Detection of Architectural Dysplastic Features from Histopathological Imagery of Oral Mucosa Using Neural Networks

Oral cancer is a serious illness, but it is potentially curable if early detection can be achieved successfully. Oral epithelial dysplasia (OED), which is a precursor to oral squamous cell carcinoma (OSCC), can provide abnormal characteristics to diagnose the risk of developing oral cancer. This pap...

Full description

Saved in:
Bibliographic Details
Main Authors: Watchanan Chantapakul, Sirikanlaya Vetchaporn, Sansanee Auephanwiriyakul, Nipon Theera-Umpon, Ritipong Wongkhuenkaew, Uklid Yeesarapat, Nutchapon Chamusri, Mansuang Wongsapai
Format: Article
Language:English
Published: MDPI AG 2025-02-01
Series:Bioengineering
Subjects:
Online Access:https://www.mdpi.com/2306-5354/12/3/216
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850203986108350464
author Watchanan Chantapakul
Sirikanlaya Vetchaporn
Sansanee Auephanwiriyakul
Nipon Theera-Umpon
Ritipong Wongkhuenkaew
Uklid Yeesarapat
Nutchapon Chamusri
Mansuang Wongsapai
author_facet Watchanan Chantapakul
Sirikanlaya Vetchaporn
Sansanee Auephanwiriyakul
Nipon Theera-Umpon
Ritipong Wongkhuenkaew
Uklid Yeesarapat
Nutchapon Chamusri
Mansuang Wongsapai
author_sort Watchanan Chantapakul
collection DOAJ
description Oral cancer is a serious illness, but it is potentially curable if early detection can be achieved successfully. Oral epithelial dysplasia (OED), which is a precursor to oral squamous cell carcinoma (OSCC), can provide abnormal characteristics to diagnose the risk of developing oral cancer. This paper proposes a neural network architecture for detecting dysplastic features of epithelial architecture, including irregular epithelial stratification and bulbous rete ridges. The different combinations of atrous convolution, batch normalization, global pooling, and dropout are discussed regarding their effects, along with an ablation study. A signature library containing image patches was constructed and utilized to train the models. The best-performing model in the validation set attained an average accuracy of 97.52%. The results of the blind test from the receiver operating characteristic (ROC) curves show that the best model reached the best probability of detection, 0.8571, for irregular epithelial stratifications and 0.8462 for the bulbous rete ridges.
format Article
id doaj-art-7d010deae2b94adc8ebcfd88ff9aa5b1
institution OA Journals
issn 2306-5354
language English
publishDate 2025-02-01
publisher MDPI AG
record_format Article
series Bioengineering
spelling doaj-art-7d010deae2b94adc8ebcfd88ff9aa5b12025-08-20T02:11:22ZengMDPI AGBioengineering2306-53542025-02-0112321610.3390/bioengineering12030216Detection of Architectural Dysplastic Features from Histopathological Imagery of Oral Mucosa Using Neural NetworksWatchanan Chantapakul0Sirikanlaya Vetchaporn1Sansanee Auephanwiriyakul2Nipon Theera-Umpon3Ritipong Wongkhuenkaew4Uklid Yeesarapat5Nutchapon Chamusri6Mansuang Wongsapai7Biomedical Engineering Institute, and the Biomedical Engineering and Innovation Research Center, Chiang Mai University, Chiang Mai 50200, ThailandIntercountry Centre for Oral Health, Department of Health, Ministry of Public Health, Chiang Mai 50000, ThailandBiomedical Engineering Institute, and the Biomedical Engineering and Innovation Research Center, Chiang Mai University, Chiang Mai 50200, ThailandBiomedical Engineering Institute, and the Biomedical Engineering and Innovation Research Center, Chiang Mai University, Chiang Mai 50200, ThailandBiomedical Engineering Institute, and the Biomedical Engineering and Innovation Research Center, Chiang Mai University, Chiang Mai 50200, ThailandDepartment of Computer Engineering, Faculty of Engineering, Chiang Mai University, Chiang Mai 50200, ThailandDepartment of Oral Biology and Diagnostic