Combined Support Vector Machine Classifier and Brain Structural Network Features for the Individual Classification of Amnestic Mild Cognitive Impairment and Subjective Cognitive Decline Patients

White matter alterations and their associations with biomarkers and behavior in subjective cognitive decline individuals: a fixel-based analysis

BACKGROUND

Subjective cognitive decline (SCD) is an early stage of dementia linked to Alzheimer's disease pathology. White matter changes were found in SCD using diffusion tensor imaging, but there are known limitations in voxel-wise tensor-based methods. Fixel-based analysis (FBA) can help understand changes in white matter fibers and how they relate to neurodegenerative proteins and multidomain behavior data in individuals with SCD.

METHODS

Healthy adults with normal cognition were recruited in the Northeastern Taiwan Community Medicine Research Cohort in 2018-2022 and divided into SCD and normal control (NC). Participants underwent evaluations to assess cognitive abilities, mental states, physical activity levels, and susceptibility to fatigue. Neurodegenerative proteins were measured using an immunomagnetic reduction technique. Multi-shell diffusion MRI data were collected and analyzed using whole-brain FBA, comparing results between groups and correlating them with multidomain assessments.

RESULTS

The final enrollment included 33 SCD and 46 NC participants, with no significant differences in age, sex, or education between the groups. SCD had a greater fiber-bundle cross-section than NC (pFWE < 0.05) at bilateral frontal superior longitudinal fasciculus II (SLFII). These white matter changes correlate negatively with plasma Aβ42 level (r = -0.38, p = 0.01) and positively with the AD8 score for subjective cognitive complaints (r = 0.42, p = 0.004) and the Hamilton Anxiety Rating Scale score for the degree of anxiety (Ham-A, r = 0.35, p = 0.019). The dimensional analysis of FBA metrics and blood biomarkers found positive correlations of plasma neurofilament light chain with fiber density at the splenium of corpus callosum (pFWE < 0.05) and with fiber-bundle cross-section at the right thalamus (pFWE < 0.05). Further examination of how SCD grouping interacts between the correlations of FBA metrics and multidomain assessments showed interactions between the fiber density at the corpus callosum with letter-number sequencing cognitive score (pFWE < 0.01) and with fatigue to leisure activities (pFWE < 0.05).

CONCLUSION

Based on FBA, our investigation suggests white matter structural alterations in SCD. The enlargement of SLFII's fiber cross-section is linked to plasma Aβ42 and neuropsychiatric symptoms, which suggests potential early axonal dystrophy associated with Alzheimer's pathology in SCD. The splenium of the corpus callosum is also a critical region of axonal degeneration and cognitive alteration for SCD.

Topological properties analysis and identification of mild cognitive impairment based on individual morphological brain network connectome

Mild cognitive impairment is considered the prodromal stage of Alzheimer's disease. Accurate diagnosis and the exploration of the pathological mechanism of mild cognitive impairment are extremely valuable for targeted Alzheimer's disease prevention and early intervention. In all, 100 mild cognitive impairment patients and 86 normal controls were recruited in this study. We innovatively constructed the individual morphological brain networks and derived multiple brain connectome features based on 3D-T1 structural magnetic resonance imaging with the Jensen-Shannon divergence similarity estimation method. Our results showed that the most distinguishing morphological brain connectome features in mild cognitive impairment patients were consensus connections and nodal graph metrics, mainly located in the frontal, occipital, limbic lobes, and subcortical gray matter nuclei, corresponding to the default mode network. Topological properties analysis revealed that mild cognitive impairment patients exhibited compensatory changes in the frontal lobe, while abnormal cortical-subcortical circuits associated with cognition were present. Moreover, the combination of multidimensional brain connectome features using multiple kernel-support vector machine achieved the best classification performance in distinguishing mild cognitive impairment patients and normal controls, with an accuracy of 84.21%. Therefore, our findings are of significant importance for developing potential brain imaging biomarkers for early detection of Alzheimer's disease and understanding the neuroimaging mechanisms of the disease.

Structural connectivity from DTI to predict mild cognitive impairment in de novo Parkinson's disease

BACKGROUND

Early detection of Parkinson's disease (PD) patients at high risk for mild cognitive impairment (MCI) can help with timely intervention. White matter structural connectivity is considered an early and sensitive indicator of neurodegenerative disease.

OBJECTIVES

To investigate whether baseline white matter structural connectivity features from diffusion tensor imaging (DTI) of de novo PD patients can help predict PD-MCI conversion at an individual level using machine learning methods.

