Enhancing deep learning classification performance of tongue lesions in imbalanced data - mosaic-based soft labeling with curriculum learning

Sep 21, 2026

Uploaded by Logic

Oral potentially malignant disorders (OPMDs) elevate the risk of tongue and other oral cavity cancers, making early detection essential to reduce cancer-specific morbidity and mortality. Deep learning has been increasingly applied to diagnostic tasks in oral imaging, yet training is hindered by class imbalance in imbalanced lesion datasets. This study evaluates an artificial intelligence technique designed to strengthen classification performance on tongue lesions despite unequal class representation.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages10

Reading Time25 min

AccessFree to download

Abstract

Background

Methods

Results

Conclusions

Introduction

Lee et al. BMCOral Health (2024) 24:161 [https://doi.org/10.1186/s12903-024-03898-3](https://doi.org/10.1186/s12903-024-03898-3) BMC Oral Health RESEARCH Open Access Enhancing deep learning classification performance of tongue lesions in imbalanced data: mosaic-based soft labeling with curriculum learning Sung‑Jae Lee 1†, Hyun Jun Oh2†, Young‑Don Son 1,3, Jong‑Hoon Kim3,4, Ik‑Jae Kwon5,7, Bongju Kim6, Jong‑Ho Lee2,6* and Hang‑Keun Kim 1,3* Abstract Background Oral potentially malignant disorders (OPMDs) are associated with an increased risk of cancer of the oral cavity including the tongue. The early detection of oral cavity cancers and OPMDs is critical for reducing cancer‑ specific morbidity and mortality. Recently, there have been studies to apply the rapidly advancing technology of deep learning for diagnosing oral cavity cancer and OPMDs. However, several challenging issues such as class imbalance must be resolved to effectively train a deep learning model for medical imaging classification tasks. The aim of this study is to evaluate a new technique of artificial intelligence to improve the classification performance in an imbal‑ anced tongue lesion dataset. Methods A total of 1,810 tongue images were used for the classification. The class‑imbalanced dataset consisted of 372 instances of cancer, 141 instances of OPMDs, and 1,297 instances of noncancerous lesions. The EfficientNet model was used as the feature extraction model for classification. Mosaic data augmentation, soft labeling, and cur‑ riculum learning (CL) were employed to improve the classification performance of the convolutional neural network. Results Utilizing a mosaic‑augmented dataset in conjunction with CL, the final model achieved an accuracy rate of 0 . 9444, surpassing conventional oversampling and weight balancing methods. The relative precision improvement rate for the minority class OPMD was 21. 2%, while the relative score improvement rate of OPMD was 4. 9% . Conclusions The present study demonstrates that the integration of mosaic‑based soft labeling and curriculum learning improves the classification performance of tongue lesions compared to previous methods, establishing a foundation for future research on effectively learning from imbalanced data. Keywords Tongue cancer, Class imbalance, Deep learning, Mosaic augmentation, Curriculum learning †Sung‑Jae Lee and Hyun Jun Oh contributed equally to this work. *Correspondence: Jong‑Ho Lee [leejongh@snu.ac.kr](leejongh@snu.ac.kr)[ ](leejongh@snu.ac.kr)Hang‑Keun Kim [dsaint31@gachon.ac.kr](dsaint31@gachon.ac.kr) Full list of author information is available at the end of the article © The Author(s) 2024. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit [http://creativecommons.org/licenses/by/4.0/](http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver ()[. The Creative Commons Public Domain Dedication waiver (](http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver ()[http://creativecom](http://creativecom)‑ [mons.org/publicdomain/zero/1.0/](mons.org/publicdomain/zero/1.0/)) applies to the data made available in this article, unless otherwise sta

FAQs

Why is early detection of tongue lesions important?

OPMDs and oral cavity cancers are associated with increased morbidity and mortality. Early detection supports timely treatment and can reduce cancer-specific outcomes.

What dataset size and class distribution were used?

The study used 1,810 tongue images with a class-imbalanced distribution: 372 cancer, 141 OPMD, and 1,297 noncancerous lesion instances.

Which techniques improved classification performance under class imbalance?

The method combined mosaic data augmentation, soft labeling, and curriculum learning with an EfficientNet-based feature extraction model.

Sep 21, 2026

Uploaded by Logic