The relative importance of imaging markers for the prediction of Alzheimer's disease dementia in mild cognitive impairment - Beyond classical regression

AI for the prediction of early stages of Alzheimer's disease from neuroimaging biomarkers - A narrative review of a growing field

OBJECTIVES

The objectives of this narrative review are to summarize the current state of AI applications in neuroimaging for early Alzheimer's disease (AD) prediction and to highlight the potential of AI techniques in improving early AD diagnosis, prognosis, and management.

METHODS

We conducted a narrative review of studies using AI techniques applied to neuroimaging data for early AD prediction. We examined single-modality studies using structural MRI and PET imaging, as well as multi-modality studies integrating multiple neuroimaging techniques and biomarkers. Furthermore, they reviewed longitudinal studies that model AD progression and identify individuals at risk of rapid decline.

RESULTS

Single-modality studies using structural MRI and PET imaging have demonstrated high accuracy in classifying AD and predicting progression from mild cognitive impairment (MCI) to AD. Multi-modality studies, integrating multiple neuroimaging techniques and biomarkers, have shown improved performance and robustness compared to single-modality approaches. Longitudinal studies have highlighted the value of AI in modeling AD progression and identifying individuals at risk of rapid decline. However, challenges remain in data standardization, model interpretability, generalizability, clinical integration, and ethical considerations.

CONCLUSION

AI techniques applied to neuroimaging data have the potential to improve early AD diagnosis, prognosis, and management. Addressing challenges related to data standardization, model interpretability, generalizability, clinical integration, and ethical considerations is crucial for realizing the full potential of AI in AD research and clinical practice. Collaborative efforts among researchers, clinicians, and regulatory agencies are needed to develop reliable, robust, and ethical AI tools that can benefit AD patients and society.

Identification of subjective cognitive decline due to Alzheimer's disease using multimodal MRI combining with machine learning

Subjective cognitive decline (SCD) is a preclinical asymptomatic stage of Alzheimer's disease (AD). Accurate diagnosis of SCD represents the greatest challenge for current clinical practice. The multimodal magnetic resonance imaging (MRI) features of 7 brain networks and 90 regions of interests from Chinese and ANDI cohorts were calculated. Machine learning (ML) methods based on support vector machine (SVM) were used to classify SCD plus and normal control. To assure the robustness of ML model, above analyses were repeated in amyloid β (Aβ) and apolipoprotein E (APOE) ɛ4 subgroups. We found that the accuracy of the proposed multimodal SVM method achieved 79.49% and 83.13%, respectively, in Chinese and ANDI cohorts for the diagnosis of the SCD plus individuals. Furthermore, adding Aβ pathology and ApoE ɛ4 genotype information can further improve the accuracy to 85.36% and 82.52%. More importantly, the classification model exhibited the robustness in the crossracial cohorts and different subgroups, which outperforms any single and 2 modalities. The study indicates that multimodal MRI imaging combining with ML classification method yields excellent and powerful performances at categorizing SCD due to AD, suggesting potential for clinical utility.

Analyzing Hierarchical Multi-View MRI Data With StaPLR: An Application to Alzheimer's Disease Classification

Multi-view data refers to a setting where features are divided into feature sets, for example because they correspond to different sources. Stacked penalized logistic regression (StaPLR) is a recently introduced method that can be used for classification and automatically selecting the views that are most important for prediction. We introduce an extension of this method to a setting where the data has a hierarchical multi-view structure. We also introduce a new view importance measure for StaPLR, which allows us to compare the importance of views at any level of the hierarchy. We apply our extended StaPLR algorithm to Alzheimer's disease classification where different MRI measures have been calculated from three scan types: structural MRI, diffusion-weighted MRI, and resting-state fMRI. StaPLR can identify which scan types and which derived MRI measures are most important for classification, and it outperforms elastic net regression in classification performance.

Dynamics of the "Cognitive" Brain Wave P3b at Rest for Alzheimer Dementia Prediction in Mild Cognitive Impairment

Alzheimer's disease (AD) is the most common cause of dementia that involves a progressive and irrevocable decline in cognitive abilities and social behavior, thus annihilating the patient's autonomy. The theoretical assumption that disease-modifying drugs are most effective in the early stages hopefully in the prodromal stage called mild cognitive impairment (MCI) urgently pushes toward the identification of robust and individualized markers of cognitive decline to establish an early pharmacological intervention. This requires the combination of well-established neural mechanisms and the development of increasingly sensitive methodologies. Among the neurophysiological markers of attention and cognition, one of the sub-components of the 'cognitive brain wave' P300 recordable in an odd-ball paradigm -namely the P3b- is extensively regarded as a sensitive indicator of cognitive performance. Several studies have reliably shown that changes in the amplitude and latency of the P3b are strongly related to cognitive decline and aging both healthy and pathological. Here, we used a P3b spatial filter to enhance the electroencephalographic (EEG) characteristics underlying 175 subjects divided into 135 MCI subjects, 20 elderly controls (EC), and 20 young volunteers (Y). The Y group served to extract the P3b spatial filter from EEG data, which was later applied to the other groups during resting conditions with eyes open and without being asked to perform any task. The group of 135 MCI subjects could be divided into two subgroups at the end of a month follow-up: 75 with stable MCI (MCI-S, not converted to AD), 60 converted to AD (MCI-C). The P3b spatial filter was built by means of a signal processing method called Functional Source Separation (FSS), which increases signal-to-noise ratio by using a weighted sum of all EEG recording channels rather than relying on a single, or a small sub-set, of channels. A clear difference was observed for the P3b dynamics at rest between groups. Moreover, a machine learning approach showed that P3b at rest could correctly distinguish MCI from EC (80.6% accuracy) and MCI-S from MCI-C (74.1% accuracy), with an accuracy as high as 93.8% in discriminating between MCI-C and EC. Finally, a comparison of the Bayes factor revealed that the group differences among MCI-S and MCI-C were 138 times more likely to be detected using the P3b dynamics compared with the best performing single electrode (Pz) approach. In conclusion, we propose that P3b as measured through spatial filters can be safely regarded as a simple and sensitive marker to predict the conversion from an MCI to AD status eventually combined with other non-neurophysiological biomarkers for a more precise definition of dementia having neuropathological Alzheimer characteristics.

Performance of binary prediction models in high-correlation low-dimensional settings: a comparison of methods

BACKGROUND

Clinical prediction models are developed widely across medical disciplines. When predictors in such models are highly collinear, unexpected or spurious predictor-outcome associations may occur, thereby potentially reducing face-validity of the prediction model. Collinearity can be dealt with by exclusion of collinear predictors, but when there is no a priori motivation (besides collinearity) to include or exclude specific predictors, such an approach is arbitrary and possibly inappropriate.

METHODS

We compare different methods to address collinearity, including shrinkage, dimensionality reduction, and constrained optimization. The effectiveness of these methods is illustrated via simulations.

RESULTS

In the conducted simulations, no effect of collinearity was observed on predictive outcomes (AUC, R², Intercept, Slope) across methods. However, a negative effect of collinearity on the stability of predictor selection was found, affecting all compared methods, but in particular methods that perform strong predictor selection (e.g., Lasso). Methods for which the included set of predictors remained most stable under increased collinearity were Ridge, PCLR, LAELR, and Dropout.

CONCLUSIONS

Based on the results, we would recommend refraining from data-driven predictor selection approaches in the presence of high collinearity, because of the increased instability of predictor selection, even in relatively high events-per-variable settings. The selection of certain predictors over others may disproportionally give the impression that included predictors have a stronger association with the outcome than excluded predictors.

