Diagnosis and Subtyping of Autoimmune Encephalitis Using an Attention-Based Multi-Instance Learning Model: A Multi-Center 18F-FDG PET Study

IF 5 1区医学 Q1 NEUROSCIENCES

CNS Neuroscience & Therapeutics Pub Date : 2025-08-04 DOI:10.1111/cns.70513

Yueqian Sun, Ruizhe Sun, Jiahua Lv, Qingxia Kong, Cixiang Dai, Bin Wang, Xiong Han, Min Chen, Ruihan Liu, Yan Jiang, Leilei Yuan, Lin Ai, Xiaodong Yang, Yiqiang Chen, Qun Wang

{"title":"Diagnosis and Subtyping of Autoimmune Encephalitis Using an Attention-Based Multi-Instance Learning Model: A Multi-Center 18F-FDG PET Study","authors":"Yueqian Sun, Ruizhe Sun, Jiahua Lv, Qingxia Kong, Cixiang Dai, Bin Wang, Xiong Han, Min Chen, Ruihan Liu, Yan Jiang, Leilei Yuan, Lin Ai, Xiaodong Yang, Yiqiang Chen, Qun Wang","doi":"10.1111/cns.70513","DOIUrl":null,"url":null,"abstract":"<div>\n \n \n <section>\n \n <h3> Background</h3>\n \n <p>The aim was to develop an attention-based model using <sup>18</sup>F-fluorodeoxyglucose (<sup>18</sup>F-FDG) PET imaging to differentiate autoimmune encephalitis (AE) patients from controls and to discriminate among different AE subtypes.</p>\n </section>\n \n <section>\n \n <h3> Methods</h3>\n \n <p>This multi-center retrospective study enrolled 390 participants: 222 definite AE patients (comprising four subtypes: LGI1-AE, NMDAR-AE, GABAB-AE, GAD65-AE), 122 age- and sex-matched healthy controls, and 33 age- and sex-matched antibody-negative AE patients along with 13 age- and sex-matched viral encephalitis patients, both serving as disease controls. An attention-based multi-instance learning (MIL) model was trained using data from one hospital and underwent external validation with data from other institutions. Additionally, a multi-modal MIL (m-MIL) model integrating imaging features, age, and sex parameters was evaluated alongside logistic regression (LR) and random forest (RF) models for comparative analysis.</p>\n </section>\n \n <section>\n \n <h3> Results</h3>\n \n <p>The attention-based m-MIL model outperformed classical algorithms (LR, RF) and single-modal MIL in AE vs. all controls binary classification, achieving the highest accuracy (84.00% internal, 67.38% external) and sensitivity (90.91% internal, 71.19% external). For multiclass AE subtype classification, the MIL-based model achieved 95.05% (internal) and 77.97% (external) accuracy. Heatmap analysis revealed that NMDAR-AE involved broader brain regions, including the medial temporal lobe (MTL) and basal ganglia (BG), whereas LGI1-AE and GABAB-AE showed focal attention on the MTL and BG. In contrast, GAD65-AE demonstrated concentrated attention exclusively in the MTL.</p>\n </section>\n \n <section>\n \n <h3> Conclusion</h3>\n \n <p>The m-MIL model effectively discriminates AE patients from controls and enables subtyping of different AE subtypes, offering a valuable diagnostic tool for the clinical assessment and classification of AE.</p>\n </section>\n </div>","PeriodicalId":154,"journal":{"name":"CNS Neuroscience & Therapeutics","volume":"31 8","pages":""},"PeriodicalIF":5.0000,"publicationDate":"2025-08-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://onlinelibrary.wiley.com/doi/epdf/10.1111/cns.70513","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"CNS Neuroscience & Therapeutics","FirstCategoryId":"3","ListUrlMain":"https://onlinelibrary.wiley.com/doi/10.1111/cns.70513","RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"NEUROSCIENCES","Score":null,"Total":0}

引用次数: 0

Abstract

Background

The aim was to develop an attention-based model using ¹⁸F-fluorodeoxyglucose (¹⁸F-FDG) PET imaging to differentiate autoimmune encephalitis (AE) patients from controls and to discriminate among different AE subtypes.

