PDF(10102 KB)
Preliminary study on multimodal magnetic resonance imaging for Alzheimer's disease based on machine learning
Nuerbiya·Keranmu, Jun LIU, Yushanjiang·Niyazi, Ying LIU
Chinese Journal of Alzheimer's Disease and Related Disorders ›› 2026, Vol. 9 ›› Issue (3) : 186-195.
PDF(10102 KB)
Abbreviation (ISO4): Chinese Journal of Alzheimer's Disease and Related Disorders
Editor in chief: Jun WANG
PDF(10102 KB)
Preliminary study on multimodal magnetic resonance imaging for Alzheimer's disease based on machine learning
Objective: Based on interpretable machine learning methods, to explore the application value of multimodal MRI radiomics in the diagnosis of Alzheimer's disease (AD) and provide imaging tools for accurate clinical diagnosis of AD. Methods: A retrospective study was conducted on 110 subjects, including 48 in the AD group and 62 in the control group (HC), all of whom completed 3D-T1WI, DWI, and T2WI MRI sequence scans. Randomly stratified sampling was conducted at a ratio of 7:3 to divide the data into training and testing sets, and 8 AD related core brain regions (hippocampus, entorhinal cortex, etc.) were segmented. Extract 107 radiomics features, screen the core features, and construct 8 diagnostic models (2 algorithms x 3 sequences, 2 algorithms x 2 sequences combined) based on logistic regression (LR) and random forest (RF) algorithms. Evaluate the model performance using area under the curve (AUC), and analyze the model interpretability using SHAP analysis. Results: 16 core radiomics features were ultimately selected. The joint sequence model has the best diagnostic performance, with AUC values of 0.989 (95% CI: 0.960~1.00) and 0.970 (95% CI: 0.920~1.00) for the LR and RF algorithm test sets, respectively, which are significantly higher than those of the single sequence model; The accuracy, sensitivity, and specificity of the LR joint model test set were 0.882, 0.800, and 0.947, respectively. SHAP analysis shows that the 3D-T1WI sequence with short run length and high grayscale emphasized features, and the DWI sequence with grayscale co-occurrence matrix information measurement are the core indicators for AD diagnosis. Conclusions: The multimodal MRI radiomics model can efficiently achieve AD diagnosis, and the LR combined model has the best comprehensive performance. SHAP analysis can clearly analyze the decision-making basis of the model, providing strong support for the clinical translation of the model and accurate diagnosis of AD.
Alzheimer's disease / Magnetic resonance imaging / Machine learning / Explainability analysis
| [1] |
王英全, 梁景宏, 贾瑞霞, 等. 2020-2050年中国阿尔茨海默病患病情况预测研究[J]. 阿尔茨海默病及相关病杂志, 2019, 2(1):289-298.
目的: 对中国未来老年人群阿尔茨海默病（AD）患病情况进行预测，为我国相关公共卫生政策的制定和实施提供参考。方法: 检索国内外数据库，对符合纳入标准的AD流行病学研究结果进行meta分析，计算过去30年不同时期老年人AD合并患病率。以2015年全国1%人口普查数据为基础，采用年龄移算法预测我国未来60岁以上老年人口数。将近年（2015-2018）全国不同年龄组AD合并患病率与未来人口相结合，预测我国2050年老年人AD患病情况。结果: 我国1985-2018年AD合并患病率为3.9%(95%CI：3.4-4.4)，其中2015-2018年合并患病率为6.6%(95%CI：4.7-8.6)。2020年、2030年、2040年60岁以上老年人患病人数分别为1450万、2075万、2687万。2050年患病人数为3003万，其中年龄组60~69岁、70~79岁、≥80岁患病人数分别为614万、907万、1482万人。2050年AD患病人数为2015年的2.35倍。结论: 如果没有相应的有效预防措施，中国AD患病人数在未来30年将大幅增长。
|
| [2] |
徐勇, 王军, 王虹峥, 等. 2023中国阿尔茨海默病数据与防控策略[J]. 阿尔茨海默病及相关病杂志, 2023, 6(3):175-192,173.