OUP accepted manuscript

The Coronavirus disease of 2019 (COVID-19) and measures to curb it created population-level changes in male-dominant impulsive and risky behaviors such as violent crimes and gambling. One possible explanation for this is that the pandemic has been stressful, and males, more so than females, tend to respond to stress by altering their focus on immediate versus delayed rewards, as reflected in their delay discounting rates. Delay discounting rates from healthy undergraduate students were collected twice during the pandemic. Discounting rates of males (n=190) but not of females (n=493) increased during the pandemic. Using machine learning, we show that prepandemic functional connectome predict increased discounting rates in males (n=88). Moreover, considering that delay discounting is associated with multiple psychiatric disorders, we found the same neural pattern that predicted increased discounting rates in this study, in secondary datasets of patients with major depression and schizophrenia. The findings point to sex-based differences in maladaptive delay discounting under real-world stress events, and to connectome-based neuromarkers of such effects. They can explain why there was a population-level increase in several impulsive and risky behaviors during the pandemic and point to intriguing questions about the shared underlying mechanisms of stress responses, psychiatric disorders and delay discounting.

A review on omics-based biomarkers discovery for Alzheimer's disease from the bioinformatics perspectives: Statistical approach vs machine learning approach

Alzheimer's Disease (AD) is a neurodegenerative disease that affects cognition and is the most common cause of dementia in the elderly. As the number of elderly individuals increases globally, the incidence and prevalence of AD are expected to increase. At present, AD is diagnosed clinically, according to accepted criteria. The essential elements in the diagnosis of AD include a patients history, a physical examination and neuropsychological testing, in addition to appropriate investigations such as neuroimaging. The omics-based approach is an emerging field of study that may not only aid in the diagnosis of AD but also facilitate the exploration of factors that influence the development of the disease. Omics techniques, including genomics, transcriptomics, proteomics and metabolomics, may reveal the pathways that lead to neuronal death and identify biomolecular markers associated with AD. This will further facilitate an understanding of AD neuropathology. In this review, omics-based approaches that were implemented in studies on AD were assessed from a bioinformatics perspective. Current state-of-the-art statistical and machine learning approaches used in the single omics analysis of AD were compared based on correlations of variants, differential expression, functional analysis and network analysis. This was followed by a review of the approaches used in the integration and analysis of multi-omics of AD. The strengths and limitations of multi-omics analysis methods were explored and the issues and challenges associated with omics studies of AD were highlighted. Lastly, future studies in this area of research were justified.

Unique regional patterns of amyloid burden predict progression to prodromal and clinical stages of Alzheimer's disease

Although beta-amyloid (Aβ) positivity has shown to be associated with higher risk of progression to Alzheimer's disease (AD) in mild cognitive impairment (MCI), information on the time to conversion to manifest dementia cannot be readily deduced from this binary classification. Here, we assessed if regional patterns of Aβ deposition measured with ¹⁸F-florbetapir may serve as biomarker for progression risk in Aβ-positive cognitively normal (CN) and MCI patients, including clinical follow-up data and cerebrospinal fluid (CSF) biomarkers. Voxel-wise group comparisons between age and sex-matched Aβ-positive groups (i.e., CN-stables [n = 38] vs. CN-to-MCI/AD progressors [n = 38], MCI-stables [n = 104] versus MCI-to-AD progressors [n = 104]) revealed higher Aβ burden in precuneus, subcortical, and parietal regions in CN-to-MCI/AD progressors and cingulate, temporal, and frontal regions in MCI-to-AD progressors. Importantly, these regional patterns predicted progression to advanced stages on the AD spectrum in the short and the long-term beyond global Aβ burden and CSF biomarkers. These results suggest that distinct regional patterns of Aβ burden are a valuable biomarker for risk of disease progression in CN and MCI.

A comparison of feature extraction methods for prediction of neuropsychological scores from functional connectivity data of stroke patients

Multivariate prediction of human behavior from resting state data is gaining increasing popularity in the neuroimaging community, with far-reaching translational implications in neurology and psychiatry. However, the high dimensionality of neuroimaging data increases the risk of overfitting, calling for the use of dimensionality reduction methods to build robust predictive models. In this work, we assess the ability of four well-known dimensionality reduction techniques to extract relevant features from resting state functional connectivity matrices of stroke patients, which are then used to build a predictive model of the associated deficits based on cross-validated regularized regression. In particular, we investigated the prediction ability over different neuropsychological scores referring to language, verbal memory, and spatial memory domains. Principal Component Analysis (PCA) and Independent Component Analysis (ICA) were the two best methods at extracting representative features, followed by Dictionary Learning (DL) and Non-Negative Matrix Factorization (NNMF). Consistent with these results, features extracted by PCA and ICA were found to be the best predictors of the neuropsychological scores across all the considered cognitive domains. For each feature extraction method, we also examined the impact of the regularization method, model complexity (in terms of number of features that entered in the model) and quality of the maps that display predictive edges in the resting state networks. We conclude that PCA-based models, especially when combined with L1 (LASSO) regularization, provide optimal balance between prediction accuracy, model complexity, and interpretability.

Use of a Sparse-Response Deep Belief Network and Extreme Learning Machine to Discriminate Alzheimer's Disease, Mild Cognitive Impairment, and Normal Controls Based on Amyloid PET/MRI Images

In recent years, interest has grown in using computer-aided diagnosis (CAD) for Alzheimer's disease (AD) and its prodromal stage, mild cognitive impairment (MCI). However, existing CAD technologies often overfit data and have poor generalizability. In this study, we proposed a sparse-response deep belief network (SR-DBN) model based on rate distortion (RD) theory and an extreme learning machine (ELM) model to distinguish AD, MCI, and normal controls (NC). We used [¹⁸F]-AV45 positron emission computed tomography (PET) and magnetic resonance imaging (MRI) images from 340 subjects enrolled in the ADNI database, including 116 AD, 82 MCI, and 142 NC subjects. The model was evaluated using five-fold cross-validation. In the whole model, fast principal component analysis (PCA) served as a dimension reduction algorithm. An SR-DBN extracted features from the images, and an ELM obtained the classification. Furthermore, to evaluate the effectiveness of our method, we performed comparative trials. In contrast experiment 1, the ELM was replaced by a support vector machine (SVM). Contrast experiment 2 adopted DBN without sparsity. Contrast experiment 3 consisted of fast PCA and an ELM. Contrast experiment 4 used a classic convolutional neural network (CNN) to classify AD. Accuracy, sensitivity, specificity, and area under the curve (AUC) were examined to validate the results. Our model achieved 91.68% accuracy, 95.47% sensitivity, 86.68% specificity, and an AUC of 0.87 separating between AD and NC groups; 87.25% accuracy, 79.74% sensitivity, 91.58% specificity, and an AUC of 0.79 separating MCI and NC groups; and 80.35% accuracy, 85.65% sensitivity, 72.98% specificity, and an AUC of 0.71 separating AD and MCI groups, which gave better classification than other models assessed.

Multivariate Neural Connectivity Patterns in Early Infancy Predict Later Autism Symptoms

BACKGROUND

Functional brain connectivity is altered in children and adults with autism spectrum disorder (ASD). Functional disruption during infancy could provide earlier markers of ASD, thus providing a crucial opportunity to improve developmental outcomes. Using a whole-brain multivariate approach, we asked whether electroencephalography measures of neural connectivity at 3 months of age predict autism symptoms at 18 months.

METHODS

Spontaneous electroencephalography data were collected from 65 infants with and without familial risk for ASD at 3 months of age. Neural connectivity patterns were quantified using phase coherence in the alpha range (6-12 Hz). Support vector regression analysis was used to predict ASD symptoms at age 18 months, with ASD symptoms quantified by the Toddler Module of the Autism Diagnostic Observation Schedule, Second Edition.

RESULTS

Autism Diagnostic Observation Schedule scores predicted by support vector regression algorithms trained on 3-month electroencephalography data correlated highly with Autism Diagnostic Observation Schedule scores measured at 18 months (r = .76, p = .02, root-mean-square error = 2.38). Specifically, lower frontal connectivity and higher right temporoparietal connectivity at 3 months predicted higher ASD symptoms at 18 months. The support vector regression model did not predict cognitive abilities at 18 months (r = .15, p = .36), suggesting specificity of these brain patterns to ASD.

