Multimodal EEG, MRI and PET data fusion for Alzheimer's disease diagnosis

Studying the Alzheimer's disease continuum using EEG and fMRI in single-modality and multi-modality settings

Alzheimer's disease (AD) is a biological, clinical continuum that covers the preclinical, prodromal, and clinical phases of the disease. Early diagnosis and identification of the stages of Alzheimer's disease (AD) are crucial in clinical practice. Ideally, biomarkers should reflect the underlying process (pathological or otherwise), be reproducible and non-invasive, and allow repeated measurements over time. However, the currently known biomarkers for AD are not suitable for differentiating the stages and predicting the trajectory of disease progression. Some objective parameters extracted using electroencephalography (EEG) and functional magnetic resonance imaging (fMRI) are widely applied to diagnose the stages of the AD continuum. While electroencephalography (EEG) has a high temporal resolution, fMRI has a high spatial resolution. Combined EEG and fMRI (EEG-fMRI) can overcome single-modality drawbacks and obtain multi-dimensional information simultaneously, and it can help explore the hemodynamic changes associated with the neural oscillations that occur during information processing. This technique has been used in the cognitive field in recent years. This review focuses on the different techniques available for studying the AD continuum, including EEG and fMRI in single-modality and multi-modality settings, and the possible future directions of AD diagnosis using EEG-fMRI.

Multi-scale analysis of acupuncture mechanisms for motor and sensory cortex activity based on SEEG data

Acupuncture, a traditional Chinese therapy, is gaining attention for its impact on the brain. While existing electroencephalogram and functional magnetic resonance image research has made significant contributions, this paper utilizes stereo-electroencephalography data for a comprehensive exploration of neurophysiological effects. Employing a multi-scale approach, channel-level analysis reveals notable $\delta $-band activity changes during acupuncture. At the brain region level, acupuncture modulated connectivity between the paracentral lobule and the precentral gyrus. Whole-brain analysis indicates acupuncture's influence on network organization, and enhancing $E_{glob}$ and increased interaction between the motor and sensory cortex. Brain functional reorganization is an important basis for functional recovery or compensation after central nervous system injury. The use of acupuncture to stimulate peripheral nerve trunks, muscle motor points, acupoints, etc., in clinical practice may contribute to the reorganization of brain function. This multi-scale perspective provides diverse insights into acupuncture's effects. Remarkably, this paper pioneers the introduction of stereo-electroencephalography data, advancing our understanding of acupuncture's mechanisms and potential therapeutic benefits in clinical settings.

Artificial intelligence-based diagnosis of Alzheimer's disease with brain MRI images

Alzheimer's disease, a primary neurodegenerative condition, predominantly impacts the elderly and pre-elderly population. This progressive neurological disorder is characterized by an array of symptoms including memory loss, cognitive decline, and various physiological and psychological disturbances, significantly compromising the quality of life of patients and their caregivers. Recent advancements in Magnetic Resonance Imaging (MRI) technology have catalyzed research in AI-enhanced diagnostics for Alzheimer's disease, fostering optimism for early detection and timely interventions. This progress has paved the way for the development of sophisticated algorithms and models adept at analyzing complex brain imaging data, thereby augmenting diagnostic accuracy and efficiency. This advancement fuels optimism regarding the transformative potential of AI-driven diagnostics in revolutionizing Alzheimer's disease management, with the prospect of facilitating more effective treatment strategies and improved patient outcomes. The objective of this review is to provide a comprehensive overview of recent developments in deep learning methodologies applied to brain MRI images for the classification of various stages of Alzheimer's disease, with a particular emphasis on early diagnosis. Furthermore, this review underscores the limitations of current research, discussing potential challenges and future research directions in this dynamic field.

Computational methods of EEG signals analysis for Alzheimer's disease classification

Computational analysis of electroencephalographic (EEG) signals have shown promising results in detecting brain disorders, such as Alzheimer's disease (AD). AD is a progressive neurological illness that causes neuron cells degeneration, resulting in cognitive impairment. While there is no cure for AD, early diagnosis is critical to improving the quality of life of affected individuals. Here, we apply six computational time-series analysis methods (wavelet coherence, fractal dimension, quadratic entropy, wavelet energy, quantile graphs and visibility graphs) to EEG records from 160 AD patients and 24 healthy controls. Results from raw and wavelet-filtered (alpha, beta, theta and delta bands) EEG signals show that some of the time-series analysis methods tested here, such as wavelet coherence and quantile graphs, can robustly discriminate between AD patients from elderly healthy subjects. They represent a promising non-invasive and low-cost approach to the AD detection in elderly patients.

