计算机辅助下自体荧光图像定量结果与口腔白斑病上皮异常增生等级的相关性

doi:10.3969/j.issn.1674-8115.2024.09.009

上海交通大学学报（医学版） ›› 2024, Vol. 44 ›› Issue (9): 1146-1154.doi: 10.3969/j.issn.1674-8115.2024.09.009

• 论著 · 临床研究 • 上一篇

计算机辅助下自体荧光图像定量结果与口腔白斑病上皮异常增生等级的相关性

李晨曦¹(), 王子瑞¹, 金恬昊¹, 周曾同¹, 唐国瑶¹^,²(), 施琳俊¹()

^1.上海交通大学医学院附属第九人民医院口腔黏膜病科，上海交通大学口腔医学院，国家口腔医学中心，国家口腔疾病临床医学研究中心，上海市口腔医学重点实验室，上海市口腔医学研究所，上海 200011
^2.上海交通大学医学院附属新华医院口腔科，上海 200092

收稿日期:2024-05-03 接受日期:2024-07-30 出版日期:2024-09-28 发布日期:2024-10-09
通讯作者: 唐国瑶,施琳俊 E-mail:lichenxi0327@qq.com;tanggy@shsmu.edu.cn;drshilinjun@126.com
作者简介:李晨曦（1993—），女，博士生；电子信箱：lichenxi0327@qq.com。
基金资助:
国家自然科学基金(82170952);国家重点研发计划(2022YFC2402900);上海交通大学医学院“双百人”项目(20221813);上海市卫生健康委员会临床研究项目(20214Y0192)

Correlation between computer-assisted quantitative autofluorescence imaging results and the pathological grading of oral epithelial dysplasia in oral leukoplakia

LI Chenxi¹(), WANG Zirui¹, JIN Tianhao¹, ZHOU Zengtong¹, TANG Guoyao¹^,²(), SHI Linjun¹()

^1.Department of Oral Mucosal Diseases, Shanghai Ninth People′s Hospital, Shanghai Jiao Tong University School of Medicine; College of Stomatology, Shanghai Jiao Tong University; National Center for Stomatology; National Clinical Research Center for Oral Diseases; Shanghai Key Laboratory of Stomatology; Shanghai Research Institute of Stomatology, Shanghai 200011, China
^2.Department of stomatology, Shanghai Xin Hua Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai 200092, China

Received:2024-05-03 Accepted:2024-07-30 Online:2024-09-28 Published:2024-10-09
Contact: TANG Guoyao,SHI Linjun E-mail:lichenxi0327@qq.com;tanggy@shsmu.edu.cn;drshilinjun@126.com
Supported by:
National Natural Science Foundation of China(82170952);National Key Research and Development Program(2022YFC2402900);“Two-Hundred Talents” Program of Shanghai Jiao Tong University School of Medicine(20221813);Clinical Research Program of Shanghai Municipal Health Commission(20214Y0192)

摘要/Abstract

摘要：

目的·探究计算机辅助下自体荧光图像定量结果与口腔白斑病上皮异常增生等级的相关性。方法·纳入2016年4月—2024年1月于上海交通大学医学院附属第九人民医院口腔黏膜病科就诊的口腔白斑病患者357例。利用手持自体荧光仪器获取患者病损的自体荧光图像，将自体荧光图像转为灰度图像，获得量化指标。在Python中拟合有序多元Logistic回归模型，绘制累积概率图。将数据集划分训练集和测试集，生成决策树，调整不同的超参数，获得最佳的模型效果。计算准确度、精确度和F1分值。利用混淆矩阵对模型性能进行可视化呈现。结果·随着上皮异常增生程度的增加，相对色阶平均值呈现下降趋势。在上皮异常增生二分类中累积概率图不同类别曲线之间无重叠，在四分类中仅上皮重度异常增生与其他类别曲线有重叠，说明模型的区分能力较好。在二分类病理等级中，当训练集和测试集比例为4∶1、决策树最大深度为2时，准确度、精确度、F1分值可达到较高，分别为0.792、0.801和0.795。在四分类病理等级中，当训练集和测试集比例为9∶1、决策树最大深度为4时，准确度、精确度、F1分值可达到较高，分别为0.611、0.537和0.569。结论·口腔黏膜病专科医师可将计算机辅助下自体荧光图像定量结果作为参考依据，预测口腔白斑病患者上皮异常增生程度，监控患者癌变风险。

关键词: 自体荧光图像, 口腔白斑病, 上皮异常增生, 有序多元Logistic回归模型, 混淆矩阵

Abstract:

Objective ·To explore the correlation between the quantitative results of autofluorescence imaging under computer assistance and the grade of epithelial dysplasia in oral leukoplakia. Methods ·From April 2016 to January 2024, 357 patients with oral leukoplakia who visited the Department of Oral Mucosal Diseases at Shanghai Ninth People′s Hospital, Shanghai Jiao Tong University School of Medicine, were included. Autofluorescence images of the lesions were obtained using a handheld autofluorescence device. These images were converted to grayscale images to obtain quantitative metrics. An ordered multinomial Logistic regression model was fitted in Python, and cumulative probability plots were generated. The dataset was divided into training and testing sets, and a decision tree was generated. Different hyperparameters were adjusted to achieve optimal model performance. Accuracy, precision, and F1 scores were calculated. The model performance was visualized using a confusion matrix. Results ·As the degree of epithelial dysplasia increased, the relative mean color level showed a declining trend. In the binary classification of epithelial dysplasia, there was no overlap between the cumulative probability curves of different categories. In the four-category classification, only severe epithelial dysplasia overlapped with other category curves, indicating good discriminative ability of the model. In binary pathological grading, when the training and testing set ratio was 4∶1 and the maximum depth was 2, the accuracy, precision, and F1 scores were 0.792, 0.801, and 0.795, respectively. In the four-category pathological grading, when the training and testing set ratio was 9∶1 and the maximum depth was 4, the accuracy, precision, and F1 scores were 0.611, 0.537, and 0.569, respectively. Conclusion ·Computer-assisted quantitative analysis of autofluorescence images can be used by oral mucosal specialists as a reference to predict the degree of epithelial dysplasia in patients with oral leukoplakia and to monitor their risk of cancer.

Key words: autofluorescence imaging, oral leukoplakia, epithelial dysplasia, ordered multinomial Logistic regression model, confusion matrix

中图分类号:

R781.5⁺9

李晨曦, 王子瑞, 金恬昊, 周曾同, 唐国瑶, 施琳俊. 计算机辅助下自体荧光图像定量结果与口腔白斑病上皮异常增生等级的相关性[J]. 上海交通大学学报（医学版）, 2024, 44(9): 1146-1154.

LI Chenxi, WANG Zirui, JIN Tianhao, ZHOU Zengtong, TANG Guoyao, SHI Linjun. Correlation between computer-assisted quantitative autofluorescence imaging results and the pathological grading of oral epithelial dysplasia in oral leukoplakia[J]. Journal of Shanghai Jiao Tong University (Medical Science), 2024, 44(9): 1146-1154.

图/表 7

图1 自体荧光图像的量化Note： A. Original autofluorescence image. B. Converted grayscale image. C. Grayscale image was used to obtain a histogram of grayscale levels, which was used to obtain the mean grayscale value. D. Selection of two negative regions as background controls. E. Obtaining the mean color level of the negative regions. F. Selection of one positive region. G. Obtaining the MCL of the positive region.

Fig 1 Quantification of autofluorescence imaging

图2 RMCL分布图Note： A decreasing trend in RMCL was observed as the degree of epithelial dysplasia increased.

Fig 2 Distribution of RMCL

图3 不同级别上皮异常增生的4例口腔白斑病

Fig 3 Four cases of oral leukoplakia with different levels of epithelial dysplasia

图4 有序多元Logistic回归模型预测病理结果的累积概率图

Fig 4 Cumulative probability plot for the prediction of pathological results using an ordinal multivariate Logistic regression model

表1 二分类病理结果的准确度、精确度和F1分值

Tab 1 Accuracy， precision， and F1 scores of binary classification of pathological results

Maximum depth of decision tree	Index	The proportion of the test set in all samples
Maximum depth of decision tree	Index	0.10	0.15	0.20	0.25	0.30	0.35	0.40	0.45	0.50
2	Accuracy	0.667	0.741	0.792	0.656	0.648	0.632	0.650	0.671	0.665
	Precision	0.675	0.749	0.801	0.659	0.655	0.555	0.604	0.622	0.605
	F1 score	0.669	0.744	0.795	0.547	0.534	0.522	0.556	0.586	0.578
3	Accuracy	0.639	0.685	0.750	0.589	0.648	0.680	0.629	0.603	0.609
	Precision	0.643	0.700	0.764	0.485	0.619	0.665	0.560	0.536	0.536
	F1 score	0.640	0.690	0.755	0.508	0.596	0.663	0.551	0.554	0.554
4	Accuracy	0.667	0.704	0.750	0.711	0.620	0.664	0.630	0.634	0.631
	Precision	0.667	0.713	0.764	0.707	0.579	0.649	0.572	0.552	0.544
	F1 score	0.667	0.707	0.755	0.709	0.576	0.650	0.566	0.562	0.557

图5 模型性能最佳时的混淆矩阵和决策树Note： A. Confusion matrix of the test set prediction model for binary classification of pathological grade. Darker color blocks along the diagonal indicate a higher probability of correct predictions. B. Confusion matrix of the test set prediction model for four-class classification of pathological grade. Dark color blocks are primarily concentrated in the second and third column, indicating that the model tends to predict the degree of epithelial dysplasia as mild or moderate. C. Decision tree of the training set for binary classification of pathological grade, showing the prediction process of each sample. The red circle indicates the root node, and the red square indicates the leaf nodes, with different classification results. D. Decision tree of the training set for four-class classification of pathological grade, showing the prediction process of each sample. The red circle indicates the root node, and the red square indicates the leaf nodes, with different classification results.