CONCLUSIONS

Using a data-driven, unbiased analytic approach, neural connectivity across frontal and temporoparietal regions at 3 months predicted ASD symptoms at 18 months. Identifying early neural differences that precede an ASD diagnosis could promote closer monitoring of infants who show signs of neural risk and provide a crucial opportunity to mediate outcomes through early intervention.

The Italian INTERCEPTOR Project: From the Early Identification of Patients Eligible for Prescription of Antidementia Drugs to a Nationwide Organizational Model for Early Alzheimer's Disease Diagnosis

Alzheimer's disease is the most common age-related neurodegenerative disorder and its burden on patients, families, and society grows significantly with lifespan. Early modifications of risk-enhancing lifestyles and treatment initiation expand personal autonomy and reduce management costs. Many clinical trials with potentially disease-modifying drugs are devoted to mild cognitive impairment (MCI) prodromal-to-Alzheimer's disease. The identification of biomarkers for early diagnosis may thus be crucial for early intervention and identification of high-risk subjects, the most appropriate target of new drugs as soon as they will be discovered. INTERCEPTOR is a strategic project by the Italian Ministry of Health and the Italian Medicines Agency (AIFA), aiming to validate the best combination (highly accurate, non-invasive, available on the whole national territory and financially sustainable) of biomarkers and organizational model for early diagnosis. 500 MCI subjects will be enrolled at baseline and followed-up for 3 years for at least 400 of them in order to define a "hub & spoke" nationwide model with recruiting (spokes) centers for MCI identification and expert (hubs) centers for risk diagnosis.

Early-Stage Identification and Pathological Development of Alzheimer's Disease Using Multimodal MRI

Alzheimer's disease (AD) is one of the most common progressive and irreversible neurodegenerative diseases. The study of the pathological mechanism of AD and early-stage diagnosis is essential and important. Subjective cognitive decline (SCD), the first at-risk stage of AD occurring prior to amnestic mild cognitive impairment (aMCI), is of great research value and has gained our interest. To investigate the entire pathological development of AD pathology efficiently, we proposed a machine learning classification method based on a multimodal support vector machine (SVM) to investigate the structural and functional connectivity patterns of the three stages of AD (SCD, aMCI, and AD). Our experiments achieved an accuracy of 98.58% in the AD group, 97.76% in the aMCI group, and 80.24% in the SCD group. Moreover, in our experiments, we identified the most discriminating brain regions, which were mainly located in the default mode network and subcortical structures (SCS). Notably, with the development of AD pathology, SCS regions have become increasingly important, and structural connectivity has shown more discriminative power than functional connectivity. The current study may shed new light on the pathological mechanism of AD and suggests that whole-brain connectivity may provide potential effective biomarkers for the early-stage diagnosis of AD.

No evidence for a relationship between social closeness and similarity in resting-state functional brain connectivity in schoolchildren

Previous research suggests that the proximity of individuals in a social network predicts how similarly their brains respond to naturalistic stimuli. However, the relationship between social connectedness and brain connectivity in the absence of external stimuli has not been examined. To investigate whether neural homophily between friends exists at rest we collected resting-state functional magnetic resonance imaging (fMRI) data from 68 school-aged girls, along with social network information from all pupils in their year groups (total 5,066 social dyads). Participants were asked to rate the amount of time they voluntarily spent with each person in their year group, and directed social network matrices and community structure were then determined from these data. No statistically significant relationships between social distance, community homogeneity and similarity of global-level resting-state connectivity were observed. Nor were we able to predict social distance using a regularised regression technique (i.e. elastic net regression based on the local-level similarities in resting-state whole-brain connectivity between participants). Although neural homophily between friends exists when viewing naturalistic stimuli, this finding did not extend to functional connectivity at rest in our population. Instead, resting-state connectivity may be less susceptible to the influences of a person's social environment.

Prediction of Cognitive Decline in Temporal Lobe Epilepsy and Mild Cognitive Impairment by EEG, MRI, and Neuropsychology

Cognitive decline is a severe concern of patients with mild cognitive impairment. Also, in patients with temporal lobe epilepsy, memory problems are a frequently encountered problem with potential progression. On the background of a unifying hypothesis for cognitive decline, we merged knowledge from dementia and epilepsy research in order to identify biomarkers with a high predictive value for cognitive decline across and beyond these groups that can be fed into intelligent systems. We prospectively assessed patients with temporal lobe epilepsy (N = 9), mild cognitive impairment (N = 19), and subjective cognitive complaints (N = 4) and healthy controls (N = 18). All had structural cerebral MRI, EEG at rest and during declarative verbal memory performance, and a neuropsychological assessment which was repeated after 18 months. Cognitive decline was defined as significant change on neuropsychological subscales. We extracted volumetric and shape features from MRI and brain network measures from EEG and fed these features alongside a baseline testing in neuropsychology into a machine learning framework with feature subset selection and 5-fold cross validation. Out of 50 patients, 27 had a decline over time in executive functions, 23 in visual-verbal memory, 23 in divided attention, and 7 patients had an increase in depression scores. The best sensitivity/specificity for decline was 72%/82% for executive functions based on a feature combination from MRI volumetry and EEG partial coherence during recall of memories; 95%/74% for visual-verbal memory by combination of MRI-wavelet features and neuropsychology; 84%/76% for divided attention by combination of MRI-wavelet features and neuropsychology; and 81%/90% for increase of depression by combination of EEG partial directed coherence factor at rest and neuropsychology. Combining information from EEG, MRI, and neuropsychology in order to predict neuropsychological changes in a heterogeneous population could create a more general model of cognitive performance decline.

Neurophysiological Hallmarks of Neurodegenerative Cognitive Decline: The Study of Brain Connectivity as A Biomarker of Early Dementia

Neurodegenerative processes of various types of dementia start years before symptoms, but the presence of a "neural reserve", which continuously feeds and supports neuroplastic mechanisms, helps the aging brain to preserve most of its functions within the "normality" frame. Mild cognitive impairment (MCI) is an intermediate stage between dementia and normal brain aging. About 50% of MCI subjects are already in a stage that is prodromal-to-dementia and during the following 3 to 5 years will develop clinically evident symptoms, while the other 50% remains at MCI or returns to normal. If the risk factors favoring degenerative mechanisms are modified during early stages (i.e., in the prodromal), the degenerative process and the loss of abilities in daily living activities will be delayed. It is therefore extremely important to have biomarkers able to identify-in association with neuropsychological tests-prodromal-to-dementia MCI subjects as early as possible. MCI is a large (i.e., several million in EU) and substantially healthy population; therefore, biomarkers should be financially affordable, largely available and non-invasive, but still accurate in their diagnostic prediction. Neurodegeneration initially affects synaptic transmission and brain connectivity; methods exploring them would represent a 1st line screening. Neurophysiological techniques able to evaluate mechanisms of synaptic function and brain connectivity are attracting general interest and are described here. Results are quite encouraging and suggest that by the application of artificial intelligence (i.e., learning-machine), neurophysiological techniques represent valid biomarkers for screening campaigns of the MCI population.