METHODS

We included 90 de novo PD patients who underwent DTI and 3D T1-weighted imaging. Elastic net-based feature consensus ranking (ENFCR) was used with 1000 random training sets to select clinical and structural connectivity features. Linear discrimination analysis (LDA), support vector machine (SVM), K-nearest neighbor (KNN) and naïve Bayes (NB) classifiers were trained based on features selected more than 500 times. The area under the ROC curve (AUC), accuracy (ACC), sensitivity (SEN) and specificity (SPE) were used to evaluate model performance.

RESULTS

A total of 57 PD patients were classified as PD-MCI nonconverters, and 33 PD patients were classified as PD-MCI converters. The models trained with clinical data showed moderate performance (AUC range: 0.62-0.68; ACC range: 0.63-0.77; SEN range: 0.45-0.66; SPE range: 0.64-0.84). Models trained with structural connectivity (AUC range, 0.81-0.84; ACC range, 0.75-0.86; SEN range, 0.77-0.91; SPE range, 0.71-0.88) performed similar to models that were trained with both clinical and structural connectivity data (AUC range, 0.81-0.85; ACC range, 0.74-0.85; SEN range, 0.79-0.91; SPE range, 0.70-0.89).

CONCLUSIONS

Baseline white matter structural connectivity from DTI is helpful in predicting future MCI conversion in de novo PD patients.

Changes in Multiparametric Magnetic Resonance Imaging and Plasma Amyloid-Beta Protein in Subjective Cognitive Decline

The association between plasma amyloid-beta protein (Aβ) and subjective cognitive decline (SCD) remains controversial. We aimed to explore the correlation between neuroimaging findings, plasma Aβ, and neuropsychological scales using data from 53 SCD patients and 46 age- and sex-matched healthy controls (HCs). Magnetic resonance imaging (MRI) was used to obtain neuroimaging data for a whole-brain voxel-based morphometry analysis and cortical functional network topological features. The SCD group had slightly lower Montreal Cognitive Assessment (MoCA) scores than the HC group. The Aβ42 levels were significantly higher in the SCD group than in the HC group (p < 0.05). The SCD patients demonstrated reduced volumes in the left hippocampus, right rectal gyrus (REC.R), and right precentral gyrus (PreCG.R); an increased percentage fluctuation in the left thalamus (PerAF); and lower average small-world coefficient (aSigma) and average global efficiency (aEg) values. Correlation analyses with Aβ and neuropsychological scales revealed significant positive correlations between the volumes of the HIP.L, REC.R, PreCG.R, and MoCA scores. The HIP.L volume and Aβ42 were negatively correlated, as were the REC.R volume and Aβ42/40. PerAF and aSigma were negatively and positively correlated with the MoCA scores, respectively. The aEg was positively correlated with Aβ42/40. SCD patients may exhibit alterations in plasma biomarkers and multi-parameter MRI that resemble those observed in Alzheimer's disease, offering a theoretical foundation for early clinical intervention in SCD.

Recent contributions to the field of subjective cognitive decline in aging: A literature review

Subjective cognitive decline (SCD) is defined as self-experienced, persistent concerns of decline in cognitive capacity in the context of normal performance on objective cognitive measures. Although SCD was initially thought to represent the "worried well," these concerns can be linked to subtle brain changes prior to changes in objective cognitive performance and, therefore, in some individuals, SCD may represent the early stages of an underlying neurodegenerative disease process (e.g., Alzheimer's disease). The field of SCD research has expanded rapidly over the years, and this review aims to provide an update on new advances in, and contributions to, the field of SCD in key areas and themes identified by researchers in this field as particularly important and impactful. First, we highlight recent studies examining sociodemographic and genetic risk factors for SCD, including explorations of SCD across racial and ethnic minoritized groups, and examinations of sex and gender considerations. Next, we review new findings on relationships between SCD and in vivo markers of pathophysiology, utilizing neuroimaging and biofluid data, as well as associations between SCD and objective cognitive tests and neuropsychiatric measures. Finally, we summarize recent work on interventions for SCD and areas of future growth in the field of SCD.