本报告汇总了阿尔茨海默病及痴呆相关领域现有的最新数据， 分析了我国阿尔茨海默病的流行病学、疾病负担、诊断治疗、风险因素、康复护理和疾病筛查等各方面的现状、问题以及趋势，针对目前中国民众和政策制定者最关心的问题，例如：阿尔茨海默病基本数据是怎样的?为什么知晓率相对提高，但是患者就诊愿望相对不高?如何采用有效的措施来应对，等等，提出了目前阿尔茨海默病的防控策略。特别强调了要大力开发经过验证的早期筛查和诊断工具，提供更有效治疗的创新药物，以及鼓励和建立全国性属地化社会互助支持网络。本报告期望能对医学专业人士、患者、家属和照护者、政府政策制定人员、养老机构等有所帮助，能为阿尔茨海默病的防治工作提供支撑，为相关卫生政策的制定提供依据和建议，有助于提高公众对阿尔茨海默病的认识，以期缓解我国阿尔茨海默病疾病的整体负担，推动我国健康老龄化的实现。
|
| [3] |
Alzheimer's disease (AD) is pathologically defined by extracellular accumulation of amyloid-β (Aβ) peptides generated by the cleavage of amyloid precursor protein (APP), strings of hyperphosphorylated Tau proteins accumulating inside neurons known as neurofibrillary tangles (NFTs) and neuronal loss. The association between the two hallmarks and cognitive decline has been known since the beginning of the 20th century when the first description of the disease was carried out by Alois Alzheimer. Today, more than 40 million people worldwide are affected by AD that represents the most common cause of dementia and there is still no effective treatment available to cure the disease. In general, the aggregation of Aβ is considered an essential trigger in AD pathogenesis that gives rise to NFTs, neuronal dysfunction and dementia. During the process leading to AD, tau and Aβ first misfold and form aggregates in one brain region, from where they spread to interconnected areas of the brain thereby inducing its gradual morphological and functional deterioration. In this mini-review article, we present an overview of the current literature on the spreading mechanisms of Aβ and tau pathology in AD since a more profound understanding is necessary to design therapeutic approaches aimed at preventing or halting disease progression.Copyright © 2020 d’Errico and Meyer-Luehmann.
|
| [4] |
|
| [5] |
Alzheimer's disease is the leading cause of dementia. However, neither Alzheimer's disease nor Alzheimer's dementia are an inevitable consequence of aging. This review provides an overview of the issues involved in a diagnosis of Alzheimer's disease before an individual meets the criteria for Alzheimer's dementia. It examines how Alzheimer's disease diagnosis rates can be improved, the implications of an early diagnosis for the individual, carer and society, and the importance of risk reduction to prevent or delay progression. Although no disease-modifying agents capable of reversing the initial pathological changes are currently available, it may be possible to prevent or delay the development of dementia in a proportion of the population by modifying exposure to common risk factors. In other individuals, diagnosing the disease or risk of disease early is still valuable so that the individual and their carers have time to make choices and plan for the future, and to allow access to treatments that can help manage symptoms. Primary healthcare professionals play a pivotal role in recognising individuals at risk, recommending lifestyle changes in mid-adult life that can prevent or slow down the disease, and in timely diagnosis. Early intervention is the optimal strategy, because the patient's level of function is preserved for longer.© 2019 Rasmussen and Langerman.
|
| [6] |
Alzheimer’s disease (AD) is the most common neurodegenerative disease and its diagnosis has classically been based on clinical symptoms. Recently, a biological rather than a syndromic definition of the disease has been proposed that is based on biomarkers that reflect neuropathological changes. In AD, there are two main biomarker categories, namely neuroimaging and fluid biomarkers [cerebrospinal fluid (CSF) and blood]. As a complex and multifactorial disease, AD biomarkers are important for an accurate diagnosis and to stage the disease, assess the prognosis, test target engagement, and measure the response to treatment. In addition, biomarkers provide us with information that, even if it does not have a current clinical use, helps us to understand the mechanisms of the disease. In addition to the pathological hallmarks of AD, which include amyloid-β and tau deposition, there are multiple concomitant pathological events that play a key role in the disease. These include, but are not limited to, neurodegeneration, inflammation, vascular dysregulation or synaptic dysfunction. In addition, AD patients often have an accumulation of other proteins including α-synuclein and TDP-43, which may have a pathogenic effect on AD. In combination, there is a need to have biomarkers that reflect different aspects of AD pathogenesis and this will be important in the future to establish what are the most suitable applications for each of these AD-related biomarkers. It is unclear whether sex, gender, or both have an effect on the causes of AD. There may be differences in fluid biomarkers due to sex but this issue has often been neglected and warrants further research. In this review, we summarize the current state of the principal AD fluid biomarkers and discuss the effect of sex on these biomarkers.