Gaussian Graphical Models Reveal Inter-Modal and Inter-Regional Conditional Dependencies of Brain Alterations in Alzheimer's Disease

Alzheimer's disease (AD) is characterized by a sequence of pathological changes, which are commonly assessed in vivo using various brain imaging modalities such as magnetic resonance imaging (MRI) and positron emission tomography (PET). Currently, the most approaches to analyze statistical associations between regions and imaging modalities rely on Pearson correlation or linear regression models. However, these models are prone to spurious correlations arising from uninformative shared variance and multicollinearity. Notably, there are no appropriate multivariate statistical models available that can easily integrate dozens of multicollinear variables derived from such data, being able to utilize the additional information provided from the combination of data sources. Gaussian graphical models (GGMs) can estimate the conditional dependency from given data, which is conceptually expected to closely reflect the underlying causal relationships between various variables. Hence, we applied GGMs to assess multimodal regional brain alterations in AD. We obtained data from N = 972 subjects from the Alzheimer's Disease Neuroimaging Initiative. The mean amyloid load (AV45-PET), glucose metabolism (FDG-PET), and gray matter volume (MRI) were calculated for each of the 108 cortical and subcortical brain regions. GGMs were estimated using a Bayesian framework for the combined multimodal data and the resulted conditional dependency networks were compared to classical covariance networks based on Pearson correlation. Additionally, graph-theoretical network statistics were calculated to determine network alterations associated with disease status. The resulting conditional dependency matrices were much sparser (≈10% density) than Pearson correlation matrices (≈50% density). Within imaging modalities, conditional dependency networks yielded clusters connecting anatomically adjacent regions. For the associations between different modalities, only few region-specific connections were detected. Network measures such as small-world coefficient were significantly altered across diagnostic groups, with a biphasic u-shape trajectory, i.e., increased small-world coefficient in early mild cognitive impairment (MCI), similar values in late MCI, and decreased values in AD dementia patients compared to cognitively normal controls. In conclusion, GGMs removed commonly shared variance among multimodal measures of regional brain alterations in MCI and AD, and yielded sparser matrices compared to correlation networks based on the Pearson coefficient. Therefore, GGMs may be used as alternative to thresholding-approaches typically applied to correlation networks to obtain the most informative relations between variables.

Structural magnetic resonance imaging for the early diagnosis of dementia due to Alzheimer's disease in people with mild cognitive impairment

BACKGROUND

Mild cognitive impairment (MCI) due to Alzheimer's disease is the symptomatic predementia phase of Alzheimer's disease dementia, characterised by cognitive and functional impairment not severe enough to fulfil the criteria for dementia. In clinical samples, people with amnestic MCI are at high risk of developing Alzheimer's disease dementia, with annual rates of progression from MCI to Alzheimer's disease estimated at approximately 10% to 15% compared with the base incidence rates of Alzheimer's disease dementia of 1% to 2% per year.

OBJECTIVES

To assess the diagnostic accuracy of structural magnetic resonance imaging (MRI) for the early diagnosis of dementia due to Alzheimer's disease in people with MCI versus the clinical follow-up diagnosis of Alzheimer's disease dementia as a reference standard (delayed verification). To investigate sources of heterogeneity in accuracy, such as the use of qualitative visual assessment or quantitative volumetric measurements, including manual or automatic (MRI) techniques, or the length of follow-up, and age of participants. MRI was evaluated as an add-on test in addition to clinical diagnosis of MCI to improve early diagnosis of dementia due to Alzheimer's disease in people with MCI.

SEARCH METHODS

On 29 January 2019 we searched Cochrane Dementia and Cognitive Improvement's Specialised Register and the databases, MEDLINE, Embase, BIOSIS Previews, Science Citation Index, PsycINFO, and LILACS. We also searched the reference lists of all eligible studies identified by the electronic searches.

SELECTION CRITERIA

We considered cohort studies of any size that included prospectively recruited people of any age with a diagnosis of MCI. We included studies that compared the diagnostic test accuracy of baseline structural MRI versus the clinical follow-up diagnosis of Alzheimer's disease dementia (delayed verification). We did not exclude studies on the basis of length of follow-up. We included studies that used either qualitative visual assessment or quantitative volumetric measurements of MRI to detect atrophy in the whole brain or in specific brain regions, such as the hippocampus, medial temporal lobe, lateral ventricles, entorhinal cortex, medial temporal gyrus, lateral temporal lobe, amygdala, and cortical grey matter.

DATA COLLECTION AND ANALYSIS

Four teams of two review authors each independently reviewed titles and abstracts of articles identified by the search strategy. Two teams of two review authors each independently assessed the selected full-text articles for eligibility, extracted data and solved disagreements by consensus. Two review authors independently assessed the quality of studies using the QUADAS-2 tool. We used the hierarchical summary receiver operating characteristic (HSROC) model to fit summary ROC curves and to obtain overall measures of relative accuracy in subgroup analyses. We also used these models to obtain pooled estimates of sensitivity and specificity when sufficient data sets were available.

MAIN RESULTS

We included 33 studies, published from 1999 to 2019, with 3935 participants of whom 1341 (34%) progressed to Alzheimer's disease dementia and 2594 (66%) did not. Of the participants who did not progress to Alzheimer's disease dementia, 2561 (99%) remained stable MCI and 33 (1%) progressed to other types of dementia. The median proportion of women was 53% and the mean age of participants ranged from 63 to 87 years (median 73 years). The mean length of clinical follow-up ranged from 1 to 7.6 years (median 2 years). Most studies were of poor methodological quality due to risk of bias for participant selection or the index test, or both. Most of the included studies reported data on the volume of the total hippocampus (pooled mean sensitivity 0.73 (95% confidence interval (CI) 0.64 to 0.80); pooled mean specificity 0.71 (95% CI 0.65 to 0.77); 22 studies, 2209 participants). This evidence was of low certainty due to risk of bias and inconsistency. Seven studies reported data on the atrophy of the medial temporal lobe (mean sensitivity 0.64 (95% CI 0.53 to 0.73); mean specificity 0.65 (95% CI 0.51 to 0.76); 1077 participants) and five studies on the volume of the lateral ventricles (mean sensitivity 0.57 (95% CI 0.49 to 0.65); mean specificity 0.64 (95% CI 0.59 to 0.70); 1077 participants). This evidence was of moderate certainty due to risk of bias. Four studies with 529 participants analysed the volume of the total entorhinal cortex and four studies with 424 participants analysed the volume of the whole brain. We did not estimate pooled sensitivity and specificity for the volume of these two regions because available data were sparse and heterogeneous. We could not statistically evaluate the volumes of the lateral temporal lobe, amygdala, medial temporal gyrus, or cortical grey matter assessed in small individual studies. We found no evidence of a difference between studies in the accuracy of the total hippocampal volume with regards to duration of follow-up or age of participants, but the manual MRI technique was superior to automatic techniques in mixed (mostly indirect) comparisons. We did not assess the relative accuracy of the volumes of different brain regions measured by MRI because only indirect comparisons were available, studies were heterogeneous, and the overall accuracy of all regions was moderate.

AUTHORS' CONCLUSIONS

The volume of hippocampus or medial temporal lobe, the most studied brain regions, showed low sensitivity and specificity and did not qualify structural MRI as a stand-alone add-on test for an early diagnosis of dementia due to Alzheimer's disease in people with MCI. This is consistent with international guidelines, which recommend imaging to exclude non-degenerative or surgical causes of cognitive impairment and not to diagnose dementia due to Alzheimer's disease. In view of the low quality of most of the included studies, the findings of this review should be interpreted with caution. Future research should not focus on a single biomarker, but rather on combinations of biomarkers to improve an early diagnosis of Alzheimer's disease dementia.

Regional brain atrophy predicts time to conversion to Alzheimer's disease, dependent on baseline volume

A key question for the design of clinical trials for Alzheimer's disease (AD) is whether the timing of conversion from mild cognitive impairment (MCI) to AD can be predicted. This is also an important question for the clinical management of MCI. This study aims to address this question by exploring the contribution of baseline brain volume and annual volume change, using Cox regression, in predicting the time to conversion. Individuals with MCI, who converted to AD (n = 198), reverted to normal (n = 38), or remained stable (n = 96) for at least five years, were included in this study. The results revealed that the volumes of all the brain areas considered were predictive of the time to conversion from MCI to AD. Annual change in volume was also predictive of the time to conversion but only when initial volumes were above a certain threshold. This is important because it suggests that reduction in atrophy rate, which is the outcome of some clinical trials, is not inevitably associated with delay in conversion from MCI to AD.

