MRI-based radiomics and deep learning model construction: non-invasive differentiation of molecular subtypes in primary intracranial diffuse large B-cell lymphoma

Yanwei ZENG; Zhijian XU; Xin CAO; Kun LÜ; Huiming LI; Min GAO; Shenghong JU; Jun LIU; Daoying GENG

doi:10.19401/j.cnki.1007-3639.2025.08.001

您当前的位置：

首页 >

文章列表页 >

MRI-based radiomics and deep learning model construction: non-invasive differentiation of molecular subtypes in primary intracranial diffuse large B-cell lymphoma

Article | 更新时间：2025-12-31

- MRI-based radiomics and deep learning model construction: non-invasive differentiation of molecular subtypes in primary intracranial diffuse large B-cell lymphoma
- China Oncology Vol. 35, Issue 8, Pages: 735-742(2025)
- 作者机构：
  
  1. 复旦大学附属华山医院放射科，上海 200040
  2. 上海脑重大疾病智能影像工程技术研究中心，上海 200040
  3. 复旦大学医学功能与分子影像研究所，上海 200040
  4. 复旦大学工程与应用技术研究院，上海 200433
  5. 东南大学附属中大医院放射科，江苏南京 210009
  6. 中南大学湘雅二医院放射科，湖南长沙 410011
- 作者简介：
- 基金信息：
  
  National Natural Science Foundation of China(82372048);Science and Technology Commission of Shanghai Municipality Fund(23S31904100)
- DOI：10.19401/j.cnki.1007-3639.2025.08.001
  CLC： R733.4
- Received：06 May 2025，
  
  Revised：2025-06-13，
  
  Published：30 August 2025
- 稿件说明：
移动端阅览
Yanwei ZENG, Zhijian XU, Xin CAO, et al. MRI-based radiomics and deep learning model construction: non-invasive differentiation of molecular subtypes in primary intracranial diffuse large B-cell lymphoma[J]. China Oncology, 2025, 35(8): 735-742.
DOI：

Yanwei ZENG, Zhijian XU, Xin CAO, et al. MRI-based radiomics and deep learning model construction: non-invasive differentiation of molecular subtypes in primary intracranial diffuse large B-cell lymphoma[J]. China Oncology, 2025, 35(8): 735-742. DOI： 10.19401/j.cnki.1007-3639.2025.08.001.

摘要

背景与目的：

弥漫大B细胞淋巴瘤（diffuse large B-cell lymphoma，DLBCL）的生发中心B细胞样（germinal center B-cell-like，GCB）亚型和非GCB（non-GCB）亚型在患者预后和治疗上存在差异，但目前依赖有创病理学检查。本研究基于多参数MRI构建影像组学和深度学习模型，旨在于术前无创性区分这两种亚型。

方法：

本研究回顾性分析2013年3月—2024年12月在复旦大学附属华山医院及外院经病理学检查确诊的DLBCL患者。使用多参数MRI扫描数据，结合4种影像组学机器学习[支持向量机（support vector machine，SVM）、逻辑回归（logistic regression，LR）、高斯过程（Gaussian process，GP）和朴素贝叶斯（Naive Bayes，NB）

]

和3种深度学习[密集连接卷积网络121（densely-connected convolutional networks 121，DenseNet121）、残差网络101（residual network 101，ResNet101）和高效网络B5（EfficientNet-b5）

]

建立DLBCL亚型分类模型。此外，两名经验不同的放射科医师在盲法下基于MRI图像独立分类DLBCL。模型和医师的诊断性能均通过接收者操作特征曲线下面积（area under the curve，AUC）、准确度（accuracy，ACC）和F1分数（F1-score，F1）等指标进行量化评估，以衡量其区分GCB和non-GCB亚型的能力。本研究经复旦大学附属华山医院伦理委员会批准（KY2024-663），所有患者均知情同意。

结果：

本研究共纳入173例患者（GCB型55例，non-GCB型118例）。影像组学和深度学习方法能有效地区分DLBCL亚型。其中，GP影像组学模型（基于T1-CE+T2-FLAIR+ADC序列）和DenseNet121深度学习模型（基于T1-CE+T2-FLAIR+ADC序列）表现最佳，在内部验证集上分别取得优异性能（GP: AUC=0.900，ACC=0.896，F1=0.840；DenseNet121: AUC=0.846，ACC=0.854，F1=0.774），并在外部验证集上保持稳健。并且，最优AI模型的分类效能优于经验丰富的放射科医师（医师最高AUC=0.678）。

结论：

基于多参数MRI特征的影像组学与深度学习模型可有效地鉴别DLBCL的GCB与non-GCB亚型。其中，GP与DenseNet121模型在处理复杂图像数据、特别是融合多序列特征组进行亚型分类时，呈现出优异的性能。

Abstract

Background and purpose:

Diffuse large B-cell lymphoma (DLBCL) is subclassified into germinal center B-cell-like (GCB) and non-GCB subtypes

which differ in prognosis a

nd treatment response. However

current distinction still relies on invasive pathological assays. This study developed radiomics and deep-learning models based on multiparametric magnetic resonance imaging (MRI) to non-invasively differentiate the two subtypes preoperatively

thereby reducing dependence on histopathological examination.

