基于深度学习的胰腺黏液性和浆液性囊性肿瘤的多源特征分类模型

doi:10.11938/cjmr20233064

摘要/Abstract

摘要：

术前精准预测胰腺囊性肿瘤的类型，对制定个体化诊疗方案具有重要的临床价值．针对胰腺浆液性和黏液性囊性肿瘤的分类鉴别问题，本文探讨了基于深度学习的多源特征分类模型在胰腺囊性肿瘤的术前辅助诊断中的应用．首先，通过深度学习和影像组学技术从分割图像中提取深度学习特征和影像组学特征，并对病例的临床特征进行评估和量化，然后采用最小绝对收缩选择算子（LASSO）及交叉验证的方法筛选特征，随之构建出两个多源特征模型，即影像组学联合深度学习（RAD_DL）模型、临床特征联合RAD_DL（Clinical_RAD_DL）模型，把传统的影像组学（RAD）模型和深度学习（DL）模型作为对照，最后选用支持向量机（SVM）、自适应提升算法（ADAboost）、随机森林（Random Forest）以及逻辑回归（Logistic）进行分类．采用准确率、召回率、精确率、曲线下面积（AUC）值以及精确率和召回率的调和平均数（F1值）作为评价指标，比较上述4种不同特征模型的分类效能，用校准曲线和决策曲线来评估其临床应用价值．结果显示Clinical_RAD_DL特征模型的分类效能表现最佳，准确率是0.923 1，召回率是0.882 4，精确率是0.882 0，F1是0.882 2，AUC是0.912 6，校准曲线和决策曲线显示出Clinical_RAD_DL特征模型的临床应用价值是最高的．实验表明基于深度学习的多源特征分类模型，对胰腺黏液性和浆液性囊性肿瘤具有较好的分类效果，可以为临床上精准诊疗提供帮助．

关键词: 磁共振成像（MRI）, 胰腺囊性肿瘤, 临床特征, 深度学习, 影像组学

Abstract:

This study aims to classify and differentiate mucinous and serous cystic neoplasms of the pancreas using a multi-source feature classification model based on deep learning for preoperative auxiliary diagnosis. Deep learning features and radiomics features were extracted from segmented images using deep learning and radiomics technology, respectively. Clinical features were also evaluated and quantified. LASSO (least absolute shrinkage and selection operator) and cross-validation methods were applied to screen the features, and two multi-source feature models were constructed: the radiomics combined with deep learning (RAD_DL) model and the clinical feature combined with RAD_DL (Clinical_RAD_DL) model. Traditional radiomics (RAD) and deep learning (DL) models were used as controls. SVM (support vector machine), ADAboost (adaptive boosting), Random Forest, and Logistic were selected for classification. The Clinical_RAD_DL feature model shows the best classification performance, with the accuracy of 0.923 1, recall rate of 0.882 4, precision of 0.882 0, F1-score of 0.882 2, and AUC value of 0.912 6. The experimental results indicate that the multi-source feature classification model based on deep learning has good performance in classifying pancreatic serous cystic neoplasms and pancreatic mucinous cystic neoplasms, and can assist clinical accurate diagnosis and treatment.

Key words: magnetic resonance imaging (MRI), pancreatic cystic neoplasms, clinical features, deep learning, radiomics

中图分类号:

O482.53

徐真顺, 袁小涵, 黄子珩, 邵成伟, 武杰, 边云. 基于深度学习的胰腺黏液性和浆液性囊性肿瘤的多源特征分类模型[J]. 波谱学杂志, 2024, 41(1): 19-29.

XU Zhenshun, YUAN Xiaohan, HUANG Ziheng, SHAO Chengwei, WU Jie, BIAN Yun. Multi-source Feature Classification Model of Pancreatic Mucinous and Serous Cystic Neoplasms Based on Deep Learning[J]. Chinese Journal of Magnetic Resonance, 2024, 41(1): 19-29.