Visualizing Alzheimer's disease progression in low dimensional manifolds

While tomographic neuroimaging data is information rich, objective, and with high sensitivity in the study of brain diseases such as Alzheimer's disease (AD), its direct use in clinical practice and in regulated clinical trial (CT) still has many challenges. Taking CT as an example, unless the relevant policy and the perception of the primary outcome measures change, the need to construct univariate indices (out of the 3-D imaging data) to serve as CT's primary outcome measures will remain the focus of active research. More relevant to this current study, an overall global index that summarizes multiple complicated features from neuroimages should be developed in order to provide high diagnostic accuracy and sensitivity in tracking AD progression over time in clinical setting. Such index should also be practically intuitive and logically explainable to patients and their families. In this research, we propose a new visualization tool, derived from the manifold-based nonlinear dimension reduction of brain MRI features, to track AD progression over time. In specific, we investigate the locally linear embedding (LLE) method using a dataset from Alzheimer's Disease Neuroimaging Initiative (ADNI), which includes the longitudinal MRIs from 562 subjects. About 20% of them progressed to the next stage of dementia. Using only the baseline data of cognitively unimpaired (CU) and AD subjects, LLE reduces the feature dimension to two and a subject's AD progression path can be plotted in this low dimensional LLE feature space. In addition, the likelihood of being categorized to AD is indicated by color. This LLE map is a new data visualization tool that can assist in tracking AD progression over time.

Multicenter Resting State Functional Connectivity in Prodromal and Dementia Stages of Alzheimer's Disease

BACKGROUND

Alterations of intrinsic networks from resting state fMRI (rs-fMRI) have been suggested as functional biomarkers of Alzheimer's disease (AD).

OBJECTIVE

To determine the diagnostic accuracy of multicenter rs-fMRI for prodromal and preclinical stages of AD.

METHODS

We determined rs-fMRI functional connectivity based on Pearson's correlation coefficients and amplitude of low-frequency fluctuation in people with subjective cognitive decline, people with mild cognitive impairment, and people with AD dementia compared with healthy controls. We used data of 247 participants of the prospective DELCODE study, a longitudinal multicenter observational study, imposing a unified fMRI acquisition protocol across sites. We determined cross-validated discrimination accuracy based on penalized logistic regression to account for multicollinearity of predictors.

RESULTS

Resting state functional connectivity reached significant cross-validated group discrimination only for the comparison of AD dementia cases with healthy controls, but not for the other diagnostic groups. AD dementia cases showed alterations in a large range of intrinsic resting state networks, including the default mode and salience networks, but also executive and language networks. When groups were stratified according to their CSF amyloid status that was available in a subset of cases, diagnostic accuracy was increased for amyloid positive mild cognitive impairment cases compared with amyloid negative controls, but still inferior to the accuracy of hippocampus volume.

CONCLUSION

Even when following a strictly harmonized data acquisition protocol and rigorous scan quality control, widely used connectivity measures of multicenter rs-fMRI do not reach levels of diagnostic accuracy sufficient for a useful biomarker in prodromal stages of AD.

Diffusion tensor tractography in children with sensory processing disorder: Potentials for devising machine learning classifiers

The "sensory processing disorder" (SPD) refers to brain's inability to organize sensory input for appropriate use. In this study, we determined the diffusion tensor imaging (DTI) microstructural and connectivity correlates of SPD, and apply machine learning algorithms for identification of children with SPD based on DTI/tractography metrics. A total of 44 children with SPD and 41 typically developing children (TDC) were prospectively recruited and scanned. In addition to fractional anisotropy (FA), mean diffusivity (MD), and radial diffusivity (RD), we applied probabilistic tractography to generate edge density (ED) and track density (TD) from DTI maps. For identification of children with SPD, accurate classification rates from a combination of DTI microstructural (FA, MD, AD, and RD), connectivity (TD) and connectomic (ED) metrics with different machine learning algorithms - including naïve Bayes, random forest, support vector machine, and neural networks - were determined. In voxel-wise analysis, children with SPD had lower FA, ED, and TD but higher MD and RD compared to TDC - predominantly in posterior white matter tracts including posterior corona radiata, posterior thalamic radiation, and posterior body and splenium of corpus callosum. In stepwise penalized logistic regression, the only independent variable distinguishing children with SPD from TDC was the average TD in the splenium (p < 0.001). Among different combinations of machine learning algorithms and DTI/connectivity metrics, random forest models using tract-based TD yielded the highest accuracy in classification of SPD - 77.5% accuracy, 73.8% sensitivity, and 81.6% specificity. Our findings demonstrate impaired microstructural and connectivity/connectomic integrity in children with SPD, predominantly in posterior white matter tracts, and with reduced TD of the splenium of corpus callosum as the most distinctive pattern. Applying machine learning algorithms, these connectivity metrics can be used to devise novel imaging biomarkers for neurodevelopmental disorders.

Forty years of structural brain imaging in mental disorders: is it clinically useful or not?

Structural brain imaging was introduced into routine clinical practice more than 40 years ago with the hope that it would support the diagnosis and treatment of mental disorders. It is now widely used to exclude organic brain disease (eg, brain tumors, cardiovascular, and inflammatory processes) in mental disorders. However, questions have been raised about whether structural brain imaging is still needed today and whether it could also be clinically useful to apply new biostatistical methods, such as machine learning. Therefore, the current paper not only reviews structural findings in Alzheimer disease, depression, bipolar disorder, and schizophrenia but also discusses the role of structural imaging in supporting diagnostic, prognostic, and therapeutic processes in mental disorders. Thus, it attempts to answer the questions whether, after four decades of use, structural brain imaging is clinically useful in mental disorders or whether it will become so in the future.

The relevance of hippocampal subfield integrity and clock drawing test performance for the diagnosis of Alzheimer's disease and mild cognitive impairment

OBJECTIVES

The clock drawing test (CDT) is one of the worldwide most used screening tests for Alzheimer's disease (AD). MRI studies have identified temporo-parietal regions being involved in CDT impairment. However, the contributions of specific hippocampal subfields and adjacent extrahippocampal structures to CDT performance in AD and mild cognitive impairment (MCI) have not been investigated so far. It is unclear whether morphological alterations or CDT score, or a combination of both, are able to predict AD.

METHODS

38 AD patients, 38 MCI individuals and 31 healthy controls underwent neuropsychological assessment and MRI at 3 Tesla. FreeSurfer 5.3 was used to perform hippocampal parcellation. We used a collection of statistical methods to better understand the relationship between CDT and hippocampal formation. We also tested the clinical feasibility of this relationship when predicting AD.

RESULTS

Impaired CDT performance in AD was associated with widespread atrophy of the cornu ammonis, presubiculum, and subiculum, whereas MCI subjects showed CDT-related alterations of the CA4-dentate gyrus and subiculum. CDT correlates in AD and MCI showed regional and quantitative overlap. Importantly, CDT score was the best predictor of AD.

CONCLUSIONS

Our findings lend support for an involvement of different hippocampal subfields in impaired CDT performance in AD and MCI. CDT seems to be more efficient than subfield imaging for predicting AD.

Instance-Based Representation Using Multiple Kernel Learning for Predicting Conversion to Alzheimer Disease

The early detection of Alzheimer’s disease and quantification of its progression poses multiple difficulties for machine learning algorithms. Two of the most relevant issues are related to missing data and results interpretability. To deal with both issues, we introduce a methodology to predict conversion of mild cognitive impairment patients to Alzheimer’s from structural brain MRI volumes. First, we use morphological measures of each brain structure to build an instance-based feature mapping that copes with missed follow-up visits. Then, the extracted multiple feature mappings are combined into a single representation through the convex combination of reproducing kernels. The weighting parameters per structure are tuned based on the maximization of the centered-kernel alignment criterion. We evaluate the proposed methodology on a couple of well-known classification machines employing the ADNI database devoted to assessing the combined prognostic value of several AD biomarkers. The obtained experimental results show that our proposed method of Instance-based representation using multiple kernel learning enables detecting mild cognitive impairment as well as predicting conversion to Alzheimers disease within three years from the initial screening. Besides, the brain structures with larger combination weights are directly related to memory and cognitive functions.