|
| [7] |
To compare the diagnostic accuracy of CSF biomarkers and amyloid PET for diagnosing early-stage Alzheimer disease (AD).From the prospective, longitudinal BioFINDER study, we included 122 healthy elderly and 34 patients with mild cognitive impairment who developed AD dementia within 3 years (MCI-AD). β-Amyloid (Aβ) deposition in 9 brain regions was examined with [18F]-flutemetamol PET. CSF was analyzed with INNOTEST and EUROIMMUN ELISAs. The results were replicated in 146 controls and 64 patients with MCI-AD from the Alzheimer's Disease Neuroimaging Initiative study.The best CSF measures for identifying MCI-AD were Aβ42/total tau (t-tau) and Aβ42/hyperphosphorylated tau (p-tau) (area under the curve [AUC] 0.93-0.94). The best PET measures performed similarly (AUC 0.92-0.93; anterior cingulate, posterior cingulate/precuneus, and global neocortical uptake). CSF Aβ42/t-tau and Aβ42/p-tau performed better than CSF Aβ42 and Aβ42/40 (AUC difference 0.03-0.12, p<0.05). Using nonoptimized cutoffs, CSF Aβ42/t-tau had the highest accuracy of all CSF/PET biomarkers (sensitivity 97%, specificity 83%). The combination of CSF and PET was not better than using either biomarker separately.Amyloid PET and CSF biomarkers can identify early AD with high accuracy. There were no differences between the best CSF and PET measures and no improvement when combining them. Regional PET measures were not better than assessing the global Aβ deposition. The results were replicated in an independent cohort using another CSF assay and PET tracer. The choice between CSF and amyloid PET biomarkers for identifying early AD can be based on availability, costs, and doctor/patient preferences since both have equally high diagnostic accuracy.This study provides Class III evidence that amyloid PET and CSF biomarkers identify early-stage AD equally accurately.© 2015 American Academy of Neurology.
|
| [8] |
杨慧娟, 刘顺英, 苏晶, 等. 人工智能在阿尔茨海默病全程管理中的应用进展[J]. 实用老年医学, 2025, 39(8):768-772.
|
| [9] |
金贺, 王蓉. 基于多组学技术的阿尔茨海默病研究进展[J]. 神经疾病与精神卫生, 2025, 25(9):609-615.
|
| [10] |
|
| [11] |
Alzheimer’s disease (AD) poses a significant challenge for individuals aged 65 and older, being the most prevalent form of dementia. Although existing AD risk prediction tools demonstrate high accuracy, their complexity and limited accessibility restrict practical application. This study aimed to develop a convenience, efficient prediction model for AD risk using machine learning techniques.
|
| [12] |
Alzheimer’s disease (AD) is a progressive neurodegenerative condition that impairs memory and cognition, presenting a growing global healthcare burden. Despite major research efforts, no cure exists, and treatments remain focused on symptom relief. This narrative review highlights recent advancements in artificial intelligence (AI), particularly machine learning (ML) and deep learning (DL), which enhance early diagnosis, predict disease progression, and support personalized treatment strategies. AI applications are reshaping healthcare by enabling early detection, predicting disease progression, and developing personalized treatment plans. In particular, AI’s ability to analyze complex datasets, including genetic and imaging data, has shown promise in identifying early biomarkers of AD. Additionally, AI-driven cognitive training and rehabilitation programs are emerging as effective tools to improve cognitive function and slow down the progression of cognitive impairment. The paper also discusses the potential of AI in drug discovery and clinical trial optimization, offering new avenues for the development of AD treatments. The paper emphasizes the need for ongoing interdisciplinary collaboration and regulatory oversight to harness AI’s full potential in transforming AD care and improving patient outcomes.
