基于ViT-B深度学习模型的口腔良恶性病变图像分类研究

doi:10.13701/j.cnki.kqyxyj.2025.01.004

口腔医学研究 ›› 2025, Vol. 41 ›› Issue (1): 16-20.DOI: 10.13701/j.cnki.kqyxyj.2025.01.004

基于ViT-B深度学习模型的口腔良恶性病变图像分类研究

崔宇琛¹, 谢元栋², 吴聿淼¹, 牛凌霄³, 常路广达¹, 朱宪春^1*

1.吉林大学口腔医院正畸科吉林长春 130021;
2.吉林大学牙发育与颌骨重塑吉林省重点实验室吉林长春 130021;
3.吉林大学口腔医院颌面外科吉林长春 130021

收稿日期:2024-08-13 出版日期:2025-01-28 发布日期:2025-01-24
通讯作者: * 朱宪春,E-mail:zhuxc@jlu.edu.cn
作者简介:崔宇琛(1999~ ),女,山西人,硕士在读,研究方向:口腔正畸治疗的生物力学机制及临床优化应用。
基金资助:
吉林省科技厅自然科学基金项目(编号:YDZJ202201ZYTS057)

Research on Classification of Benign and Malignant Oral Lesions Using ViT-B Deep Learning Model

CUI Yuchen¹, XIE Yuandong², WU Yumiao¹, NIU Lingxiao³, CHANG Luguangda¹, ZHU Xianchun^1*

1. Department of Orthodontics, Hospital of Stomatology, Jilin University, Changchun 130021, China;
2. Jilin Provincial Key Laboratory of Tooth Development and Bone Remodeling, Changchun 130021, China;
3. Department of Oral and Maxillofacial Surgery, School and Hospital of Stomatology, Jilin University, Changchun 130021, China

Received:2024-08-13 Online:2025-01-28 Published:2025-01-24

摘要/Abstract

摘要： 目的:基于深度学习算法,对ViT-B模型检测口腔良性和恶性病变图像的性能进行分析,旨在为临床医生早期发现和准确诊断口腔癌提供有效工具。方法:使用包含口腔良性和恶性病变图像的公共数据集,对数据进行预处理和数据增强,按7∶2∶1的比例将数据随机划分为训练集、验证集和测试集。选取ViT-B、VGG16、ResNet101、DenseNet121和EfficientNetV2 5种深度学习模型,对模型进行训练和性能比较。通过外部数据对ViT-B模型的泛化能力进行评估,并基于注意力权重的可视化方法对ViT-B模型进行分析。结果:ViT-B在5种模型中分类性能最佳,受试者工作特征曲线下面积为0.9715,准确率为91.00%。该模型可以有效区分口腔良性和恶性病变图像,具有较强的泛化能力和临床实用性。结论:ViT-B模型在口腔良性和恶性病变图像识别中表现良好,可以为口腔癌的早期发现和准确诊断提供支持。

关键词: 口腔癌, 口腔病变, 深度学习, ViT-B

Abstract: Objective: To analyze the performance of ViT-B model in detecting oral benign and malignant lesions based on deep learning algorithms. Methods: A public dataset containing images of oral benign and malignant lesions was used, with preprocessing and data augmentation applied. The data was randomly divided into training, validation, and test sets in a 7∶2∶1 ratio. Five deep learning models, including ViT-B, VGG16, ResNet101, DenseNet121, and EfficientNetV2, were selected for training and evaluation. The generalization ability of the ViT-B model was evaluated using external data, and the model was analyzed based on the visualization of attention weights. Results: The ViT-B model demonstrated the best performance among five models, with an area under the receiver operating characteristic curve (AUC) of 0.9715 and an accuracy of 91.00%. The model effectively distinguished between images of oral benign and malignant lesions, demonstrating strong generalization ability and clinical applicability. Conclusion: The ViT-B model performs well in the recognition of oral benign and malignant lesions, supporting the early detection and accurate diagnosis of oral cancer.