Methods

This multi-center retrospective study enrolled 390 participants: 222 definite AE patients (comprising four subtypes: LGI1-AE, NMDAR-AE, GABAB-AE, GAD65-AE), 122 age- and sex-matched healthy controls, and 33 age- and sex-matched antibody-negative AE patients along with 13 age- and sex-matched viral encephalitis patients, both serving as disease controls. An attention-based multi-instance learning (MIL) model was trained using data from one hospital and underwent external validation with data from other institutions. Additionally, a multi-modal MIL (m-MIL) model integrating imaging features, age, and sex parameters was evaluated alongside logistic regression (LR) and random forest (RF) models for comparative analysis.

Results

The attention-based m-MIL model outperformed classical algorithms (LR, RF) and single-modal MIL in AE vs. all controls binary classification, achieving the highest accuracy (84.00% internal, 67.38% external) and sensitivity (90.91% internal, 71.19% external). For multiclass AE subtype classification, the MIL-based model achieved 95.05% (internal) and 77.97% (external) accuracy. Heatmap analysis revealed that NMDAR-AE involved broader brain regions, including the medial temporal lobe (MTL) and basal ganglia (BG), whereas LGI1-AE and GABAB-AE showed focal attention on the MTL and BG. In contrast, GAD65-AE demonstrated concentrated attention exclusively in the MTL.

Conclusion

The m-MIL model effectively discriminates AE patients from controls and enables subtyping of different AE subtypes, offering a valuable diagnostic tool for the clinical assessment and classification of AE.

Abstract Image

查看原文本刊更多论文

使用基于注意的多实例学习模型诊断和分型自身免疫性脑炎：一项多中心18F-FDG PET研究

目的是建立一个基于注意力的模型，利用18f -氟脱氧葡萄糖（18F-FDG） PET成像来区分自身免疫性脑炎（AE）患者和对照组，并区分不同的AE亚型。方法本多中心回顾性研究共纳入390名受试者：222例确诊AE患者（包括LGI1-AE、NMDAR-AE、GABAB-AE、GAD65-AE四种亚型），122例年龄和性别匹配的健康对照，33例年龄和性别匹配的抗体阴性AE患者以及13例年龄和性别匹配的病毒性脑炎患者作为疾病对照。使用来自一家医院的数据训练了基于注意的多实例学习（MIL）模型，并使用来自其他机构的数据进行了外部验证。此外，结合影像特征、年龄和性别参数的多模态MIL （m-MIL）模型与逻辑回归（LR）和随机森林（RF）模型进行比较分析。结果基于注意力的m-MIL模型在AE分类中优于经典算法（LR、RF）和单模态MIL，准确率最高（内部84.00%，外部67.38%），灵敏度最高（内部90.91%，外部71.19%）。对于多类AE亚型分类，基于mil的模型准确率达到95.05%（内部）和77.97%（外部）。热图分析显示，NMDAR-AE累及更广泛的脑区，包括内侧颞叶（MTL）和基底神经节（BG），而LGI1-AE和GABAB-AE主要累及内侧颞叶和基底神经节。相反，GAD65-AE仅在MTL中表现出集中的注意力。结论m-MIL模型能有效区分AE患者和对照组，并能对AE的不同亚型进行分型，为AE的临床评估和分型提供了有价值的诊断工具。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文求助全文

来源期刊

CNS Neuroscience & Therapeutics 医学-神经科学

CiteScore

7.30

自引率

12.70%

发文量

240

审稿时长

2 months

期刊介绍： CNS Neuroscience & Therapeutics provides a medium for rapid publication of original clinical, experimental, and translational research papers, timely reviews and reports of novel findings of therapeutic relevance to the central nervous system, as well as papers related to clinical pharmacology, drug development and novel methodologies for drug evaluation. The journal focuses on neurological and psychiatric diseases such as stroke, Parkinson’s disease, Alzheimer’s disease, depression, schizophrenia, epilepsy, and drug abuse.