Discrimination and conversion prediction of mild cognitive impairment using convolutional neural networks

Background

Recently, studies have demonstrated that machine learning techniques, particularly cutting-edge deep learning technology, have achieved significant progression on the classification of Alzheimer's disease (AD) and its prodromal phase, mild cognitive impairment (MCI). Moreover, accurate prediction of the progress and the conversion risk from MCI to probable AD has been of great importance in clinical application.

Methods

In this study, the baseline MR images and follow-up information during 3 years of 150 normal controls (NC), 150 patients with stable MCI (sMCI) and 157 converted MCI (cMCI) were collected from the Alzheimer's Disease Neuroimaging Initiative (ADNI). The deep convolutional neural networks (CNNs) were adopted to distinguish different stages of MCI from the NC group, and predict the conversion time from MCI to AD. Two CNN architectures including GoogleNet and CaffeNet were explored and evaluated in multiple classifications and estimations of conversion risk using transfer learning from pre-trained ImageNet (via fine-tuning) and five-fold cross-validation. A novel data augmentation approach using random views aggregation was applied to generate abundant image patches from the original MR scans.

Results

The GoogleNet acquired accuracies with 97.58%, 67.33% and 84.71% in three-way discrimination among the NC, sMCI and cMCI groups respectively, whereas the CaffeNet obtained promising accuracies of 98.71%, 72.04% and 92.35% in the NC, sMCI and cMCI classifications. Furthermore, the accuracy measures of conversion risk of patients with cMCI ranged from 71.25% to 83.25% in different time points using GoogleNet, whereas the CaffeNet achieved remarkable accuracy measures from 95.42% to 97.01% in conversion risk prediction.

Conclusions

The experimental results demonstrated that the proposed methods had prominent capability in classification among the 3 groups such as sMCI, cMCI and NC, and exhibited significant ability in conversion risk prediction of patients with MCI.

Role of GTPases in the Regulation of Mitochondrial Dynamics in Alzheimer's Disease and CNS-Related Disorders

Data obtained from several studies have shown that mitochondria are involved and play a central role in the progression of several distinct pathological conditions. Morphological alterations and disruptions on the functionality of mitochondria may be related to metabolic and energy deficiency in neurons in a neurodegenerative disorder. Several recent studies demonstrate the linkage between neurodegeneration and mitochondrial dynamics in the spectrum of a promising era called precision mitochondrial medicine. In this review paper, an analysis of the correlation between mitochondria, Alzheimer's disease, and other central nervous system (CNS)-related disorders like the Parkinson's disease and the autism spectrum disorder is under discussion. The role of GTPases like the mfn1, mfn2, opa1, and dlp1 in mitochondrial fission and fusion is also under investigation, influencing mitochondrial population and leading to oxidative stress and neuronal damage.

Sustainable method for Alzheimer dementia prediction in mild cognitive impairment: Electroencephalographic connectivity and graph theory combined with apolipoprotein E

OBJECTIVE

Mild cognitive impairment (MCI) is a condition intermediate between physiological brain aging and dementia. Amnesic-MCI (aMCI) subjects progress to dementia (typically to Alzheimer-Dementia = AD) at an annual rate which is 20 times higher than that of cognitively intact elderly. The present study aims to investigate whether EEG network Small World properties (SW) combined with Apo-E genotyping, could reliably discriminate aMCI subjects who will convert to AD after approximately a year.

METHODS

145 aMCI subjects were divided into two sub-groups and, according to the clinical follow-up, were classified as Converted to AD (C-MCI, 71) or Stable (S-MCI, 74).

RESULTS

Results showed significant differences in SW in delta, alpha1, alpha2, beta2, gamma bands, with C-MCI in the baseline similar to AD. Receiver Operating Characteristic(ROC) curve, based on a first-order polynomial regression of SW, showed 57% sensitivity, 66% specificity and 61% accuracy(area under the curve: AUC=0.64). In 97 out of 145 MCI, Apo-E allele testing was also available. Combining this genetic risk factor with Small Word EEG, results showed: 96.7% sensitivity, 86% specificity and 91.7% accuracy(AUC=0.97). Moreover, using only the Small World values in these 97 subjects, the ROC showed an AUC of 0.63; the resulting classifier presented 50% sensitivity, 69% specificity and 59.6% accuracy. When different types of EEG analysis (power density spectrum) were tested, the accuracy levels were lower (68.86%).

INTERPRETATION

Concluding, this innovative EEG analysis, in combination with a genetic test (both low-cost and widely available), could evaluate on an individual basis with great precision the risk of MCI progression. This evaluation could then be used to screen large populations and quickly identify aMCI in a prodromal stage of dementia. Ann Neurol 2018 Ann Neurol 2018;84:302-314.

Reproducible evaluation of classification methods in Alzheimer's disease: Framework and application to MRI and PET data

A large number of papers have introduced novel machine learning and feature extraction methods for automatic classification of Alzheimer's disease (AD). However, while the vast majority of these works use the public dataset ADNI for evaluation, they are difficult to reproduce because different key components of the validation are often not readily available. These components include selected participants and input data, image preprocessing and cross-validation procedures. The performance of the different approaches is also difficult to compare objectively. In particular, it is often difficult to assess which part of the method (e.g. preprocessing, feature extraction or classification algorithms) provides a real improvement, if any. In the present paper, we propose a framework for reproducible and objective classification experiments in AD using three publicly available datasets (ADNI, AIBL and OASIS). The framework comprises: i) automatic conversion of the three datasets into a standard format (BIDS); ii) a modular set of preprocessing pipelines, feature extraction and classification methods, together with an evaluation framework, that provide a baseline for benchmarking the different components. We demonstrate the use of the framework for a large-scale evaluation on 1960 participants using T1 MRI and FDG PET data. In this evaluation, we assess the influence of different modalities, preprocessing, feature types (regional or voxel-based features), classifiers, training set sizes and datasets. Performances were in line with the state-of-the-art. FDG PET outperformed T1 MRI for all classification tasks. No difference in performance was found for the use of different atlases, image smoothing, partial volume correction of FDG PET images, or feature type. Linear SVM and L2-logistic regression resulted in similar performance and both outperformed random forests. The classification performance increased along with the number of subjects used for training. Classifiers trained on ADNI generalized well to AIBL and OASIS. All the code of the framework and the experiments is publicly available: general-purpose tools have been integrated into the Clinica software (www.clinica.run) and the paper-specific code is available at: https://gitlab.icm-institute.org/aramislab/AD-ML.

Quasi-Periodic Patterns of Neural Activity improve Classification of Alzheimer's Disease in Mice

Resting state (rs)fMRI allows measurement of brain functional connectivity and has identified default mode (DMN) and task positive (TPN) network disruptions as promising biomarkers for Alzheimer's disease (AD). Quasi-periodic patterns (QPPs) of neural activity describe recurring spatiotemporal patterns that display DMN with TPN anti-correlation. We reasoned that QPPs could provide new insights into AD network dysfunction and improve disease diagnosis. We therefore used rsfMRI to investigate QPPs in old TG2576 mice, a model of amyloidosis, and age-matched controls. Multiple QPPs were determined and compared across groups. Using linear regression, we removed their contribution from the functional scans and assessed how they reflected functional connectivity. Lastly, we used elastic net regression to determine if QPPs improved disease classification. We present three prominent findings: (1) Compared to controls, TG2576 mice were marked by opposing neural dynamics in which DMN areas were anti-correlated and displayed diminished anti-correlation with the TPN. (2) QPPs reflected lowered DMN functional connectivity in TG2576 mice and revealed significantly decreased DMN-TPN anti-correlations. (3) QPP-derived measures significantly improved classification compared to conventional functional connectivity measures. Altogether, our findings provide insight into the neural dynamics of aberrant network connectivity in AD and indicate that QPPs might serve as a translational diagnostic tool.