Fig 5 Confusion matrix and decision tree with optimal model performance

表2 四分类病理结果的准确度、精确度和F1分值

Tab 2 Accuracy， precision， and F1 scores of four-class classification of pathological results

Max imum depth of decision tree	Index	The proportion of the test set in all samples
Max imum depth of decision tree	Index	0.10	0.15	0.20	0.25	0.30	0.35	0.40	0.45	0.50
3	Accuracy	0.528	0.537	0.611	0.556	0.519	0.512	0.546	0.565	0.553
	Precision	0.302	0.344	0.512	0.395	0.385	0.394	0.476	0.352	0.346
	F1 score	0.384	0.419	0.556	0.425	0.424	0.414	0.421	0.433	0.423
4	Accuracy	0.611	0.574	0.514	0.522	0.537	0.536	0.552	0.571	0.559
	Precision	0.537	0.500	0.337	0.422	0.418	0.459	0.541	0.515	0.455
	F1 score	0.569	0.533	0.407	0.454	0.427	0.438	0.465	0.489	0.455
5	Accuracy	0.528	0.537	0.542	0.544	0.556	0.528	0.483	0.528	0.525
	Precision	0.414	0.386	0.340	0.443	0.454	0.550	0.468	0.525	0.465
	F1 score	0.422	0.446	0.417	0.460	0.467	0.453	0.423	0.506	0.460