Methods:

This study retrospectively included patients with pathologically confirmed DLBCL diagnosed at Huashan Hospital

Fudan University

and other institutions between March 2013 and December 2024. Using multiparametric MRI data

we developed DLBCL-subtype classification models that combined 4 radiomics-based machine-learning algorithms: support vector machine (SVM)

logistic regression (LR)

Gaussian process (GP) and Naive Bayes (NB)

with 3 deep-learning architectures [densely-connected convolutional networks 121 (DenseNet121)

residual network 101 (ResNet101) and EfficientNet-b5

]

. Additionally

two radiologists with different experience levels independently classified DLBCL on MRI in a blinded fashion. Model and radiologist performance were quantified using the area under the receiver operating characteristic curve (AUC)

accuracy (ACC)

and F1-score to evaluate their ability to distinguish GCB from non-GCB subtypes. This study was approved by the Ethics Committee of Huashan Hospital of Fudan University (No. KY2024-663)

and all patients signed informed consents.

Results:

A total of 173 patients were enrolled (55 with GCB subtype and 118 with non-GCB subtype). Radiomics and deep learning methods effectively distinguished DLBCL subtypes. Among these

the GP radiomics model (based on T1-CE+T2-FLAIR+ADC sequences) and DenseNet121 deep learning model (based on T1-CE+T2-FLAIR+ADC sequences) demonstrated optimal performance. Both achieved excellent results on the internal validation set (GP: AUC=0.900

ACC=0.896

F1=0.840; DenseNet121: AUC=0.846

ACC=0.854

F1=0.774) and maintained robustness on the external validation set. Furthermore

the classification

efficacy of the optimal AI model surpassed that of experienced radiologists (highest physician AUC=0.678).

Conclusion:

Radiomics and deep-learning models based on multiparametric MRI features can effectively differentiate GCB from non-GCB subtypes of DLBCL. Among them

GP and DenseNet121 exhibit outstanding performance

especially when integrating multi-sequence feature sets for classifying DLBCL subtypes on complex imaging data.

关键词

Keywords

references

SCHAFF L R , GROMMES C . Primary central nervous system lymphoma [J ] . Blood , 2022 , 140 ( 9 ): 971 - 979 .

中华医学会血液学分会淋巴细胞疾病学组 , 中国临床肿瘤学会(CSCO)淋巴瘤专家委员会 . 原发性中枢神经系统淋巴瘤诊断及治疗专家共识(2024年版) [J ] . 白血病·淋巴瘤 , 2024 , 33 ( 3 ): 129 - 137 .

The Lymphocytic Disease Group of the Hematology Branch of the Chinese Medical Association and the Lymphoma Expert Committee of the Chinese Society of Clinical Oncology (CSCO) Expert . Consensus on diagnosis and treatment of primary central nervous system lymphoma (2024 edition) [J ] . J Leuk Lymph , 2024 , 33 ( 3 ): 129 - 137 .

HANS C P , WEISENBURGER D D , GREINER T C , et al . Confirmation of the molecular classification of diffuse large B-cell lymphoma by immunohistochemistry using a tissue microarray [J ] . Blood , 2004 , 103 ( 1 ): 275 - 282 . DOI: 10.1182/blood-2003-05-1545 http://doi.org/10.1182/blood-2003-05-1545

WILSON W H , WRIGHT G W , HUANG D W , et al . Effect of ibrutinib with R-CHOP chemotherapy in genetic subtypes of DLBCL [J ] . Cancer Cell , 2021 , 39 ( 12 ): 1643 - 1653 .e3. DOI: 10.1016/j.ccell.2021.10.006 http://doi.org/10.1016/j.ccell.2021.10.006

MARCUS C , MARAGKOS G A , ALTERMAN R L , et al . GCB-type is a favorable prognostic factor in primary CNS diffuse large B-cell lymphomas [J ] . J Clin Neurosci , 2021 , 83 : 49 - 55 . DOI: 10.1016/j.jocn.2020.11.031 http://doi.org/10.1016/j.jocn.2020.11.031

NIPARUCK P , BOONSAKAN P , SUTTHIPPINGKIAT T , et al . Treatment outcome and prognostic factors in PCNSL [J ] . Diagn Pathol , 2019 , 14 ( 1 ): 56 . DOI: 10.1186/s13000-019-0833-1 http://doi.org/10.1186/s13000-019-0833-1

曾延玮 , 曹鑫 , 吕锟 , 等 . 多参数磁共振成像鉴别中枢神经系统弥漫大B细胞淋巴瘤的亚型 [J ] . 中国医学计算机成像杂志 , 2025 , 31 ( 2 ): 156 - 161 .

