{"title":"基于变压器的磁共振诊断阿尔茨海默病混合网络","authors":"Zhentao Hu , Yihan Wang , Yanyang Li","doi":"10.1016/j.dsp.2025.105562","DOIUrl":null,"url":null,"abstract":"<div><div>Treating Alzheimer's disease (AD) is currently considered highly challenging among various neurodegenerative diseases. The precision of AD diagnosis can be confounded by multiple factors. Magnetic resonance imaging (MRI) is a critical tool for diagnosing AD. To assist physicians in clinical diagnosis, a new hybrid model, CTM-Net, is proposed based on MRI. CTM-Net incorporates a CNN enhanced by a channel attention mechanism to extract local fine-grained features from MRI slices, which are then mapped into high-level representations. Subsequently, the model integrates a Transformer's multi-head attention mechanism to capture long-range dependencies across MRI slices. The local continuity of features between two adjacent MRI slices is enhanced using a one-dimensional convolution operation, which gradually fuse spatially adjacent features to ultimately obtain global MRI information. CTM-Net was validated on the ADNI dataset. It achieved 92.70%, 83.00%, and 79.07% accuracy on the three classification tasks of AD vs. CN, AD vs. MCI, and MCI vs. CN, respectively. Compared to other models applied to AD classification tasks, the proposed model yielded superior results in terms of accuracy. CTM-Net is a Convolution-Transformer hybrid model for AD classification and diagnosis tasks, which can combine the advantages of the CNN and attention mechanism to make the most of interactive information between local lesion features and global context features for improving diagnosis efficiency.</div></div>","PeriodicalId":51011,"journal":{"name":"Digital Signal Processing","volume":"168 ","pages":"Article 105562"},"PeriodicalIF":3.0000,"publicationDate":"2025-09-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"A transformer-based hybrid network for Alzheimer's disease diagnosis via MRI\",\"authors\":\"Zhentao Hu , Yihan Wang , Yanyang Li\",\"doi\":\"10.1016/j.dsp.2025.105562\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<div><div>Treating Alzheimer's disease (AD) is currently considered highly challenging among various neurodegenerative diseases. The precision of AD diagnosis can be confounded by multiple factors. Magnetic resonance imaging (MRI) is a critical tool for diagnosing AD. To assist physicians in clinical diagnosis, a new hybrid model, CTM-Net, is proposed based on MRI. CTM-Net incorporates a CNN enhanced by a channel attention mechanism to extract local fine-grained features from MRI slices, which are then mapped into high-level representations. Subsequently, the model integrates a Transformer's multi-head attention mechanism to capture long-range dependencies across MRI slices. The local continuity of features between two adjacent MRI slices is enhanced using a one-dimensional convolution operation, which gradually fuse spatially adjacent features to ultimately obtain global MRI information. CTM-Net was validated on the ADNI dataset. It achieved 92.70%, 83.00%, and 79.07% accuracy on the three classification tasks of AD vs. CN, AD vs. MCI, and MCI vs. CN, respectively. Compared to other models applied to AD classification tasks, the proposed model yielded superior results in terms of accuracy. CTM-Net is a Convolution-Transformer hybrid model for AD classification and diagnosis tasks, which can combine the advantages of the CNN and attention mechanism to make the most of interactive information between local lesion features and global context features for improving diagnosis efficiency.</div></div>\",\"PeriodicalId\":51011,\"journal\":{\"name\":\"Digital Signal Processing\",\"volume\":\"168 \",\"pages\":\"Article 105562\"},\"PeriodicalIF\":3.0000,\"publicationDate\":\"2025-09-02\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Digital Signal Processing\",\"FirstCategoryId\":\"5\",\"ListUrlMain\":\"https://www.sciencedirect.com/science/article/pii/S1051200425005846\",\"RegionNum\":3,\"RegionCategory\":\"工程技术\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q2\",\"JCRName\":\"ENGINEERING, ELECTRICAL & ELECTRONIC\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Digital Signal Processing","FirstCategoryId":"5","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S1051200425005846","RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"ENGINEERING, ELECTRICAL & ELECTRONIC","Score":null,"Total":0}
引用次数: 0
摘要
治疗阿尔茨海默病(AD)目前被认为是各种神经退行性疾病中极具挑战性的。阿尔茨海默病的诊断精度可能受到多种因素的影响。磁共振成像(MRI)是诊断AD的重要工具。为了帮助医生进行临床诊断,提出了一种基于MRI的混合模型CTM-Net。CTM-Net结合了一个由通道注意机制增强的CNN,从MRI切片中提取局部细粒度特征,然后将其映射为高级表示。随后,该模型集成了Transformer的多头注意机制,以捕获跨MRI切片的远程依赖关系。通过一维卷积运算增强两个相邻MRI切片之间特征的局部连续性,逐步融合空间相邻特征,最终获得全局MRI信息。CTM-Net在ADNI数据集上进行了验证。在AD vs. CN、AD vs. MCI、MCI vs. CN三个分类任务上,准确率分别达到了92.70%、83.00%、79.07%。与应用于AD分类任务的其他模型相比,本文提出的模型在准确率方面取得了更好的结果。CTM-Net是一种用于AD分类和诊断任务的卷积-变压器混合模型,它可以结合CNN和注意机制的优点,最大限度地利用局部病变特征和全局上下文特征之间的交互信息,提高诊断效率。
A transformer-based hybrid network for Alzheimer's disease diagnosis via MRI
Treating Alzheimer's disease (AD) is currently considered highly challenging among various neurodegenerative diseases. The precision of AD diagnosis can be confounded by multiple factors. Magnetic resonance imaging (MRI) is a critical tool for diagnosing AD. To assist physicians in clinical diagnosis, a new hybrid model, CTM-Net, is proposed based on MRI. CTM-Net incorporates a CNN enhanced by a channel attention mechanism to extract local fine-grained features from MRI slices, which are then mapped into high-level representations. Subsequently, the model integrates a Transformer's multi-head attention mechanism to capture long-range dependencies across MRI slices. The local continuity of features between two adjacent MRI slices is enhanced using a one-dimensional convolution operation, which gradually fuse spatially adjacent features to ultimately obtain global MRI information. CTM-Net was validated on the ADNI dataset. It achieved 92.70%, 83.00%, and 79.07% accuracy on the three classification tasks of AD vs. CN, AD vs. MCI, and MCI vs. CN, respectively. Compared to other models applied to AD classification tasks, the proposed model yielded superior results in terms of accuracy. CTM-Net is a Convolution-Transformer hybrid model for AD classification and diagnosis tasks, which can combine the advantages of the CNN and attention mechanism to make the most of interactive information between local lesion features and global context features for improving diagnosis efficiency.
期刊介绍:
Digital Signal Processing: A Review Journal is one of the oldest and most established journals in the field of signal processing yet it aims to be the most innovative. The Journal invites top quality research articles at the frontiers of research in all aspects of signal processing. Our objective is to provide a platform for the publication of ground-breaking research in signal processing with both academic and industrial appeal.
The journal has a special emphasis on statistical signal processing methodology such as Bayesian signal processing, and encourages articles on emerging applications of signal processing such as:
• big data• machine learning• internet of things• information security• systems biology and computational biology,• financial time series analysis,• autonomous vehicles,• quantum computing,• neuromorphic engineering,• human-computer interaction and intelligent user interfaces,• environmental signal processing,• geophysical signal processing including seismic signal processing,• chemioinformatics and bioinformatics,• audio, visual and performance arts,• disaster management and prevention,• renewable energy,