Electroencephalogram Analysis of Magnetic Stimulation at Different Acupoints

Magnetic stimulation has some similarities with acupuncture, and it has broad application prospects because of its non-invasiveness and easy quantification. This paper combines magnetic stimulation technology with electroencephalography to analyze the time-frequency and the brain functional network results elicited by magnetic stimulation at different acupoints. This paper hopes to observe the different effects of stimulating different acupoints on the brain from the perspective of EEG. The EEG signals during magnetic stimulation at ST36, ST40, and GB37 were recorded, respectively. The time-frequency results showed that the magnetic stimulation at ST36 and ST40 on the Foot Yangming Stomach Meridian increased the energy in the left parietal lobe and the right central region, and the energy increased mainly in the theta and alpha bands. However, during the magnetic stimulation at GB37 on the Foot Shaoyang Gallbladder Meridian, the energy in the central region and the frontal lobe increased, and the energy increased mainly in the delta, theta, and alpha bands. Moreover, the energy in the right parietal lobe decreased during magnetic stimulation at GB37. The results of brain functional network were also consistent with time-frequency results. The brain network connections of GB37 stimulation in the central region were significantly less than that of ST36 and ST40 (p < 0.01). In addition, the connections between central region and frontal lobe and the connections between central region and parietal lobe of GB37 stimulation were significantly different from that of ST36 and ST40 (p < 0.01). The above results indicate that ST36 and ST40 on the same meridian have similar effects on the brain, while GB37 on the other meridian has completely different effects from ST36 and ST40. The results of this paper explain the reason why stimulating ST36 and ST40 can treat similar diseases from the perspective of EEG, and also explain that stimulating GB37 has significantly different effects on the brain from that of ST36 and ST40.

An insight into the role of Artificial Intelligence in the early diagnosis of Alzheimer's disease

BACKGROUND

The complication of Alzheimer's disease (AD) has made the development of its therapeutic a challenging task. Even after decades of research, we have achieved no more than a few years of symptomatic relief. The inability to diagnose the disease early is the foremost hurdle behind its treatment. Several studies have aimed to identify potential biomarkers that can be detected in body fluids (CSF, blood, urine, etc) or assessed by neuroimaging (i.e., PET and MRI). However, the clinical implementation of these biomarkers is incomplete as they cannot be validated.

METHOD

To overcome the limitation, the use of artificial intelligence along with technical tools has been extensively investigated for AD diagnosis. For developing a promising artificial intelligence strategy that can diagnose AD early, it is critical to supervise neuropsychological outcomes and imaging-based readouts with a proper clinical review.

CONCLUSION

Profound knowledge, a large data pool, and detailed investigations are required for the successful implementation of this tool. This review will enlighten various aspects of early diagnosis of AD using artificial intelligence.

Stacked autoencoders as new models for an accurate Alzheimer's disease classification support using resting-state EEG and MRI measurements

OBJECTIVE

This retrospective and exploratory study tested the accuracy of artificial neural networks (ANNs) at detecting Alzheimer's disease patients with dementia (ADD) based on input variables extracted from resting-state electroencephalogram (rsEEG), structural magnetic resonance imaging (sMRI) or both.

METHODS

For the classification exercise, the ANNs had two architectures that included stacked (autoencoding) hidden layers recreating input data in the output. The classification was based on LORETA source estimates from rsEEG activity recorded with 10-20 montage system (19 electrodes) and standard sMRI variables in 89 ADD and 45 healthy control participants taken from a national database.

RESULTS

The ANN with stacked autoencoders and a deep leaning model representing both ADD and control participants showed classification accuracies in discriminating them of 80%, 85%, and 89% using rsEEG, sMRI, and rsEEG + sMRI features, respectively. The two ANNs with stacked autoencoders and a deep leaning model specialized for either ADD or control participants showed classification accuracies of 77%, 83%, and 86% using the same input features.

CONCLUSIONS

The two architectures of ANNs using stacked (autoencoding) hidden layers consistently reached moderate to high accuracy in the discrimination between ADD and healthy control participants as a function of the rsEEG and sMRI features employed.