图/表 10

图1

图2

表1

表2

图3

表3

图4

表4

图5

参考文献 20

[1]	LI Z S, JIN Z D, LI X. Chinese guidelines for the diagnosis and treatment of pancreas cystic neoplasm(2022)[J]. J Clin Hepatol, 2023, 39(2): 290-298.
	李兆申, 金震东, 李汛. 中国胰腺囊性肿瘤诊断指南(2022年)[J]. 临床肝胆病杂志, 2023, 39(2): 290-298.
[2]	BIAN Y, JIANG H, CAO K, et al. The relationship between microscopic tumor size and CT tumor size in pancreatic ductal adenocarcinoma[J]. Clin Imag, 2021, 76: 30-37. doi: 10.1016/j.clinimag.2020.11.039 pmid: 33548890
[3]	JAYASREE C, ABHISHEK M, LIOR G, et al. CT radiomics to predict high-risk intraductal papillary mucinous neoplasms of the pancreas[J]. Med phys, 2018, 45(11): 5019-5029. doi: 10.1002/mp.13159 pmid: 30176047
[4]	AVANZO M, WEI L S, STANCANELLO J, et al. Machine and deep learning methods for radiomics[J]. Med Phys, 2020, 47(5): 185-202. doi: 10.1002/mp.13678 pmid: 32418336
[5]	HAN B, XU J, WANG Y J, et al. Classification of BI-RADS 3-5 breast lesions based on MRI radiomics[J]. Chinese J Magn Reson, 2023, 40(1): 52-67.
	韩冰, 徐晶, 王远军, 等. 基于MRI影像组学的BI-RADS 3-5类乳腺病变三分类[J]. 波谱学杂志, 2023, 40(1): 52-67.
[6]	SHBOUL Z A, ALAM M, VIDYARATNE L, et al. Feature-guided deep radiomics for glioblastoma patient survival prediction[J]. Front Neurosci, 2019, 13: 966-982. doi: 10.3389/fnins.2019.00966
[7]	LI J, LIU F, FANG X, et al. CT Radiomics features in differentiation of focal-type autoimmune pancreatitis from pancreatic ductal adenocarcinoma: a propensity score analysis[J]. Acad Radiol, 2021, 29(3): 358-366. doi: 10.1016/j.acra.2021.04.014 pmid: 34108115
[8]	PREUSS K, THACH N, LIANG X, et al. Using quantitative imaging for personalized medicine in pancreatic cancer: a review of radiomics and deep learning applications[J]. Cancers (Basel), 2022, 14(7): 1654. doi: 10.3390/cancers14071654
[9]	NGUON L S, SEO K, LIM J H, et al. Deep learning-based differentiation between mucinous cystic neoplasm and serous cystic neoplasm in the pancreas using endoscopic ultrasonography[J]. Diagnostics (Basel), 2021, 11(6): 1052.
[10]	YANG Y F, QI Z X, NIE S D. Differentiation of benign and malignant breast lesions based on multimodal MRI and deep learning[J]. Chinese J Magn Reson, 2022, 39(4): 401-412.
	杨一风, 祁章璇, 聂生东. 基于多模态MRI与深度学习的乳腺病变良恶性鉴别[J]. 波谱学杂志, 2022, 39(4): 401-412.
[11]	ZHANG Y F, XU S S, WU J, et al. Value of CT texture analysis in differentiating pancreatic serous cystadenoma from mucinous cystadenoma[J]. Journal of Southeast University, 2022, 41(3): 308-316.
	张怡帆, 徐姗姗, 吴锦, 等. CT纹理分析在鉴别胰腺浆液性囊腺瘤与黏液性囊腺瘤中的价值[J]. 东南大学学报, 2022, 41(3): 308-316.
[12]	PARK H J, SHIN K, YOU M W, et al. Deep learning-based detection of solid and cystic pancreatic neoplasms at contrast-enhanced CT[J]. Radiology, 2023, 306(1): 140-149. doi: 10.1148/radiol.220171
[13]	ISLAM W, JONES M, FAIZ R, et al. Improving performance of breast lesion classification using a ResNet50 model optimized with a novel attention mechanism[J]. Tomography, 2022, 8(5): 2411-2425. doi: 10.3390/tomography8050200 pmid: 36287799
[14]	LU J, WU Y, XIONG Y, et al. Breast tumor computer-aided detection system based on magnetic resonance imaging using convolutional neural network[J]. Comput Model Eng Sci, 2022, 130(1): 365-377.
[15]	中国人民解放军海军军医大学第一附属医院. 胰腺外分泌功能评价模型的训练方法、系统、电子设备及介质: 中国, CN2022107020513[P], 2022-08-09.
[16]	LIANG W, TIAN W, WANG Y, et al. Classification prediction of pancreatic cystic neoplasms based on radiomics deep learning models[J]. BMC Cancer, 2022, 22(1): 1237-1246. doi: 10.1186/s12885-022-10273-4 pmid: 36447168
[17]	WANG D, HU Y, ZHAN C, et al. A nomogram based on radiomics signature and deep-learning signature for preoperative prediction of axillary lymph node metastasis in breast cancer[J]. Front Oncol, 2022, 12(20): 940-950.
[18]	XIE T, WANG X, ZHANG Z, et al. CT-based radiomics analysis for preoperative diagnosis of pancreatic mucinous cystic neoplasm and atypical serous cystadenomas[J]. Front Oncol, 2021, 11(6): 621520. doi: 10.3389/fonc.2021.621520
[19]	SHAO C, FENG X, YU J, et al. A nomogram for predicting pancreatic mucinous cystic neoplasm and serous cystic neoplasm[J]. Abdom Radiol (NY), 2021, 46(8): 3963-3973.
[20]	CHEN S. Application of CT radiomics in differential diagnosis of pancreatic serous and mucinous cystic neoplasm[J]. Chinese Journal of CT and MRI, 2022, 20(10): 92-105.
	陈帅. 基于CT影像组学对胰腺浆液及黏液性囊性肿瘤鉴别诊断[J]. 中国CT和MRI杂志, 2022, 20(10): 92-105.