|
| [13] |
|
| [14] |
|
| [15] |
Radiomics, the high-throughput mining of quantitative image features from standard-of-care medical imaging that enables data to be extracted and applied within clinical-decision support systems to improve diagnostic, prognostic, and predictive accuracy, is gaining importance in cancer research. Radiomic analysis exploits sophisticated image analysis tools and the rapid development and validation of medical imaging data that uses image-based signatures for precision diagnosis and treatment, providing a powerful tool in modern medicine. Herein, we describe the process of radiomics, its pitfalls, challenges, opportunities, and its capacity to improve clinical decision making, emphasizing the utility for patients with cancer. Currently, the field of radiomics lacks standardized evaluation of both the scientific integrity and the clinical relevance of the numerous published radiomics investigations resulting from the rapid growth of this area. Rigorous evaluation criteria and reporting guidelines need to be established in order for radiomics to mature as a discipline. Herein, we provide guidance for investigations to meet this urgent need in the field of radiomics.
|
| [16] |
Radiomics is a quantitative approach to medical imaging, which aims at enhancing the existing data available to clinicians by means of advanced mathematical analysis. Through mathematical extraction of the spatial distribution of signal intensities and pixel interrelationships, radiomics quantifies textural information by using analysis methods from the field of artificial intelligence. Various studies from different fields in imaging have been published so far, highlighting the potential of radiomics to enhance clinical decision-making. However, the field faces several important challenges, which are mainly caused by the various technical factors influencing the extracted radiomic features.
|
| [17] |
|
| [18] |
|
| [19] |
With the increase in large multimodal cohorts and high-throughput technologies, the potential for discovering novel biomarkers is no longer limited by data set size. Artificial intelligence (AI) and machine learning approaches have been developed to detect novel biomarkers and interactions in complex data sets. We discuss exemplar uses and evaluate current applications and limitations of AI to discover novel biomarkers. Remaining challenges include a lack of diversity in the data sets available, the sheer complexity of investigating interactions, the invasiveness and cost of some biomarkers, and poor reporting in some studies. Overcoming these challenges will involve collecting data from underrepresented populations, developing more powerful AI approaches, validating the use of noninvasive biomarkers, and adhering to reporting guidelines. By harnessing rich multimodal data through AI approaches and international collaborative innovation, we are well positioned to identify clinically useful biomarkers that are accurate, generalizable, unbiased, and acceptable in clinical practice. HIGHLIGHTS: Artificial intelligence and machine learning approaches may accelerate dementia biomarker discovery. Remaining challenges include data set suitability due to size and bias in cohort selection. Multimodal data, diverse data sets, improved machine learning approaches, real-world validation, and interdisciplinary collaboration are required.© 2023 The Authors. Alzheimer's & Dementia published by Wiley Periodicals LLC on behalf of Alzheimer's Association.
|
| [20] |
|
| [21] |
|
| [22] |
To develop a computed tomography (CT) radiomics-based interpretable machine learning (ML) model to predict the pathological grade of pancreatic neuroendocrine tumors (pNETs) in a non-invasive manner.
|
| [23] |
To develop a computed tomography (CT) radiomics-based interpretable machine learning (ML) model to preoperatively predict human epidermal growth factor receptor 2 (HER2) status in bladder cancer (BCa) with multicenter validation.
