao ; Du, Hongliu ; Li, Xun ; Xu, Youming ; Luo, Chaijie ; Wang, Junxiao ; Tao, Xiao ; Zhang, Chenxia ; Zhu, Yijie ; Jiang, Ruiqing ; Yao, Liwen ; Wu, Lianlian ; Jin, Peng ; Yu, Honggang</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-c3135-72cd84863f2ae0082033070911457e08a22e039cb28ff2b5918a23b4e1198ae53</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2024</creationdate><topic>Deep learning</topic><topic>Differential diagnosis</topic><topic>domain knowledge</topic><topic>feature extraction</topic><topic>Tumors</topic><topic>whitish gastric lesions</topic><toplevel>peer_reviewed</toplevel><toplevel>online_resources</toplevel><creatorcontrib>Zeng, Xiaoquan</creatorcontrib><creatorcontrib>Yang, Lang</creatorcontrib><creatorcontrib>Dong, Zehua</creatorcontrib><creatorcontrib>Gong, Dexin</creatorcontrib><creatorcontrib>Li, Yanxia</creatorcontrib><creatorcontrib>Deng, Yunchao</creatorcontrib><creatorcontrib>Du, Hongliu</creatorcontrib><creatorcontrib>Li, Xun</creatorcontrib><creatorcontrib>Xu, Youming</creatorcontrib><creatorcontrib>Luo, Chaijie</creatorcontrib><creatorcontrib>Wang, Junxiao</creatorcontrib><creatorcontrib>Tao, Xiao</creatorcontrib><creatorcontrib>Zhang, Chenxia</creatorcontrib><creatorcontrib>Zhu, Yijie</creatorcontrib><creatorcontrib>Jiang, Ruiqing</creatorcontrib><creatorcontrib>Yao, Liwen</creatorcontrib><creatorcontrib>Wu, Lianlian</creatorcontrib><creatorcontrib>Jin, Peng</creatorcontrib><creatorcontrib>Yu, Honggang</creatorcontrib><collection>PubMed</collection><collection>CrossRef</collection><collection>Immunology Abstracts</collection><collection>Virology and AIDS Abstracts</collection><collection>AIDS and Cancer Research Abstracts</collection><collection>MEDLINE - Academic</collection><jtitle>Journal of gastroenterology and hepatology</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Zeng, Xiaoquan</au><au>Yang, Lang</au><au>Dong, Zehua</au><au>Gong, Dexin</au><au>Li, Yanxia</au><au>Deng, Yunchao</au><au>Du, Hongliu</au><au>Li, Xun</au><au>Xu, Youming</au><au>Luo, Chaijie</au><au>Wang, Junxiao</au><au>Tao, Xiao</au><au>Zhang, Chenxia</au><au>Zhu, Yijie</au><au>Jiang, Ruiqing</au><au>Yao, Liwen</au><au>Wu, Lianlian</au><au>Jin, Peng</au><au>Yu, Honggang</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>The effect of incorporating domain knowledge with deep learning in identifying benign and malignant gastric whitish lesions: A retrospective study</atitle><jtitle>Journal of gastroenterology and hepatology</jtitle><addtitle>J Gastroenterol Hepatol</addtitle><date>2024-07</date><risdate>2024</risdate><volume>39</volume><issue>7</issue><spage>1343</spage><epage>1351</epage><pages>1343-1351</pages><issn>0815-9319</issn><issn>1440-1746</issn><eissn>1440-1746</eissn><abstract>Background and Aim
Early whitish gastric neoplasms can be easily misdiagnosed; differential diagnosis of gastric whitish lesions remains a challenge. We aim to build a deep learning (DL) model to diagnose whitish gastric neoplasms and explore the effect of adding domain knowledge in model construction.
Methods
We collected 4558 images from two institutions to train and test models. We first developed two sole DL models (1 and 2) using supervised and semi‐supervised algorithms. Then we selected diagnosis‐related features through literature research and developed feature‐extraction models to determine features including boundary, surface, roundness, depression, and location. Then predictions of the five feature‐extraction models and sole DL model were combined and inputted into seven machine‐learning (ML) based fitting‐diagnosis models. The optimal model was selected as ENDOANGEL‐WD (whitish‐diagnosis) and compared with endoscopists.
Results
Sole DL 2 had higher sensitivity (83.12% vs 68.67%, Bonferroni adjusted P = 0.024) than sole DL 1. Adding domain knowledge, the decision tree performed best among the seven ML models, achieving higher specificity than DL 1 (84.38% vs 72.27%, Bonferroni adjusted P < 0.05) and higher accuracy than DL 2 (80.47%, Bonferroni adjusted P < 0.001) and was selected as ENDOANGEL‐WD. ENDOANGEL‐WD showed better accuracy compared with 10 endoscopists (75.70%, P < 0.001).
Conclusions
We developed a novel system ENDOANGEL‐WD combining domain knowledge and traditional DL to detect gastric whitish neoplasms. Adding domain knowledge improved the performance of traditional DL, which provided a novel solution for establishing diagnostic models for other rare diseases potentially.</abstract><cop>Australia</cop><pub>Wiley Subscription Services, Inc</pub><pmid>38414305</pmid><doi>10.1111/jgh.16525</doi><tpages>9</tpages><orcidid>https://orcid.org/0000-0002-1882-7060</orcidid><orcidid>https://orcid.org/0000-0002-5234-7394</orcidid></addata></record> |
| fulltext | fulltext |
| identifier | ISSN: 0815-9319 |
| ispartof | Journal of gastroenterology and hepatology, 2024-07, Vol.39 (7), p.1343-1351 |
| issn | 0815-9319 1440-1746 1440-1746 |
| language | eng |
| recordid | cdi_proquest_miscellaneous_2932937720 |
| source | Wiley |
| subjects | Deep learning Differential diagnosis domain knowledge feature extraction Tumors whitish gastric lesions |
| title | The effect of incorporating domain knowledge with deep learning in identifying benign and malignant gastric whitish lesions: A retrospective study |
| url | http://sfxeu10.hosted.exlibrisgroup.com/loughborough?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-07T14%3A46%3A23IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest_cross&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=The%20effect%20of%20incorporating%20domain%20knowledge%20with%20deep%20learning%20in%20identifying%20benign%20and%20malignant%20gastric%20whitish%20lesions:%20A%20retrospective%20study&rft.jtitle=Journal%20of%20gastroenterology%20and%20hepatology&rft.au=Zeng,%20Xiaoquan&rft.date=2024-07&rft.volume=39&rft.issue=7&rft.spage=1343&rft.epage=1351&rft.pages=1343-1351&rft.issn=0815-9319&rft.eissn=1440-1746&rft_id=info:doi/10.1111/jgh.16525&rft_dat=%3Cproquest_cross%3E2932937720%3C/proquest_cross%3E%3Cgrp_id%3Ecdi_FETCH-LOGICAL-c3135-72cd84863f2ae0082033070911457e08a22e039cb28ff2b5918a23b4e1198ae53%3C/grp_id%3E%3Coa%3E%3C/oa%3E%3Curl%3E%3C/url%3E&rft_id=info:oai/&rft_pqid=3078923297&rft_id=info:pmid/38414305&rfr_iscdi=true |