Deep Learning for Alzheimer's Disease Prediction: A Comprehensive Review

Alzheimer's disease (AD) is a neurological disorder that significantly impairs cognitive function, leading to memory loss and eventually death. AD progresses through three stages: early stage, mild cognitive impairment (MCI) (middle stage), and dementia. Early diagnosis of Alzheimer's disease is crucial and can improve survival rates among patients. Traditional methods for diagnosing AD through regular checkups and manual examinations are challenging. Advances in computer-aided diagnosis systems (CADs) have led to the development of various artificial intelligence and deep learning-based methods for rapid AD detection. This survey aims to explore the different modalities, feature extraction methods, datasets, machine learning techniques, and validation methods used in AD detection. We reviewed 116 relevant papers from repositories including Elsevier (45), IEEE (25), Springer (19), Wiley (6), PLOS One (5), MDPI (3), World Scientific (3), Frontiers (3), PeerJ (2), Hindawi (2), IO Press (1), and other multiple sources (2). The review is presented in tables for ease of reference, allowing readers to quickly grasp the key findings of each study. Additionally, this review addresses the challenges in the current literature and emphasizes the importance of interpretability and explainability in understanding deep learning model predictions. The primary goal is to assess existing techniques for AD identification and highlight obstacles to guide future research.

Textural features of the frontal white matter could be used to discriminate amnestic mild cognitive impairment patients from the normal population

OBJECTIVE

We aim to develop a radiomics model based on 3-dimensional (3D)-T1WI images to discriminate amnestic mild cognitive impairment (aMCI) patients from the normal population by measuring changes in frontal white matter.

METHODS

In this study, 126 patients with aMCI and 174 normal controls (NC) were recruited from the local community. All subjects underwent routine magnetic resonance imaging examination (including 3D-T1WI ). Participants were randomly divided into a training set (n = 242, aMCI:102, NC:140) and a testing set (n = 58, aMCI:24, NC:34). Texture features of the frontal lobe were extracted from 3D-T1WI images. The least absolute shrinkage and selection operator (LASSO) was used to reduce feature dimensions and develop a radiomics signature model. Diagnostic performance was assessed in the training and testing sets using the receiver operating characteristic (ROC) curve analysis. The area under the ROC curve (AUC), sensitivity, and specificity were also calculated. The efficacy of the radiomics model in discriminating aMCI patients from the normal population was assessed by decision curve analysis (DCA).

RESULTS

A total of 108 frontal lobe texture features were extracted from 3D-T1WI images. LASSO selected 58 radiomic features for the final model, including log-sigma (n = 18), original (n = 8), and wavelet (n = 32) features. The performance of radiomic features extracted from 3D T1 imaging for distinguishing aMCI patients from controls was: in the training set, AUC was 1.00, and the accuracy, sensitivity, and specificity were 100%, 98%, and 100%, respectively. In the testing set, AUC was 0.82 (95% CI:0.69-0.95), and the accuracy, sensitivity, and specificity were 69%, 92%, and 55%, respectively. The DCA demonstrated that the model had favorable clinical predictive value.

CONCLUSIONS

Textural features of white matter in the frontal lobe showed potential for distinguishing aMCI from the normal population, which could be a surrogate protocol to aid aMCI screening in clinical setting.

Graph Theory-Based Approach in Brain Connectivity Modeling and Alzheimer's Disease Detection

There is strong evidence that the pathological findings of Alzheimer's disease (AD), consisting of accumulated amyloid plaques and neurofibrillary tangles, could spread around the brain through synapses and neural connections of neighboring brain sections. Graph theory is a helpful tool in depicting the complex human brain divided into various regions of interest (ROIs) and the connections among them. Thus, applying graph theory-based models in the study of brain connectivity comes natural in the study of AD propagation mechanisms. Moreover, graph theory-based computational approaches have been lately applied in order to boost data-driven analysis, extract model measures and robustness-effectiveness indexes, and provide insights on casual interactions between regions of interest (ROI), as imposed by the models' architecture.