Abnormal white matter changes in Alzheimer's disease based on diffusion tensor imaging: A systematic review

Alzheimer's disease (AD) is a degenerative neurological disease in elderly individuals. Subjective cognitive decline (SCD), mild cognitive impairment (MCI) and further development to dementia (d-AD) are considered to be major stages of the progressive pathological development of AD. Diffusion tensor imaging (DTI), one of the most important modalities of MRI, can describe the microstructure of white matter through its tensor model. It is widely used in understanding the central nervous system mechanism and finding appropriate potential biomarkers for the early stages of AD. Based on the multilevel analysis methods of DTI (voxelwise, fiberwise and networkwise), we summarized that AD patients mainly showed extensive microstructural damage, structural disconnection and topological abnormalities in the corpus callosum, fornix, and medial temporal lobe, including the hippocampus and cingulum. The diffusion features and structural connectomics of specific regions can provide information for the early assisted recognition of AD. The classification accuracy of SCD and normal controls can reach 92.68% at present. And due to the further changes of brain structure and function, the classification accuracy of MCI, d-AD and normal controls can reach more than 97%. Finally, we summarized the limitations of current DTI-based AD research and propose possible future research directions.

An attention-based deep learning approach for the classification of subjective cognitive decline and mild cognitive impairment using resting-state EEG

Objective. This study aims to design and implement the first deep learning (DL) model to classify subjects in the prodromic states of Alzheimer's disease (AD) based on resting-state electroencephalographic (EEG) signals.Approach. EEG recordings of 17 healthy controls (HCs), 56 subjective cognitive decline (SCD) and 45 mild cognitive impairment (MCI) subjects were acquired at resting state. After preprocessing, we selected sections corresponding to eyes-closed condition. Five different datasets were created by extracting delta, theta, alpha, beta and delta-to-theta frequency bands using bandpass filters. To classify SCDvsMCI and HCvsSCDvsMCI, we propose a framework based on the transformer architecture, which uses multi-head attention to focus on the most relevant parts of the input signals. We trained and validated the model on each dataset with a leave-one-subject-out cross-validation approach, splitting the signals into 10 s epochs. Subjects were assigned to the same class as the majority of their epochs. Classification performances of the transformer were assessed for both epochs and subjects and compared with other DL models.Main results. Results showed that the delta dataset allowed our model to achieve the best performances for the discrimination of SCD and MCI, reaching an Area Under the ROC Curve (AUC) of 0.807, while the highest results for the HCvsSCDvsMCI classification were obtained on alpha and theta with a micro-AUC higher than 0.74.Significance. We demonstrated that DL approaches can support the adoption of non-invasive and economic techniques as EEG to stratify patients in the clinical population at risk for AD. This result was achieved since the attention mechanism was able to learn temporal dependencies of the signal, focusing on the most discriminative patterns, achieving state-of-the-art results by using a deep model of reduced complexity. Our results were consistent with clinical evidence that changes in brain activity are progressive when considering early stages of AD.

Altered pattern analysis and identification of subjective cognitive decline based on morphological brain network

Subjective cognitive decline (SCD) is considered the first stage of Alzheimer's disease (AD). Accurate diagnosis and the exploration of the pathological mechanism of SCD are extremely valuable for targeted AD prevention. However, there is little knowledge of the specific altered morphological network patterns in SCD individuals. In this present study, 36 SCD cases and 34 paired-matched normal controls (NCs) were recruited. The Jensen-Shannon distance-based similarity (JSS) method was implemented to construct and derive the attributes of multiple brain connectomes (i.e., morphological brain connections and global and nodal graph metrics) of individual morphological brain networks. A t-test was used to discriminate between the selected nodal graph metrics, while the leave-one-out cross-validation (LOOCV) was used to obtain consensus connections. Comparisons were performed to explore the altered patterns of connectome features. Further, the multiple kernel support vector machine (MK-SVM) was used for combining brain connectomes and differentiating SCD from NCs. We showed that the consensus connections and nodal graph metrics with the most discriminative ability were mostly found in the frontal, limbic, and parietal lobes, corresponding to the default mode network (DMN) and frontoparietal task control (FTC) network. Altered pattern analysis demonstrated that SCD cases had a tendency for modularity and local efficiency enhancement. Additionally, using the MK-SVM to combine the features of multiple brain connectomes was associated with optimal classification performance [area under the curve (AUC): 0.9510, sensitivity: 97.22%, specificity: 85.29%, and accuracy: 91.43%]. Therefore, our study highlighted the combination of multiple connectome attributes based on morphological brain networks and offered a valuable method for distinguishing SCD individuals from NCs. Moreover, the altered patterns of multidimensional connectome attributes provided a promising insight into the neuroimaging mechanism and early intervention in SCD subjects.