Sciences, Faculty of Dentistry, Chiang Mai University, Chiang Mai 50200, ThailandIntercountry Centre for Oral Health, Department of Health, Ministry of Public Health, Chiang Mai 50000, ThailandOral cancer is a serious illness, but it is potentially curable if early detection can be achieved successfully. Oral epithelial dysplasia (OED), which is a precursor to oral squamous cell carcinoma (OSCC), can provide abnormal characteristics to diagnose the risk of developing oral cancer. This paper proposes a neural network architecture for detecting dysplastic features of epithelial architecture, including irregular epithelial stratification and bulbous rete ridges. The different combinations of atrous convolution, batch normalization, global pooling, and dropout are discussed regarding their effects, along with an ablation study. A signature library containing image patches was constructed and utilized to train the models. The best-performing model in the validation set attained an average accuracy of 97.52%. The results of the blind test from the receiver operating characteristic (ROC) curves show that the best model reached the best probability of detection, 0.8571, for irregular epithelial stratifications and 0.8462 for the bulbous rete ridges.https://www.mdpi.com/2306-5354/12/3/216bulbous rete ridgesconvolutional neural networkshistopathological imageirregular epithelial stratificationsoral epithelial dysplasia
spellingShingle Watchanan Chantapakul
Sirikanlaya Vetchaporn
Sansanee Auephanwiriyakul
Nipon Theera-Umpon
Ritipong Wongkhuenkaew
Uklid Yeesarapat
Nutchapon Chamusri
Mansuang Wongsapai
Detection of Architectural Dysplastic Features from Histopathological Imagery of Oral Mucosa Using Neural Networks
Bioengineering
bulbous rete ridges
convolutional neural networks
histopathological image
irregular epithelial stratifications
oral epithelial dysplasia
title Detection of Architectural Dysplastic Features from Histopathological Imagery of Oral Mucosa Using Neural Networks
title_full Detection of Architectural Dysplastic Features from Histopathological Imagery of Oral Mucosa Using Neural Networks
title_fullStr Detection of Architectural Dysplastic Features from Histopathological Imagery of Oral Mucosa Using Neural Networks
title_full_unstemmed Detection of Architectural Dysplastic Features from Histopathological Imagery of Oral Mucosa Using Neural Networks
title_short Detection of Architectural Dysplastic Features from Histopathological Imagery of Oral Mucosa Using Neural Networks
title_sort detection of architectural dysplastic features from histopathological imagery of oral mucosa using neural networks
topic bulbous rete ridges
convolutional neural networks
histopathological image
irregular epithelial stratifications
oral epithelial dysplasia
url https://www.mdpi.com/2306-5354/12/3/216
work_keys_str_mv AT watchananchantapakul detectionofarchitecturaldysplasticfeaturesfromhistopathologicalimageryoforalmucosausingneuralnetworks
AT sirikanlayavetchaporn detectionofarchitecturaldysplasticfeaturesfromhistopathologicalimageryoforalmucosausingneuralnetworks
AT sansaneeauephanwiriyakul detectionofarchitecturaldysplasticfeaturesfromhistopathologicalimageryoforalmucosausingneuralnetworks
AT nipontheeraumpon detectionofarchitecturaldysplasticfeaturesfromhistopathologicalimageryoforalmucosausingneuralnetworks
AT ritipongwongkhuenkaew detectionofarchitecturaldysplasticfeaturesfromhistopathologicalimageryoforalmucosausingneuralnetworks
AT uklidyeesarapat detectionofarchitecturaldysplasticfeaturesfromhistopathologicalimageryoforalmucosausingneuralnetworks
AT nutchaponchamusri detectionofarchitecturaldysplasticfeaturesfromhistopathologicalimageryoforalmucosausingneuralnetworks
AT mansuangwongsapai detectionofarchitecturaldysplasticfeaturesfromhistopathologicalimageryoforalmucosausingneuralnetworks