Individualized Prediction of Reading Comprehension Ability Using Gray Matter Volume

Reading comprehension is a crucial reading skill for learning and putatively contains 2 key components: reading decoding and linguistic comprehension. Current understanding of the neural mechanism underlying these reading comprehension components is lacking, and whether and how neuroanatomical features can be used to predict these 2 skills remain largely unexplored. In the present study, we analyzed a large sample from the Human Connectome Project (HCP) dataset and successfully built multivariate predictive models for these 2 skills using whole-brain gray matter volume features. The results showed that these models effectively captured individual differences in these 2 skills and were able to significantly predict these components of reading comprehension for unseen individuals. The strict cross-validation using the HCP cohort and another independent cohort of children demonstrated the model generalizability. The identified gray matter regions contributing to the skill prediction consisted of a wide range of regions covering the putative reading, cerebellum, and subcortical systems. Interestingly, there were gender differences in the predictive models, with the female-specific model overestimating the males' abilities. Moreover, the identified contributing gray matter regions for the female-specific and male-specific models exhibited considerable differences, supporting a gender-dependent neuroanatomical substrate for reading comprehension.

Cerebral glucose metabolic prediction from amnestic mild cognitive impairment to Alzheimer's dementia: a meta-analysis

Brain ¹⁸F-fluorodeoxyglucose positron emission tomography (FDG-PET) has been utilized to monitor disease conversion from amnestic mild cognitive impairment (aMCI) to Alzheimer’s dementia (AD). However, the conversion patterns of FDG-PET metabolism across studies are not conclusive. We conducted a voxel-wise meta-analysis using Seed-based d Mapping that included 10 baseline voxel-wise FDG-PET comparisons between 93 aMCI converters and 129 aMCI non-converters from nine longitudinal studies. The most robust and reliable metabolic alterations that predicted conversion from aMCI to AD were localized in the left posterior cingulate cortex (PCC)/precuneus. Furthermore, meta-regression analyses indicated that baseline mean age and severity of cognitive impairment, and follow-up duration were significant moderators for metabolic alterations in aMCI converters. Our study revealed hypometabolism in the left PCC/precuneus as an early feature in the development of AD. This finding has important implications in understanding the neural substrates for AD conversion and could serve as a potential imaging biomarker for early detection of AD as well as for tracking disease progression at the predementia stage.

Efficient ℓ0 -norm feature selection based on augmented and penalized minimization

Advances in high-throughput technologies in genomics and imaging yield unprecedentedly large numbers of prognostic biomarkers. To accommodate the scale of biomarkers and study their association with disease outcomes, penalized regression is often used to identify important biomarkers. The ideal variable selection procedure would search for the best subset of predictors, which is equivalent to imposing an ℓ0 -penalty on the regression coefficients. Since this optimization is a nondeterministic polynomial-time hard (NP-hard) problem that does not scale with number of biomarkers, alternative methods mostly place smooth penalties on the regression parameters, which lead to computationally feasible optimization problems. However, empirical studies and theoretical analyses show that convex approximation of ℓ0 -norm (eg, ℓ1 ) does not outperform their ℓ0 counterpart. The progress for ℓ0 -norm feature selection is relatively slower, where the main methods are greedy algorithms such as stepwise regression or orthogonal matching pursuit. Penalized regression based on regularizing ℓ0 -norm remains much less explored in the literature. In this work, inspired by the recently popular augmenting and data splitting algorithms including alternating direction method of multipliers, we propose a 2-stage procedure for ℓ0 -penalty variable selection, referred to as augmented penalized minimization-L0 (APM-L0 ). The APM-L0 targets ℓ0 -norm as closely as possible while keeping computation tractable, efficient, and simple, which is achieved by iterating between a convex regularized regression and a simple hard-thresholding estimation. The procedure can be viewed as arising from regularized optimization with truncated ℓ1 norm. Thus, we propose to treat regularization parameter and thresholding parameter as tuning parameters and select based on cross-validation. A 1-step coordinate descent algorithm is used in the first stage to significantly improve computational efficiency. Through extensive simulation studies and real data application, we demonstrate superior performance of the proposed method in terms of selection accuracy and computational speed as compared to existing methods. The proposed APM-L0 procedure is implemented in the R-package APML0.

Multimodal Classification of Mild Cognitive Impairment Based on Partial Least Squares

In recent years, increasing attention has been given to the identification of the conversion of mild cognitive impairment (MCI) to Alzheimer's disease (AD). Brain neuroimaging techniques have been widely used to support the classification or prediction of MCI. The present study combined magnetic resonance imaging (MRI), 18F-fluorodeoxyglucose PET (FDG-PET), and 18F-florbetapir PET (florbetapir-PET) to discriminate MCI converters (MCI-c, individuals with MCI who convert to AD) from MCI non-converters (MCI-nc, individuals with MCI who have not converted to AD in the follow-up period) based on the partial least squares (PLS) method. Two types of PLS models (informed PLS and agnostic PLS) were built based on 64 MCI-c and 65 MCI-nc from the Alzheimer's Disease Neuroimaging Initiative (ADNI) database. The results showed that the three-modality informed PLS model achieved better classification accuracy of 81.40%, sensitivity of 79.69%, and specificity of 83.08% compared with the single-modality model, and the three-modality agnostic PLS model also achieved better classification compared with the two-modality model. Moreover, combining the three modalities with clinical test score (ADAS-cog), the agnostic PLS model (independent data: florbetapir-PET; dependent data: FDG-PET and MRI) achieved optimal accuracy of 86.05%, sensitivity of 81.25%, and specificity of 90.77%. In addition, the comparison of PLS, support vector machine (SVM), and random forest (RF) showed greater diagnostic power of PLS. These results suggested that our multimodal PLS model has the potential to discriminate MCI-c from the MCI-nc and may therefore be helpful in the early diagnosis of AD.

Patients with Amyloid-Negative Mild Cognitive Impairment have Cortical Hypometabolism but the Hippocampus is Preserved

BACKGROUND

Patients with mild cognitive impairment (MCI) are at risk for Alzheimer's dementia but the presence of amyloid (Aβ) strongly increases this risk. In clinical settings, when Aβ status is not available, different neurodegenerative markers are used to characterize MCI. The accuracy of these markers to discriminate between Aβ-and Aβ+ MCI is not yet determined.

OBJECTIVE

To compare different markers of neurodegeneration in Aβ-and Aβ+ MCI, with an Aβ-elderly control (EC) group.

METHODS

Patients with MCI (n = 39) and EC (n = 28) underwent MRI, 18F-FDG PET, and Aβ PET (18F-flutemetamol). We compared FDG and MRI biomarker values in cortical and hippocampal regions of interest, and using voxel-wise surface maps. We computed ROC curves discriminating between the three groups for each biomarker.

RESULTS

All biomarker values were reduced in Aβ+ MCI compared to EC (p < 0.001). Aβ-MCI had low cortical metabolism (p = 0.002), but hippocampal volume, cortical thickness, and hippocampal metabolism were not significantly different between Aβ-MCI and EC (p > 0.40). Cortical metabolism best discriminated between MCI and EC (AUC = 0.92/0.86, Aβ+/Aβ-) while hippocampal volume best discriminated between Aβ-MCI and Aβ+ MCI (AUC = 0.79).

CONCLUSIONS

Cortical hypometabolism was observed in both Aβ-MCI and Aβ+ MCI whereas hippocampal atrophy was mostly found in Aβ+ MCI. For MCI patients without available Aβ information, hippocampal atrophy is thus more informative about Aβ status than cortical hypometabolism.