Machine learning algorithms for the diagnosis of Alzheimer and Parkinson disease

Dementia is a general term used to indicate any disorder related to human memory. The various memory-related problems severely affect the human brain and so the individual feels difficulty in doing their normal physical as well as mental activities. There are different types of dementia that exist, but the commonly seen and fatal types of dementia are Alzheimer's disease (AD) and Parkinson's disease (PD). In this paper different efficient Machine Learning Techniques are selected analysed their behaviours in the diagnosis of AD and PD using Positron Emission Tomography (PET). The PET image dataset used in this work consists of 1050 images with AD, PD and Healthy Brain images. The total number of images is split into two different categories in the ratio of 7:3 for training and testing respectively. The different machine learning classifiers used are Bagged Ensemble, ID3, Naive Bayes and Multiclass Support Vector Machine. The classification of the AD and PD with the reference of a healthy brain is done by comparing the input image with the trained samples in the PET image database. In the comparison of trained samples with the input image for the PET images, the bagged ensemble learning classifier worked better than the other classification algorithms and yielded an accuracy of 90.3%.

An enhanced multi-modal brain graph network for classifying neuropsychiatric disorders

It has been proven that neuropsychiatric disorders (NDs) can be associated with both structures and functions of brain regions. Thus, data about structures and functions could be usefully combined in a comprehensive analysis. While brain structural MRI (sMRI) images contain anatomic and morphological information about NDs, functional MRI (fMRI) images carry complementary information. However, efficient extraction and fusion of sMRI and fMRI data remains challenging. In this study, we develop an enhanced multi-modal graph convolutional network (MME-GCN) in a binary classification between patients with NDs and healthy controls, based on the fusion of the structural and functional graphs of the brain region. First, based on the same brain atlas, we construct structural and functional graphs from sMRI and fMRI data, respectively. Second, we use machine learning to extract important features from the structural graph network. Third, we use these extracted features to adjust the corresponding edge weights in the functional graph network. Finally, we train a multi-layer GCN and use it in binary classification task. MME-GCN achieved 93.71% classification accuracy on the open data set provided by the Consortium for Neuropsychiatric Phenomics. In addition, we analyzed the important features selected from the structural graph and verified them in the functional graph. Using MME-GCN, we found several specific brain connections important to NDs.

A predictive model using the mesoscopic architecture of the living brain to detect Alzheimer's disease

Background

Alzheimer's disease, the most common cause of dementia, causes a progressive and irreversible deterioration of cognition that can sometimes be difficult to diagnose, leading to suboptimal patient care.

Methods

We developed a predictive model that computes multi-regional statistical morpho-functional mesoscopic traits from T1-weighted MRI scans, with or without cognitive scores. For each patient, a biomarker called "Alzheimer's Predictive Vector" (ApV) was derived using a two-stage least absolute shrinkage and selection operator (LASSO).

Results

The ApV reliably discriminates between people with (ADrp) and without (nADrp) Alzheimer's related pathologies (98% and 81% accuracy between ADrp - including the early form, mild cognitive impairment - and nADrp in internal and external hold-out test sets, respectively), without any a priori assumptions or need for neuroradiology reads. The new test is superior to standard hippocampal atrophy (26% accuracy) and cerebrospinal fluid beta amyloid measure (62% accuracy). A multiparametric analysis compared DTI-MRI derived fractional anisotropy, whose readout of neuronal loss agrees with ADrp phenotype, and SNPrs2075650 is significantly altered in patients with ADrp-like phenotype.

Conclusions

This new data analytic method demonstrates potential for increasing accuracy of Alzheimer diagnosis.

Radiomics-Based Artificial Intelligence Differentiation of Neurodegenerative Diseases with Reference to the Volumetry