|
| [24] |
van der Velden B H M,
|
| [25] |
Alzheimer’s disease (AD) is a neurological disease that affects numerous people. The condition causes brain atrophy, which leads to memory loss, cognitive impairment, and death. In its early stages, Alzheimer’s disease is tricky to predict. Therefore, treatment provided at an early stage of AD is more effective and causes less damage than treatment at a later stage. Although AD is a common brain condition, it is difficult to recognize, and its classification requires a discriminative feature representation to separate similar brain patterns. Multimodal neuroimage information that combines multiple medical images can classify and diagnose AD more accurately and comprehensively. Magnetic resonance imaging (MRI) has been used for decades to assist physicians in diagnosing Alzheimer’s disease. Deep models have detected AD with high accuracy in computing-assisted imaging and diagnosis by minimizing the need for hand-crafted feature extraction from MRI images. This study proposes a multimodal image fusion method to fuse MRI neuroimages with a modular set of image preprocessing procedures to automatically fuse and convert Alzheimer’s disease neuroimaging initiative (ADNI) into the BIDS standard for classifying different MRI data of Alzheimer’s subjects from normal controls. Furthermore, a 3D convolutional neural network is used to learn generic features by capturing AlD biomarkers in the fused images, resulting in richer multimodal feature information. Finally, a conventional CNN with three classifiers, including Softmax, SVM, and RF, forecasts and classifies the extracted Alzheimer’s brain multimodal traits from a normal healthy brain. The findings reveal that the proposed method can efficiently predict AD progression by combining high-dimensional MRI characteristics from different public sources with an accuracy range from 88.7% to 99% and outperforming baseline models when applied to MRI-derived voxel features.
|
| [26] |
Recent technological advancements in data acquisition tools allowed neuroscientists to acquire different modality data to diagnosis Alzheimer's disease (AD). However, how to fuse these enormous amount different modality data to improve recognizing rate and find significance brain regions is still challenging.The algorithm used multimodal medical images [structural magnetic resonance imaging (sMRI) and positron emission tomography (PET)] as experimental data. Deep feature representations of sMRI and PET images are extracted by 3D convolution neural network (3DCNN). An improved Transformer is then used to progressively learn global correlation information among features. Finally, the information from different modalities is fused for identification. A model-based visualization method is used to explain the decisions of the model and identify brain regions related to AD.The model attained a noteworthy classification accuracy of 98.1% for Alzheimer's disease (AD) using the Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset. Upon examining the visualization results, distinct brain regions associated with AD diagnosis were observed across different image modalities. Notably, the left parahippocampal region emerged consistently as a prominent and significant brain area.A large number of comparative experiments have been carried out for the model, and the experimental results verify the reliability of the model. In addition, the model adopts a visualization analysis method based on the characteristics of the model, which improves the interpretability of the model. Some disease-related brain regions were found in the visualization results, which provides reliable information for AD clinical research.© 2024. The Author(s).
|
| [27] |
|
| [28] |
Background/Objectives: Conventional logistic regression is widely used in the field of dentistry, specifically for prediction purposes in longitudinal studies. This study aimed to compare the validity of different supervised machine learning (ML) models to the conventional logistic regression (LR) model to predict the outcomes of nonsurgical periodontal treatment (NSPT). Methods: Patients diagnosed with periodontitis received full periodontal charting, including bleeding on probing (BoP), probing pocket depth (PPD), and clinical attachment loss (CAL). Furthermore, the tooth type, tooth location, tooth surface, arch type, and gingival phenotype were also collected as site-specific predictors. Later, root surface debridement was provided and treatment outcomes were evaluated after 3 months. Site-specific predictors were used to train five ML models, including random forest (RF), decision tree (DT), support vector classifier (SVC), K-nearest neighbors (KNN), and Gaussian naïve Bayes (GNB), to develop predictive models. Results: Site-specific predictors of 1108 examined sites were used, and the overall accuracy prediction of the conventional LR model was 70.4%, with PPD statistically significantly associated with the outcome of NSPT (odds ratio = 0.577, p = 0.001). Among the ML models examined, only GNB and SVC showed comparable prediction accuracy (71.0% and 70.4%, respectively) to the LR model, whereas the prediction accuracies of KNN, RF, and DT were 65.0%, 62.0%, and 61.0%, respectively. Similarly, baseline PPD was shown to be the most important featured predictor by both the RF and DT models. Conclusions: The evidence suggests that supervised ML models do not outperform the LR model in predicting the outcomes of NSPT. A larger sample size and more predictors of periodontitis are necessary to enhance the accuracy of ML models over the LR model in predicting the outcomes of NSPT.