ZENG Y W , CAO X , LÜ K , et al . Non-invasive discrimination of cerebral diffuse large B-cell lymphoma subtypes using multi-parametric MRI [J ] . Chin Comput Med Imag , 2025 , 31 ( 2 ): 156 - 161 .

FERRARI R , TRINCI M , CASINELLI A , et al . Radiomics in radiology: what the radiologist needs to know about technical aspects and clinical impact [J ] . Radiol Med , 2024 , 129 ( 12 ): 1751 - 1765 .

JOO B , AHN S S , AN C , et al . Fully automated radiomics-based machine learning models for multiclass classification of single brain tumors: glioblastoma, lymphoma, and metastasis [J ] . J Neuroradiol , 2023 , 50 ( 4 ): 388 - 395 .

LIU J P , TU J Q , XU L H , et al . MRI-based radiomics signatures for preoperative prediction of Ki-67 index in primary central nervous system lymphoma [J ] . Eur J Radiol , 2024 , 178 : 111603 .

LIU G L , ZHANG X Y , ZHANG N , et al . Detecting double expression status in primary central nervous system lymphoma using multiparametric MRI based machine learning [J ] . J Magn Reson Imaging , 2024 , 59 ( 1 ): 231 - 239 .

NASER P V , MAURER M C , FISCHER M , et al . Deep learning aided preoperative diagnosis of primary central nervous system lymphoma [J ] . iScience , 2024 , 27 ( 2 ): 109023 .

LIU X , DU P , DAI Z G , et al . SRTRP-Net: a multi-task learning network for segmentation and prediction of stereotactic radiosurgery treatment response in brain metastases [J ] . Comput Biol Med , 2024 , 175 : 108503 .

GARABA A , ASLAM N , PONZIO F , et al . Radiomics for differentiation of gliomas from primary central nervous system lymphomas: a systematic review and meta-analysis [J ] . Front Oncol , 2024 , 14 : 1291861 .

ZHAO L M , HU R , XIE F F , et al . Radiomic-based MRI for classification of solitary brain metastases subtypes from primary lymphoma of the central nervous system [J ] . J Magn Reson Imaging , 2023 , 57 ( 1 ): 227 - 235 .

CHALLIS E , HURLEY P , SERRA L , et al . Gaussian process classification of Alzheimer’s disease and mild cognitive impairment from resting-state fMRI [J ] . Neuroimage , 2015 , 112 : 232 - 243 .

PONTILLO G , TOMMASIN S , CUOCOLO R , et al . A combined radiomics and machine learning approach to overcome the clinicoradiologic paradox in multiple sclerosis [J ] . AJNR Am J Neuroradiol , 2021 , 42 ( 11 ): 1927 - 1933 .

OZTURK K , SOYLU E , CAYCI Z . Differentiation between primary CNS lymphoma and atypical glioblastoma according to major genomic alterations using diffusion and susceptibility-weighted MR imaging [J ] . Eur J Radiol , 2021 , 141 : 109784 .

ZHANG Y , NING C Y , YANG W J . An automatic cervical cell classification model based on improved DenseNet121 [J ] . Sci Rep , 2025 , 15 ( 1 ): 3240 .

MAO Y N , KIM J , PODINA L , et al . Dilated SE-DenseNet for brain tumor MRI classification [J ] . Sci Rep , 2025 , 15 ( 1 ): 3596 .

Views

2018

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Research progress in predicting the risk of lymphatic or hematologic metastasis based on chest CT in early non-small cell lung cancer

Research progress and prospects of MRI in predicting lymph node status in breast cancer

A cohort study of prognostic value of ¹⁸F-FDG PET/CT metabolic parameters in patients with diffuse large B-cell lymphoma treated with CAR-T

Artificial intelligence in gastric cancer diagnosis, treatment and prognostic prediction: current application and future perspective

A CT-based radiomics nomogram for predicting local tumor progression of colorectal cancer lung metastases treated with radiofrequency ablation

Related Author

Yuanyuan FU

Runping HOU

Xiaolong FU

Sheng CHEN

Zihan ZHAI

Chao HE

Yeye ZHOU

Bin ZHANG

Related Institution

School of Biomedical Engineering, Shanghai Jiao Tong University

Department of Radiotherapy, Shanghai Chest Hospital, Shanghai Jiao Tong University

Department of Breast Surgery, Fudan University Shanghai Cancer Center, Department of Oncology, Shanghai Medical College, Fudan University

Department of Nuclear Medicine, the First Affiliated Hospital of Soochow University

Department of Oncology, Xinjiang Military Region General Hospital of the Chinese People’s Liberation Army, Urumqi 830000, Xinjiang Uygur Autonomous Region

AI问答

⁰