Key words: oral cancer, oral lesions, deep learning, ViT-B

崔宇琛, 谢元栋, 吴聿淼, 牛凌霄, 常路广达, 朱宪春. 基于ViT-B深度学习模型的口腔良恶性病变图像分类研究[J]. 口腔医学研究, 2025, 41(1): 16-20.

CUI Yuchen, XIE Yuandong, WU Yumiao, NIU Lingxiao, CHANG Luguangda, ZHU Xianchun. Research on Classification of Benign and Malignant Oral Lesions Using ViT-B Deep Learning Model[J]. Journal of Oral Science Research, 2025, 41(1): 16-20.

参考文献

[1] Sung H, Ferlay J, Siegel RL, et al. Global Cancer Statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries [J]. CA Cancer J Clin, 2021, 71(3): 209-249.
[2] Petersen PE. Oral cancer prevention and control-The approach of the World Health Organization [J]. Oral Oncol, 2009, 45(4-5): 454-460.
[3] McCullough MJ, Prasad G, Farah CS. Oral mucosal malignancy and potentially malignant lesions: an update on the epidemiology, risk factors, diagnosis and management [J]. Aust Dent J, 2010, 55 Suppl 1: 61-65.
[4] Speight PM, Epstein J, Kujan O, et al. Screening for oral cancera perspective from the Global Oral Cancer Forum [J]. Oral Surg Oral Med Oral Pathol Oral Radiol, 2017, 123(6): 680-687.
[5] Pandey M, Choudhury H, Ying JNS, et al. Mucoadhesive nanocarriers as a promising strategy to enhance intracellular delivery against oral cavity carcinoma [J]. Pharmaceutics, 2022, 14(4): 795.
[6] Litjens G, Kooi T, Bejnordi BE, et al. A survey on deep learning in medical image analysis [J]. Med Image Anal, 2017, 42: 60-88.
[7] Afify HM, Mohammed KK, Ella Hassanien A. Novel prediction model on OSCC histopathological images via deep transfer learning combined with Grad-CAM interpretation [J]. Biomed Signal Process Control, 2023, 83: 104704.
[8] Das N, Hussain E, Mahanta LB. Automated classification of cells into multiple classes in epithelial tissue of oral squamous cell carcinoma using transfer learning and convolutional neural network [J]. Neural Netw, 2020, 128: 47-60.
[9] Warin K, Limprasert W, Suebnukarn S, et al. Performance of deep convolutional neural network for classification and detection of oral potentially malignant disorders in photographic images [J]. Int J Oral Maxillofac Surg, 2022, 51(5): 699-704.
[10] Al-hammuri K, Gebali F, Kanan A, et al. Vision transformer architecture and applications in digital health: a tutorial and survey [J]. Vis Comput Ind Biomed Art, 2023, 6(1): 14.
[11] Chandrashekar HS, Geetha Kiran A, Murali S, et al. Oral Images Dataset [DB/OL]. Mendeley Data, 2021.
[12] Dosovitskiy A, Beyer L, Kolesnikov A, et al. An image is worth 16×16 words: Transformers for image recognition at scale [J]. arXiv, 2021.
[13] Simonyan K, Zisserman A. Very deep convolutional networks for large-scale image recognition [J]. arXiv, 2015.
[14] He K, Zhang X, Ren S, et al. Deep residual learning for image recognition [C]. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2016: 770-778.
[15] Huang G, Liu Z, Van Der Maaten L, et al. Densely connected convolutional networks [C]. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2017: 2261-2269.
[16] Tan M, Le QV. EfficientNetV2: Smaller models and faster training [J]. arXiv, 2021.
[17] Shamshad F, Khan S, Zamir SW, et al. Transformers in medical imaging: A survey [J]. Med Image Anal, 2023, 88: 102802.
[18] Qiu Z, Xu J, Peng J, et al. Cross-channel dynamic spatial-spectral fusion transformer for hyperspectral image classification [J]. IEEE Transactions on Geoscience and Remote Sensing, 2023, 61: 1-12.
[19] Marzouk R, Alabdulkreem E, Dhahbi S, et al. Deep transfer learning driven oral cancer detection and classification model [J]. Computers, Materials & Continua, 2022, 73(2): 3905-3920.