This study aimed to build automated detection models—one by brain regional volume (V-model), and the other by radiomics features of the whole brain (R-model)—to differentiate mild cognitive impairment (MCI) from cognitive normal (CN), and Alzheimer’s Disease (AD) from mild cognitive impairment (MCI). The objectives are to compare the models and identify whether radiomics or volumetry can provide a better prediction for differentiating different types of dementia. Method: 582 MRI T1-weighted images were retrieved from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) database, which is a multicenter operating open source database for AD. In total, 97 images of AD, 293 images of MCI patient and 192 images of cognitive normal were divided into a training, a validation and a test group at a ratio of 70:15:15. For each T1-weighted image, volumetric segmentation was performed with the image analysis software FreeSurfer, and radiomics features were retrieved by imaging research software 3D slicers. Brain regional volume and radiomics features were used to build the V-model and R-model, respectively, using the random forest algorithm by R. The receiver operating characteristics (ROC) curve of both models were used to evaluate their diagnostic accuracy and reliability to differentiate AD, MCI and CN. Results: To differentiate MCI and CN, both V-model and R-model achieved excellent performance, with an AUC of 0.9992 ± 0.0022 and 0.9850 ± 0.0032, respectively. No significant difference was found between the two AUCs, indicating both models attained similar good performance. In MCI and AD differentiation, the V-model and R-model yielded AUC of 0.9986 ± 0.0013 and 0.9714 ± 0.0175, respectively. The best performance was to differentiate AD from CN, where the V-model and R-model yielded AUC of 0.9994 ± 0.0019 and 0.9830 ± 0.009, respectively. The results suggested that both volumetry and radiomics approaches could be used in differentiating AD, MCI and CN, based on T1 weighted MR images using random forest algorithm successfully. Conclusion: This study showed that the radiomics features from T1-weighted MR images achieved excellence performance in differentiating AD, MCI and CN. Compared to the volumetry method, the accuracy, sensitivity and specificity are slightly lower in using radiomics, but still attained very good and reliable classification of the three stages of neurodegenerations. In view of the convenience and operator independence in feature extraction, radiomics can be a quantitative biomarker to differentiate the disease groups.

Identification of early mild cognitive impairment using multi-modal data and graph convolutional networks

BACKGROUND

The identification of early mild cognitive impairment (EMCI), which is an early stage of Alzheimer's disease (AD) and is associated with brain structural and functional changes, is still a challenging task. Recent studies show great promises for improving the performance of EMCI identification by combining multiple structural and functional features, such as grey matter volume and shortest path length. However, extracting which features and how to combine multiple features to improve the performance of EMCI identification have always been a challenging problem. To address this problem, in this study we propose a new EMCI identification framework using multi-modal data and graph convolutional networks (GCNs). Firstly, we extract grey matter volume and shortest path length of each brain region based on automated anatomical labeling (AAL) atlas as feature representation from T1w MRI and rs-fMRI data of each subject, respectively. Then, in order to obtain features that are more helpful in identifying EMCI, a common multi-task feature selection method is applied. Afterwards, we construct a non-fully labelled subject graph using imaging and non-imaging phenotypic measures of each subject. Finally, a GCN model is adopted to perform the EMCI identification task.

RESULTS

Our proposed EMCI identification method is evaluated on 210 subjects, including 105 subjects with EMCI and 105 normal controls (NCs), with both T1w MRI and rs-fMRI data from the Alzheimer's Disease Neuroimaging Initiative (ADNI) database. Experimental results show that our proposed framework achieves an accuracy of 84.1% and an area under the receiver operating characteristic (ROC) curve (AUC) of 0.856 for EMCI/NC classification. In addition, by comparison, the accuracy and AUC values of our proposed framework are better than those of some existing methods in EMCI identification.

CONCLUSION

Our proposed EMCI identification framework is effective and promising for automatic diagnosis of EMCI in clinical practice.

Alzheimer's Disease Classification Based on Multi-feature Fusion

BACKGROUND

In this study, we investigated the fusion of texture and morphometric features as a possible diagnostic biomarker for Alzheimer's Disease (AD).

METHODS

In particular, we classified subjects with Alzheimer's disease, Mild Cognitive Impairment (MCI) and Normal Control (NC) based on texture and morphometric features. Currently, neuropsychiatric categorization provides the ground truth for AD and MCI diagnosis. This can then be supported by biological data such as the results of imaging studies. Cerebral atrophy has been shown to correlate strongly with cognitive symptoms. Hence, Magnetic Resonance (MR) images of the brain are important resources for AD diagnosis. In the proposed method, we used three different types of features identified from structural MR images: Gabor, hippocampus morphometric, and Two Dimensional (2D) and Three Dimensional (3D) Gray Level Co-occurrence Matrix (GLCM). The experimental results, obtained using a 5-fold cross-validated Support Vector Machine (SVM) with 2DGLCM and 3DGLCM multi-feature fusion approaches, indicate that we achieved 81.05% ±1.34, 86.61% ±1.25 correct classification rate with 95% Confidence Interval (CI) falls between (80.75-81.35) and (86.33-86.89) respectively, 83.33%±2.15, 84.21%±1.42 sensitivity and 80.95%±1.52, 85.00%±1.24 specificity in our classification of AD against NC subjects, thus outperforming recent works found in the literature. For the classification of MCI against AD, the SVM achieved a 76.31% ± 2.18, 78.95% ±2.26 correct classification rate, 75.00% ±1.34, 76.19%±1.84 sensitivity and 77.78% ±1.14, 82.35% ±1.34 specificity.