SIGNIFICANCE

The present results encourage future multi-centric, prospective and longitudinal cross-validation studies using high resolution EEG techniques and harmonized clinical procedures towards clinical applications of the present ANNs.

Early and Late Fusion Machine Learning on Multi-Frequency Electrical Impedance Data to Improve Radiofrequency Ablation Monitoring

Radiofrequency ablation (RFA) is a popular modality for tumor treatment. However, inexpensive real-time monitoring of RFA within multiple tissue types is still an ongoing research topic. The objective of this study is to utilize multi-frequency electrical impedance data within real-time RFA depth estimation through data fusion schemes that include non-linear machine learning (ML) models. Multi-frequency tissue complex electrical impedance measurements are used to provide input data to the data fusion schemes. Our results show that the fusion schemes significantly decrease both the spread of residuals and the mean of the residuals for depth estimation. Thus, data fusion can be a significant tool for use in improving the performance of ML-based monitoring for RFA.

Comparison of EEG microstates with resting state fMRI and FDG-PET measures in the default mode network via simultaneously recorded trimodal (PET/MR/EEG) data

Simultaneous trimodal positron emission tomography/magnetic resonance imaging/electroencephalography (PET/MRI/EEG) resting state (rs) brain data were acquired from 10 healthy male volunteers. The rs-functional MRI (fMRI) metrics, such as regional homogeneity (ReHo), degree centrality (DC) and fractional amplitude of low-frequency fluctuations (fALFFs), as well as 2-[18F]fluoro-2-desoxy-d-glucose (FDG)-PET standardised uptake value (SUV), were calculated and the measures were extracted from the default mode network (DMN) regions of the brain. Similarly, four microstates for each subject, showing the diverse functional states of the whole brain via topographical variations due to global field power (GFP), were estimated from artefact-corrected EEG signals. In this exploratory analysis, the GFP of microstates was nonparametrically compared to rs-fMRI metrics and FDG-PET SUV measured in the DMN of the brain. The rs-fMRI metrics (ReHO, fALFF) and FDG-PET SUV did not show any significant correlations with any of the microstates. The DC metric showed a significant positive correlation with microstate C (r_s  = 0.73, p = .01). FDG-PET SUVs indicate a trend for a negative correlation with microstates A, B and C. The positive correlation of microstate C with DC metrics suggests a functional relationship between cortical hubs in the frontal and occipital lobes. The results of this study suggest further exploration of this method in a larger sample and in patients with neuropsychiatric disorders. The aim of this exploratory pilot study is to lay the foundation for the development of such multimodal measures to be applied as biomarkers for diagnosis, disease staging, treatment response and monitoring of neuropsychiatric disorders.

Multimodal Assessment of Neural Substrates in Computerized Cognitive Training: A Preliminary Study

BACKGROUND AND PURPOSE

Several studies have validated the clinical efficacy of computerized cognitive training applications. However, few studies have investigated the neural substrates of these training applications using simultaneous multimodal neuroimaging modalities. We aimed to determine the effectiveness of computerized cognitive training and corresponding neural substrates through a multimodal approach.

METHODS

Ten patients with mild cognitive impairment (MCI), six patients with subjective memory impairment (SMI), and 10 normal controls received custom-developed computerized cognitive training in the memory clinic of a university hospital. All of the participants completed 24 sessions of computerized cognitive training, each lasting 40 minutes and performed twice weekly. They were assessed using neuropsychological tests (both computerized and conventional), electroencephalography, fluorodeoxyglucose positron-emission tomography (FDG-PET), volumetric magnetic resonance imaging (MRI), and diffusion-tensor imaging (DTI) at pre- and posttraining.

RESULTS

The patients with MCI exhibited significant improvements in the trail-making test-black & white-B, and memory domain of the computerized cognitive assessment. Subjects with normal cognition exhibited significant improvements in scores in the language and attention-/psychomotor-speed domains. There were no significant changes in subjects with SMI. In the pre- and posttraining evaluations of the MCI group, FDG-PET showed focal activation in the left anterior insula and anterior cingulate after training. Volumetric MRI showed a focal increase in the cortical thickness in the rostral anterior cingulate. DTI revealed increased fractional anisotropy in several regions, including the anterior cingulate.

CONCLUSIONS

The anterior cingulate and anterior insula, which are parts of the salience network, may be substrates for the improvements in cognitive function induced by computerized cognitive training.