特征名称	p值	OR值	95%CI
特征名称	p值	OR值	下限	上限
位置（胰头/胰体/胰尾）	<0.05	2.22	1.47	3.34
性别（男/女）	<0.05	6.30	1.82	21.82
年龄	0.14	0.98	0.96	1.01
肿瘤体积	<0.05	1.00	1.00	1.00

特征名称	p值	OR值	95%CI
特征名称	p值	OR值	下限	上限
位置（胰头/胰体/胰尾）	<0.05	2.12	1.39	3.24
性别（男/女）	<0.05	5.58	1.57	19.84

特征模型	分类器	准确率	召回率	精确率
RAD	SVM	0.8077	0.5789	0.8462	0.7592	0.6875
	ADAboost	0.8269	0.6842	0.8125	0.7967	0.7429
	Random Forest	0.7885	0.6742	0.7222	0.7663	0.6974
	Logistic	0.8269	0.6316	0.8571	0.7855	0.7273
DL	SVM	0.6731	0.1176	0.5303	0.5595	0.1925
	ADAboost	0.7692	0.9000	0.6429	0.8645	0.7500
	Random Forest	0.7115	0.7619	0.6154	0.8407	0.6809
	Logistic	0.7500	0.2857	0.5714	0.5770	0.3809
RAD_DL	SVM	0.8462	0.6111	0.7908	0.8051	0.6894
	ADAboost	0.8269	0.7727	0.8095	0.6882	0.7907
	Random Forest	0.8077	0.6364	0.8750	0.8051	0.7369
	Logistic	0.8462	0.9444	0.7083	0.8441	0.8095
Clinical_RAD_DL	SVM	0.8846	0.8235	0.8235	0.8689	0.8235
	ADAboost	0.8654	0.8823	0.7500	0.8697	0.8108
	Random Forest	0.7692	0.7059	0.6316	0.7529	0.6667
	Logistic	0.9231	0.8824	0.8820	0.9126	0.8822