Feature Fusion and Detection in Alzheimer's Disease Using a Novel Genetic Multi-Kernel SVM Based on MRI Imaging and Gene Data

Voxel-based morphometry provides an opportunity to study Alzheimer's disease (AD) at a subtle level. Therefore, identifying the important brain voxels that can classify AD, early mild cognitive impairment (EMCI) and healthy control (HC) and studying the role of these voxels in AD will be crucial to improve our understanding of the neurobiological mechanism of AD. Combining magnetic resonance imaging (MRI) imaging and gene information, we proposed a novel feature construction method and a novel genetic multi-kernel support vector machine (SVM) method to mine important features for AD detection. Specifically, to amplify the differences among AD, EMCI and HC groups, we used the eigenvalues of the top 24 Single Nucleotide Polymorphisms (SNPs) in a p-value matrix of 24 genes associated with AD for feature construction. Furthermore, a genetic multi-kernel SVM was established with the resulting features. The genetic algorithm was used to detect the optimal weights of 3 kernels and the multi-kernel SVM was used after training to explore the significant features. By analyzing the significance of the features, we identified some brain regions affected by AD, such as the right superior frontal gyrus, right inferior temporal gyrus and right superior temporal gyrus. The findings proved the good performance and generalization of the proposed model. Particularly, significant susceptibility genes associated with AD were identified, such as CSMD1, RBFOX1, PTPRD, CDH13 and WWOX. Some significant pathways were further explored, such as the calcium signaling pathway (corrected p-value = 1.35 × 10-6) and cell adhesion molecules (corrected p-value = 5.44 × 10-4). The findings offer new candidate abnormal brain features and demonstrate the contribution of these features to AD.

A Tensorized Multitask Deep Learning Network for Progression Prediction of Alzheimer's Disease

With the advances in machine learning for the diagnosis of Alzheimer's disease (AD), most studies have focused on either identifying the subject's status through classification algorithms or on predicting their cognitive scores through regression methods, neglecting the potential association between these two tasks. Motivated by the need to enhance the prospects for early diagnosis along with the ability to predict future disease states, this study proposes a deep neural network based on modality fusion, kernelization, and tensorization that perform multiclass classification and longitudinal regression simultaneously within a unified multitask framework. This relationship between multiclass classification and longitudinal regression is found to boost the efficacy of the final model in dealing with both tasks. Different multimodality scenarios are investigated, and complementary aspects of the multimodal features are exploited to simultaneously delineate the subject's label and predict related cognitive scores at future timepoints using baseline data. The main intent in this multitask framework is to consolidate the highest accuracy possible in terms of precision, sensitivity, F1 score, and area under the curve (AUC) in the multiclass classification task while maintaining the highest similarity in the MMSE score as measured through the correlation coefficient and the RMSE for all time points under the prediction task, with both tasks, run simultaneously under the same set of hyperparameters. The overall accuracy for multiclass classification of the proposed KTMnet method is 66.85 ± 3.77. The prediction results show an average RMSE of 2.32 ± 0.52 and a correlation of 0.71 ± 5.98 for predicting MMSE throughout the time points. These results are compared to state-of-the-art techniques reported in the literature. A discovery from the multitasking of this consolidated machine learning framework is that a set of hyperparameters that optimize the prediction results may not necessarily be the same as those that would optimize the multiclass classification. In other words, there is a breakpoint beyond which enhancing further the results of one process could lead to the downgrading in accuracy for the other.

Methodological evaluation of individual cognitive prediction based on the brain white matter structural connectome

An emerging trend is to use regression‐based machine learning approaches to predict cognitive functions at the individual level from neuroimaging data. However, individual prediction models are inherently influenced by the vast options for network construction and model selection in machine learning pipelines. In particular, the brain white matter (WM) structural connectome lacks a systematic evaluation of the effects of different options in the pipeline on predictive performance. Here, we focused on the methodological evaluation of brain structural connectome‐based predictions. For network construction, we considered two parcellation schemes for defining nodes and seven strategies for defining edges. For the regression algorithms, we used eight regression models. Four cognitive domains and brain age were targeted as predictive tasks based on two independent datasets (Beijing Aging Brain Rejuvenation Initiative [BABRI]: 633 healthy older adults; Human Connectome Projects in Aging [HCP‐A]: 560 healthy older adults). Based on the results, the WM structural connectome provided a satisfying predictive ability for individual age and cognitive functions, especially for executive function and attention. Second, different parcellation schemes induce a significant difference in predictive performance. Third, prediction results from different data sets showed that dMRI with distinct acquisition parameters may plausibly result in a preference for proper fiber reconstruction algorithms and different weighting options. Finally, deep learning and Elastic‐Net models are more accurate and robust in connectome‐based predictions. Together, significant effects of different options in WM network construction and regression algorithms on the predictive performances are identified in this study, which may provide important references and guidelines to select suitable options for future studies in this field.