MRI based automated diagnosis of Alzheimer's: Fusing 3D wavelet-features with clinical data

This paper presents a novel algorithm for classification of patients with Alzheimer's disease (AD) and Mild Cognitive Impairment (MCI) from the healthy controls (HC) using structural MRI. Feature extraction is based on discrete 3D wavelet transform followed by PCA for transforming the feature space into linearly uncorrelated variables. Linear SVM is used for classification purposes with clinical dementia rating used as the target vector. Proposed methodology is fully automated and independent of the annotation of region of interest. The importance of MRI, demographical data, neuro-psychiatric test scores and statistics calculated over the wavelet coefficients for the classification is studied. Proposed methodology is applied on 197 subjects from a public database. A classification accuracy of 95% was achieved for the case of HC vs AD. For the case of HC vs MCI, and MCI vs AD the classification accuracy of 78% and 81% were achieved. The results are compared with an existing state of the art technique.

Simultaneous trimodal PET-MR-EEG imaging: Do EEG caps generate artefacts in PET images?

Trimodal simultaneous acquisition of positron emission tomography (PET), magnetic resonance imaging (MRI), and electroencephalography (EEG) has become feasible due to the development of hybrid PET-MR scanners. To capture the temporal dynamics of neuronal activation on a millisecond-by-millisecond basis, an EEG system is appended to the quantitative high resolution PET-MR imaging modality already established in our institute. One of the major difficulties associated with the development of simultaneous trimodal acquisition is that the components traditionally used in each modality can cause interferences in its counterpart. The mutual interferences of MRI components and PET components on PET and MR images, and the influence of EEG electrodes on functional MRI images have been studied and reported on. Building on this, this study aims to investigate the influence of the EEG cap on the quality and quantification of PET images acquired during simultaneous PET-MR measurements. A preliminary transmission scan study on the ECAT HR+ scanner, using an Iida phantom, showed visible attenuation effect due to the EEG cap. The BrainPET-MR emission images of the Iida phantom with [¹⁸F]Fluordeoxyglucose, as well as of human subjects with the EEG cap, did not show significant effects of the EEG cap, even though the applied attenuation correction did not take into account the attenuation of the EEG cap itself.

Prediction of advertisement preference by fusing EEG response and sentiment analysis

This paper presents a novel approach to predict rating of video-advertisements based on a multimodal framework combining physiological analysis of the user and global sentiment-rating available on the internet. We have fused Electroencephalogram (EEG) waves of user and corresponding global textual comments of the video to understand the user's preference more precisely. In our framework, the users were asked to watch the video-advertisement and simultaneously EEG signals were recorded. Valence scores were obtained using self-report for each video. A higher valence corresponds to intrinsic attractiveness of the user. Furthermore, the multimedia data that comprised of the comments posted by global viewers, were retrieved and processed using Natural Language Processing (NLP) technique for sentiment analysis. Textual contents from review comments were analyzed to obtain a score to understand sentiment nature of the video. A regression technique based on Random forest was used to predict the rating of an advertisement using EEG data. Finally, EEG based rating is combined with NLP-based sentiment score to improve the overall prediction. The study was carried out using 15 video clips of advertisements available online. Twenty five participants were involved in our study to analyze our proposed system. The results are encouraging and these suggest that the proposed multimodal approach can achieve lower RMSE in rating prediction as compared to the prediction using only EEG data.

Multimodal data and machine learning for surgery outcome prediction in complicated cases of mesial temporal lobe epilepsy

BACKGROUND

This study sought to predict postsurgical seizure freedom from pre-operative diagnostic test results and clinical information using a rapid automated approach, based on supervised learning methods in patients with drug-resistant focal seizures suspected to begin in temporal lobe.

METHOD

We applied machine learning, specifically a combination of mutual information-based feature selection and supervised learning classifiers on multimodal data, to predict surgery outcome retrospectively in 20 presurgical patients (13 female; mean age±SD, in years 33±9.7 for females, and 35.3±9.4 for males) who were diagnosed with mesial temporal lobe epilepsy (MTLE) and subsequently underwent standard anteromesial temporal lobectomy. The main advantage of the present work over previous studies is the inclusion of the extent of ipsilateral neocortical gray matter atrophy and spatiotemporal properties of depth electrode-recorded seizures as training features for individual patient surgery planning.