RESULTS AND CONCLUSION

The results of the third experiment, with MCI against NC, also showed that the multiclass SVM provided highly accurate classification results. These findings suggest that this approach is efficient and may be a promising strategy for obtaining better AD, MCI and NC classification performance.

Multi-View Based Multi-Model Learning for MCI Diagnosis

Mild cognitive impairment (MCI) is the early stage of Alzheimer’s disease (AD). Automatic diagnosis of MCI by magnetic resonance imaging (MRI) images has been the focus of research in recent years. Furthermore, deep learning models based on 2D view and 3D view have been widely used in the diagnosis of MCI. The deep learning architecture can capture anatomical changes in the brain from MRI scans to extract the underlying features of brain disease. In this paper, we propose a multi-view based multi-model (MVMM) learning framework, which effectively combines the local information of 2D images with the global information of 3D images. First, we select some 2D slices from MRI images and extract the features representing 2D local information. Then, we combine them with the features representing 3D global information learned from 3D images to train the MVMM learning framework. We evaluate our model on the Alzheimer’s Disease Neuroimaging Initiative (ADNI) database. The experimental results show that our proposed model can effectively recognize MCI through MRI images (accuracy of 87.50% for MCI/HC and accuracy of 83.18% for MCI/AD).

Machine Learning Techniques for the Diagnosis of Alzheimer’s Disease

Alzheimer’s disease is an incurable neurodegenerative disease primarily affecting the elderly population. Efficient automated techniques are needed for early diagnosis of Alzheimer’s. Many novel approaches are proposed by researchers for classification of Alzheimer’s disease. However, to develop more efficient learning techniques, better understanding of the work done on Alzheimer’s is needed. Here, we provide a review on 165 papers from 2005 to 2019, using various feature extraction and machine learning techniques. The machine learning techniques are surveyed under three main categories: support vector machine (SVM), artificial neural network (ANN), and deep learning (DL) and ensemble methods. We present a detailed review on these three approaches for Alzheimer’s with possible future directions.

Schizophrenia Identification Using Multi-View Graph Measures of Functional Brain Networks

Schizophrenia (SZ) is a functional mental disorder that seriously affects the social life of patients. Therefore, accurate diagnosis of SZ has raised extensive attention of researchers. At present, study of brain network based on resting-state functional magnetic resonance imaging (rs-fMRI) has provided promising results for SZ identification by studying functional network alteration. However, previous studies based on brain network analysis are not very effective for SZ identification. Therefore, we propose an improved SZ identification method using multi-view graph measures of functional brain networks. Firstly, we construct an individual functional connectivity network based on Brainnetome atlas for each subject. Then, multi-view graph measures are calculated by the brain network analysis method as feature representations. Next, in order to consider the relationships between measures within the same brain region in feature selection, multi-view measures are grouped according to the corresponding regions and Sparse Group Lasso is applied to identify discriminative features based on this feature grouping structure. Finally, a support vector machine (SVM) classifier is employed to perform SZ identification task. To evaluate our proposed method, computational experiments are conducted on 145 subjects (71 schizophrenic patients and 74 healthy controls) using a leave-one-out cross-validation (LOOCV) scheme. The results show that our proposed method can obtain an accuracy of 93.10% for SZ identification. By comparison, our method is more effective for SZ identification than some existing methods.

Identifying Interactions Between Long Noncoding RNAs and Diseases Based on Computational Methods

With the development and improvement of next-generation sequencing technology, a great number of noncoding RNAs have been discovered. Long noncoding RNAs (lncRNAs) are the biggest kind of noncoding RNAs with more than 200 nt nucleotides in length. There are increasing evidences showing that lncRNAs play key roles in many biological processes. Therefore, the mutation and dysregulation of lncRNAs have close association with a number of complex human diseases. Identifying the most likely interaction between lncRNAs and diseases becomes a fundamental challenge in human health. A common view is that lncRNAs with similar function tend to be related to phenotypic similar diseases. In this chapter, we firstly introduce the concept of lncRNA, their biological features, and available data resources. Further, the recent computational approaches are explored to identify interactions between long noncoding RNAs and diseases, including their advantages and disadvantages. The key issues and potential future works of predicting interactions between long noncoding RNAs and diseases are also discussed.