参考文献 27

1	GONZÁLEZ-MOLES M Á, RUIZ-ÁVILA I, GONZÁLEZ-RUIZ L, et al. Malignant transformation risk of oral lichen planus: a systematic review and comprehensive meta-analysis[J]. Oral Oncol, 2019, 96: 121-130.
2	WARNAKULASURIYA S, REIBEL J, BOUQUOT J, et al. Oral epithelial dysplasia classification systems: predictive value, utility, weaknesses and scope for improvement[J]. J Oral Pathol Med, 2008, 37(3): 127-133.
3	SPERANDIO M, BROWN A L, LOCK C, et al. Predictive value of dysplasia grading and DNA ploidy in malignant transformation of oral potentially malignant disorders[J]. Cancer Prev Res, 2013, 6(8): 822-831.
4	IOCCA O, SOLLECITO T P, ALAWI F, et al. Potentially malignant disorders of the oral cavity and oral dysplasia: a systematic review and meta-analysis of malignant transformation rate by subtype[J]. Head Neck, 2020, 42(3): 539-555.
5	WARNAKULASURIYA S. Oral potentially malignant disorders: a comprehensive review on clinical aspects and management[J]. Oral Oncol, 2020, 102: 104550.
6	AWAN K H, MORGAN P R, WARNAKULASURIYA S. Evaluation of an autofluorescence based imaging system (VELscope™) in the detection of oral potentially malignant disorders and benign keratoses[J]. Oral Oncol, 2011, 47(4): 274-277.
7	LEUCI S, COPPOLA N, TURKINA A, et al. May VelScope be deemed an opportunistic oral cancer screening by general dentists? A pilot study[J]. J Clin Med, 2020, 9(6): 1754.
8	FARAH C S, DOST F, DO L. Usefulness of optical fluorescence imaging in identification and triaging of oral potentially malignant disorders: a study of VELscope in the LESIONS programme[J]. J Oral Pathol Med, 2019, 48(7): 581-587.
9	FLORES DOS SANTOS L C, FERNANDES J R, LIMA I F P, et al. Applicability of autofluorescence and fluorescent probes in early detection of oral potentially malignant disorders: a systematic review and meta-data analysis[J]. Photodiagnosis Photodyn Ther, 2022, 38: 102764.
10	李晨曦, 施琳俊, 唐国瑶. 无创筛查技术早期诊断口腔潜在恶性病患癌变的研究进展[J]. 临床口腔医学杂志, 2017, 33(6): 381-383.
	LI C X, SHI L J, TANG G Y. Advances in early diagnosis of oral potentially malignant disorders through non-invasive screening techniques[J]. Journal of Clinical Stomatology, 2017, 33(6): 381-383.
11	SHI L J, LI C X, SHEN X M, et al. Potential role of autofluorescence imaging in determining biopsy of oral potentially malignant disorders: a large prospective diagnostic study[J]. Oral Oncol, 2019, 98: 176-179.
12	LI C X, ZHANG Q Q, SUN K, et al. Autofluorescence imaging as a noninvasive tool of risk stratification for malignant transformation of oral leukoplakia: a follow-up cohort study[J]. Oral Oncol, 2022, 130: 105941.
13	中华口腔医学会口腔黏膜病专业委员会. 口腔白斑病的定义与分级标准(试行)[J]. 中华口腔医学杂志, 2011, 46(10): 579-580.
	Oral Mucosal Diseases Committee of the Chinese Stomatological Association. Definition and Grading Criteria of Oral Leukoplakia (Trial)[J]. Chinese Journal of Stomatology, 2011, 46(10): 579-580.
14	WARNAKULASURIYA S, JOHNSON N W, VAN DER WAAL I. Nomenclature and classification of potentially malignant disorders of the oral mucosa[J]. J Oral Pathol Med, 2007, 36(10): 575-580.
15	MONICI M. Cell and tissue autofluorescence research and diagnostic applications[J]. Biotechnol Annu Rev, 2005, 11: 227-256.
16	BUENAHORA M R, PERAZA-L A, DÍAZ-BÁEZ D, et al. Diagnostic accuracy of clinical visualization and light-based tests in precancerous and cancerous lesions of the oral cavity and oropharynx: a systematic review and meta-analysis[J]. Clin Oral Investig, 2021, 25(6): 4145-4159.
17	CHIANG T E, LIN Y C, LI Y H, et al. Comparative evaluation of autofluorescence imaging and histopathological investigation for oral potentially malignant disorders in Taiwan[J]. Clin Oral Investig, 2019, 23(5): 2395-2402.
18	TIWARI L, KUJAN O, FARAH C S. Optical fluorescence imaging in oral cancer and potentially malignant disorders: a systematic review[J]. Oral Dis, 2020, 26(3): 491-510.
19	吴露姝, 杨建堂, 张绍伟. 自体荧光技术在肿瘤诊断中的应用及研究进展[J]. 临床口腔医学杂志, 2019, 35(7): 438-440.
	WU L S, YANG J T, ZHANG S W. Application and research progress of autofluorescence technology in tumor diagnosis[J]. Journal of Clinical Stomatology. 2019, 35(7): 438-440.
20	KIM D H, KIM S W, HWANG S H. Autofluorescence imaging to identify oral malignant or premalignant lesions: systematic review and meta-analysis[J]. Head Neck, 2020, 42(12): 3735-3743.
21	MENDONCA P, SUNNY S P, MOHAN U, et al. Non-invasive imaging of oral potentially malignant and malignant lesions: a systematic review and meta-analysis[J]. Oral Oncol, 2022, 130: 105877.
22	李晨曦, 施芷怡, 喻仲麟, 等. 口腔白斑病自体荧光成像结果与临床病理特征相关性分析[J]. 临床口腔医学杂志, 2023, 39(3): 153-157.
	LI C X, SHI Z Y, YU Z L, et al. Correlation analysis between autofluorescence imaging results and clinicopathological characteristics of oral leukoplakia[J]. Journal of Clinical Stomatology, 2023, 39(3): 153-157.
23	SCIUBBA J, EPSTEIN J. Is no biopsy appropriate for oral potentially malignant lesion(s) without loss of autofluorescence using VELscope^®? A large prospective diagnostic study[J]. Oral Oncol, 2020, 102: 104474.
24	PENTENERO M, TODARO D, MARINO R, et al. Interobserver and intraobserver variability affecting the assessment of loss of autofluorescence of oral mucosal lesions[J]. Photodiagnosis Photodyn Ther, 2019, 28: 338-342.
25	HUANG T T, HUANG J S, WANG Y Y, et al. Novel quantitative analysis of autofluorescence images for oral cancer screening[J]. Oral Oncol, 2017, 68: 20-26.
26	QUANG T, TRAN E Q, SCHWARZ R A, et al. Prospective evaluation of multimodal optical imaging with automated image analysis to detect oral neoplasia in vivo[J]. Cancer Prev Res, 2017, 10(10): 563-570.
27	CHERRY K D, SCHWARZ R A, YANG E C, et al. Autofluorescence imaging to monitor the progression of oral potentially malignant disorders[J]. Cancer Prev Res, 2019, 12(11): 791-800.

计算机辅助下自体荧光图像定量结果与口腔白斑病上皮异常增生等级的相关性

Correlation between computer-assisted quantitative autofluorescence imaging results and the pathological grading of oral epithelial dysplasia in oral leukoplakia

RichHTML

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

图/表 7

参考文献 27

相关文章 1

编辑推荐

Metrics