RESULTS

A maximum relevance minimum redundancy (mRMR) feature selector identified the following features as the most informative predictors of postsurgical seizure freedom in this study's sample of patients: family history of epilepsy, ictal EEG onset pattern (positive correlation with seizure freedom), MRI-based gray matter thickness reduction in the hemisphere ipsilateral to seizure onset, proportion of seizures that first appeared in ipsilateral amygdala to total seizures, age, epilepsy duration, delay in the spread of ipsilateral ictal discharges from site of onset, gender, and number of electrode contacts at seizure onset (negative correlation with seizure freedom). Using these features in combination with a least square support vector machine (LS-SVM) classifier compared to other commonly used classifiers resulted in very high surgical outcome prediction accuracy (95%).

CONCLUSIONS

Supervised machine learning using multimodal compared to unimodal data accurately predicted postsurgical outcome in patients with atypical MTLE.

Classification of Alzheimer's disease from FDG-PET images using favourite class ensembles

Classification of Alzheimer's disease (AD) and Mild Cognitive Impairment (MCI) from brain images using machine learning methods has become popular. Although the large majority of the existing techniques rely on a single classifier such as the Support Vector Machine (SVM), several ensemble methods such as Adaboost or Random Forests (RF) have also been explored. The ensemble methods combine the outputs of several classifiers and aim to increase performance by exploring the diversity of the base classifiers in terms of features or examples, which are usually randomly selected. In this paper we propose using a different kind of ensemble to address the three class problem of classifying AD, MCI and Control Normals (CN) from PET brain images. We propose the favourite class ensemble of classifiers where each base classifier in the ensemble uses a different feature subset which is optimized for a given class. Since different image features correspond to different sets of brain voxels, the proposed favourite class classifiers are able to take into account the fact that the spatial pattern of brain degeneration in AD changes in time as the disease progresses. We tested this approach on FDG-PET images from The Alzheimer's Disease Neuroimaging Initiative (ADNI) database using as base classifiers both Support Vector Machines (SVM) and Random Forests (RF). The ensembles systematically outperformed the corresponding single classifier with the best result (66.78%) being obtained by the SVM ensemble.

Brain functional network in Alzheimer's disease: diagnostic markers for diagnosis and monitoring

Alzheimer's disease (AD) is the most common type of dementia that is clinically characterized by the presence of memory impairment and later by impairment in other cognitive domains. The clinical diagnosis is based on interviews with the patient and his/her relatives and on neuropsychological assessment, which are also used to monitor cognitive decline over time. Several biomarkers have been proposed for detecting AD in its earliest stages, that is, in the predementia stage. In an attempt to find noninvasive biomarkers, researchers have investigated the feasibility of neuroimaging tools, such as MR, SPECT, and FDG-PET imaging, as well as neurophysiological measurements using EEG. In this paper, we investigate the brain functional networks in AD, focusing on main neurophysiological techniques, integrating with most relevant functional brain imaging findings.

Can Functional Magnetic Resonance Imaging Improve Success Rates in CNS Drug Discovery?

INTRODUCTION: The bar for developing new treatments for CNS disease is getting progressively higher and fewer novel mechanisms are being discovered, validated and developed. The high costs of drug discovery necessitate early decisions to ensure the best molecules and hypotheses are tested in expensive late stage clinical trials. The discovery of brain imaging biomarkers that can bridge preclinical to clinical CNS drug discovery and provide a 'language of translation' affords the opportunity to improve the objectivity of decision-making. AREAS COVERED: This review discusses the benefits, challenges and potential issues of using a science based biomarker strategy to change the paradigm of CNS drug development and increase success rates in the discovery of new medicines. The authors have summarized PubMed and Google Scholar based publication searches to identify recent advances in functional, structural and chemical brain imaging and have discussed how these techniques may be useful in defining CNS disease state and drug effects during drug development. EXPERT OPINION: The use of novel brain imaging biomarkers holds the bold promise of making neuroscience drug discovery smarter by increasing the objectivity of decision making thereby improving the probability of success of identifying useful drugs to treat CNS diseases. Functional imaging holds the promise to: (1) define pharmacodynamic markers as an index of target engagement (2) improve translational medicine paradigms to predict efficacy; (3) evaluate CNS efficacy and safety based on brain activation; (4) determine brain activity drug dose-response relationships and (5) provide an objective evaluation of symptom response and disease modification.