基于ViT-B深度学习模型的口腔良恶性病变图像分类研究

Research on Classification of Benign and Malignant Oral Lesions Using ViT-B Deep Learning Model

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 15

编辑推荐

Metrics

[1]	杨青默, 魏攀, 华红. 智能化时代下的口腔黏膜病学——人工智能应用研究的现状与展望[J]. 口腔医学研究, 2025, 41(1): 1-6.
[2]	帕热克江·帕塔尔, 李晨曦, 克热木·阿巴司, 胡露露, 方昌, 龚忠诚. 预防性经皮胃造瘘对口腔癌切除并同期修复重建术后术区感染影响的研究[J]. 口腔医学研究, 2024, 40(1): 29-34.
[3]	王晨辰, 许驭. 铜死亡应用于口腔癌诊断治疗的研究进展[J]. 口腔医学研究, 2023, 39(9): 771-774.
[4]	温佳欢, 傅裕杰, 任根强, 陈宇飞, 张旗. 基于深度学习的点云补全网络修复上颌磨牙钙化根管的研究[J]. 口腔医学研究, 2023, 39(5): 455-459.
[5]	杨可, 孙雅楠, 胡雅英, 吕宜楠, 郑小风, 李易玮, 张佳莉. cyclin D1表达模式深度学习识别和预测模型对HPV阴性口腔鳞状细胞癌和口咽鳞状细胞癌患者预后的预测作用[J]. 口腔医学研究, 2023, 39(4): 308-315.
[6]	赖丹琳, 许亮, 倪涧钊, 朱小峰, 黄晓红. 不同层数的Mask-RCNN模型自动检测成釉细胞瘤效能的比较[J]. 口腔医学研究, 2023, 39(12): 1092-1096.
[7]	吴天赐, 郁佳鑫, 黄晓峰, 陈盛, 王育新, 蒲玉梅. 深度学习图片分类模型ResNet-18用于判定口腔鳞状细胞癌浸润方式的初步研究[J]. 口腔医学研究, 2023, 39(10): 917-922.
[8]	朱红标, 刘强冬, 曾子强, 娄伟明, 戴芳, 吴婧婷, 邓恬, 邓立彬, 宋莉. 基于口内数码照图像深度学习的牙周病早期筛查研究[J]. 口腔医学研究, 2022, 38(11): 1092-1095.
[9]	帕克扎提·色依提, 王铁梅, 徐子能, 白海龙, 丁鹏, 刘澍, 滕跃辉, 冯英连, 王蓉, 单珊, 钟双泽. 基于深度学习在曲面体层图像中人工智能辅助诊断系统初步研究[J]. 口腔医学研究, 2021, 37(9): 845-849.
[10]	范腾飞, 任振虎, 杨雯君, 张胜, 胡镜宙, 吴汉江, 张陈平. 胸背动脉穿支皮瓣修复口腔癌复发后软组织缺损的疗效分析[J]. 口腔医学研究, 2021, 37(11): 1036-1038.
[11]	王蓉, 王铁梅. 人工智能在口腔颌面医学影像中的研究探讨[J]. 口腔医学研究, 2020, 36(7): 619-621.
[12]	韩生伟, 韩伟. 人工智能技术在口腔医学领域的应用进展[J]. 口腔医学研究, 2020, 36(6): 519-522.
[13]	宋勇, 马婷婷, 王俊林, 刘彦杰, 陈裕聪, 何浒杰. 带蒂皮瓣法构建裸鼠低氧口腔癌模型[J]. 口腔医学研究, 2019, 35(12): 1125-1128.
[14]	孙维克, 王培源, 刘伟, 王霞. Beclin1与凋亡相关蛋白在口腔鳞癌中的表达及意义[J]. 口腔医学研究, 2019, 35(10): 979-983.
[15]	胡慧珍,王世琴,孙贵军,金佳佳,李志勇,何祥一. 口腔鳞状细胞癌外泌体mRNA表达图谱分析[J]. 口腔医学研究, 2018, 34(9): 943-947.