|
| [29] |
|
| [30] |
To investigate the potential of T1-weighted imaging (T1WI)-based hippocampal radiomics as imaging markers for the diagnosis of Alzheimer's disease (AD) and their efficacy in discriminating between mild cognitive impairment (MCI) and dementia in AD.A total of 126 AD patients underwent T1WI-based magnetic resonance imaging (MRI) examinations, along with 108 age-sex-matched healthy controls (HC). This was a retrospective, single-center study conducted from November 2021 to February 2023. AD patients were categorized into two groups based on disease progression and cognitive function: AD-MCI and dementia (AD-D). T1WI-based radiomics features of the bilateral hippocampi were extracted. To diagnose AD and differentiate between AD-MCI and AD-D, predictive models were developed using random forest (RF), logistic regression (LR), and support vector machine (SVM). We compared radiomics features between the AD and HC groups, as well as within the subgroups of AD-MCI and AD-D. Area under the curve (AUC), accuracy, sensitivity, and specificity were all used to assess model performance. Furthermore, correlations between radiomics features and Mini-Mental State Examination (MMSE) scores, tau protein phosphorylated at threonine 181 (P-tau-181), and amyloid β peptide1-42 (Aβ1-42) were analyzed.The RF model demonstrated superior performance in distinguishing AD from HC (AUC=0.961, accuracy=90.8%, sensitivity=90.7%, specificity=90.9%) and in identifying AD-MCI and AD-D (AUC=0.875, accuracy=80.7%, sensitivity=87.2%, specificity=73.2%) compared to the other models. Additionally, radiomics features were correlated with MMSE scores, P-tau-181, and Aβ1-42 levels in AD.T1WI-based hippocampal radiomics features are valuable for diagnosing AD and identifying AD-MCI and AD-D.Copyright © 2024 The Association of University Radiologists. Published by Elsevier Inc. All rights reserved.
|
| [31] |
The progressive aging of the population will notably increase the burden of those diseases which leads to a disabling situation, such as Alzheimer's disease (AD) and ophthalmological diseases that cause a visual impairment (VI). Eye diseases that cause a VI raise neuroplastic processes in the parietal lobe. Meanwhile, the aforementioned lobe suffers a severe decline throughout AD. From this perspective, diving deeper into the particularities of the parietal lobe is of paramount importance. In this article, we discuss the functions of the parietal lobe, review the parietal anatomical and pathophysiological peculiarities in AD, and also describe some of the changes in the parietal region that occur after VI. Although the alterations in the hippocampus and the temporal lobe have been well documented in AD, the alterations of the parietal lobe have been less thoroughly explored. Recent neuroimaging studies have revealed that some metabolic and perfusion impairments along with a reduction of the white and grey matter could take place in the parietal lobe during AD. Conversely, it has been speculated that blinding ocular diseases induce a remodeling of the parietal region which is observable through the improvement of the integration of multimodal stimuli and in the increase of the volume of this cortical region. Based on current findings concerning the parietal lobe in both pathologies, we hypothesize that the increased activity of the parietal lobe in people with VI may diminish the neurodegeneration of this brain region in those who are visually impaired by oculardiseases.
|
| [32] |
Atrophy of the medial temporal lobe, especially the hippocampus and the parahippocampal gyrus, is considered to be the most predictive structural brain biomarker for Alzheimer's Dementia (AD). However, recent neuroimaging studies reported a possible mismatch between structural and metabolic findings, showing medial temporal lobe atrophy and medial parietal hypoperfusion as biomarkers for AD. The role of the parietal lobe in the development of AD is only recently beginning to attract attention. The current review discusses parietal lobe involvement in the early stages of AD, viz. mild cognitive impairment, as reported from structural, functional, perfusion and metabolic neuroimaging studies. The medial and posterior parts of the parietal lobe seem to be preferentially affected, compared to the other parietal lobe parts. On the basis of the reviewed literature we propose a model showing the relationship between the various pathological events, as measured by different neuroimaging techniques, in the development of AD. In this model myelin breakdown is a beginning of the chain of pathological events leading to AD pathology and an AD diagnosis.Copyright © 2011 Elsevier Ltd. All rights reserved.
|
利益冲突声明： 所有作者在本研究中均不存在利益冲突。
/
| 〈 |
|
〉 |