Random forest feature selection, fusion and ensemble strategy: Combining multiple morphological MRI measures to discriminate among healhy elderly, MCI, cMCI and alzheimer's disease patients: From the alzheimer's disease neuroimaging initiative (ADNI) database

BACKGROUND

In the era of computer-assisted diagnostic tools for various brain diseases, Alzheimer's disease (AD) covers a large percentage of neuroimaging research, with the main scope being its use in daily practice. However, there has been no study attempting to simultaneously discriminate among Healthy Controls (HC), early mild cognitive impairment (MCI), late MCI (cMCI) and stable AD, using features derived from a single modality, namely MRI.

NEW METHOD

Based on preprocessed MRI images from the organizers of a neuroimaging challenge,³ we attempted to quantify the prediction accuracy of multiple morphological MRI features to simultaneously discriminate among HC, MCI, cMCI and AD. We explored the efficacy of a novel scheme that includes multiple feature selections via Random Forest from subsets of the whole set of features (e.g. whole set, left/right hemisphere etc.), Random Forest classification using a fusion approach and ensemble classification via majority voting. From the ADNI database, 60 HC, 60 MCI, 60 cMCI and 60 CE were used as a training set with known labels. An extra dataset of 160 subjects (HC: 40, MCI: 40, cMCI: 40 and AD: 40) was used as an external blind validation dataset to evaluate the proposed machine learning scheme.

RESULTS

In the second blind dataset, we succeeded in a four-class classification of 61.9% by combining MRI-based features with a Random Forest-based Ensemble Strategy. We achieved the best classification accuracy of all teams that participated in this neuroimaging competition.

COMPARISON WITH EXISTING METHOD(S)

The results demonstrate the effectiveness of the proposed scheme to simultaneously discriminate among four groups using morphological MRI features for the very first time in the literature.

CONCLUSIONS

Hence, the proposed machine learning scheme can be used to define single and multi-modal biomarkers for AD.

18F PET with florbetapir for the early diagnosis of Alzheimer's disease dementia and other dementias in people with mild cognitive impairment (MCI)

BACKGROUND

¹⁸F-florbetapir uptake by brain tissue measured by positron emission tomography (PET) is accepted by regulatory agencies like the Food and Drug Administration (FDA) and the European Medicine Agencies (EMA) for assessing amyloid load in people with dementia. Its added value is mainly demonstrated by excluding Alzheimer's pathology in an established dementia diagnosis. However, the National Institute on Aging and Alzheimer's Association (NIA-AA) revised the diagnostic criteria for Alzheimer's disease and confidence in the diagnosis of mild cognitive impairment (MCI) due to Alzheimer's disease may be increased when using amyloid biomarkers tests like ¹⁸F-florbetapir. These tests, added to the MCI core clinical criteria, might increase the diagnostic test accuracy (DTA) of a testing strategy. However, the DTA of ¹⁸F-florbetapir to predict the progression from MCI to Alzheimer's disease dementia (ADD) or other dementias has not yet been systematically evaluated.

OBJECTIVES

To determine the DTA of the ¹⁸F-florbetapir PET scan for detecting people with MCI at time of performing the test who will clinically progress to ADD, other forms of dementia (non-ADD), or any form of dementia at follow-up.

SEARCH METHODS

This review is current to May 2017. We searched MEDLINE (OvidSP), Embase (OvidSP), PsycINFO (OvidSP), BIOSIS Citation Index (Thomson Reuters Web of Science), Web of Science Core Collection, including the Science Citation Index (Thomson Reuters Web of Science) and the Conference Proceedings Citation Index (Thomson Reuters Web of Science), LILACS (BIREME), CINAHL (EBSCOhost), ClinicalTrials.gov (https://clinicaltrials.gov), and the World Health Organization International Clinical Trials Registry Platform (WHO ICTRP) (http://www.who.int/ictrp/search/en/). We also searched ALOIS, the Cochrane Dementia & Cognitive Improvement Group's specialised register of dementia studies (http://www.medicine.ox.ac.uk/alois/). We checked the reference lists of any relevant studies and systematic reviews, and performed citation tracking using the Science Citation Index to identify any additional relevant studies. No language or date restrictions were applied to the electronic searches.

SELECTION CRITERIA

We included studies that had prospectively defined cohorts with any accepted definition of MCI at time of performing the test and the use of ¹⁸F-florbetapir scan to evaluate the DTA of the progression from MCI to ADD or other forms of dementia. In addition, we only selected studies that applied a reference standard for Alzheimer's dementia diagnosis, for example, National Institute of Neurological and Communicative Disorders and Stroke and the Alzheimer's Disease and Related Disorders Association (NINCDS-ADRDA) or Diagnostic and Statistical Manual of Mental Disorders-IV (DSM-IV) criteria.

DATA COLLECTION AND ANALYSIS

We screened all titles and abstracts identified in electronic-database searches. Two review authors independently selected studies for inclusion and extracted data to create two-by-two tables, showing the binary test results cross-classified with the binary reference standard. We used these data to calculate sensitivities, specificities, and their 95% confidence intervals. Two independent assessors performed quality assessment using the QUADAS-2 tool plus some additional items to assess the methodological quality of the included studies.

MAIN RESULTS

We included three studies, two of which evaluated the progression from MCI to ADD, and one evaluated the progression from MCI to any form of dementia.Progression from MCI to ADD was evaluated in 448 participants. The studies reported data on 401 participants with 1.6 years of follow-up and in 47 participants with three years of follow-up. Sixty-one (15.2%) participants converted at 1.6 years follow-up; nine (19.1%) participants converted at three years of follow-up.Progression from MCI to any form of dementia was evaluated in five participants with 1.5 years of follow-up, with three (60%) participants converting to any form of dementia.There were concerns regarding applicability in the reference standard in all three studies. Regarding the domain of flow and timing, two studies were considered at high risk of bias. MCI to ADD;Progression from MCI to ADD in those with a follow-up between two to less than four years had a sensitivity of 67% (95% CI 30 to 93) and a specificity of 71% (95% CI 54 to 85) by visual assessment (n = 47, 1 study).Progression from MCI to ADD in those with a follow-up between one to less than two years had a sensitivity of 89% (95% CI 78 to 95) and a specificity of 58% (95% CI 53 to 64) by visual assessment, and a sensitivity of 87% (95% CI 76 to 94) and a specificity of 51% (95% CI 45 to 56) by quantitative assessment by the standardised uptake value ratio (SUVR)(n = 401, 1 study). MCI to any form of dementia;Progression from MCI to any form of dementia in those with a follow-up between one to less than two years had a sensitivity of 67% (95% CI 9 to 99) and a specificity of 50% (95% CI 1 to 99) by visual assessment (n = 5, 1 study). MCI to any other forms of dementia (non-ADD);There was no information regarding the progression from MCI to any other form of dementia (non-ADD).

AUTHORS' CONCLUSIONS

Although sensitivity was good in one included study, considering the poor specificity and the limited data available in the literature, we cannot recommend routine use of ¹⁸F-florbetapir PET in clinical practice to predict the progression from MCI to ADD.Because of the poor sensitivity and specificity, limited number of included participants, and the limited data available in the literature, we cannot recommend its routine use in clinical practice to predict the progression from MCI to any form of dementia.Because of the high financial costs of ¹⁸F-florbetapir, clearly demonstrating the DTA and standardising the process of this modality are important prior to its wider use.

A comprehensive analysis of resting state fMRI measures to classify individual patients with Alzheimer's disease

Alzheimer's disease (AD) patients show altered patterns of functional connectivity (FC) on resting state functional magnetic resonance imaging (RSfMRI) scans. It is yet unclear which RSfMRI measures are most informative for the individual classification of AD patients. We investigated this using RSfMRI scans from 77 AD patients (MMSE = 20.4 ± 4.5) and 173 controls (MMSE = 27.5 ± 1.8). We calculated i) FC matrices between resting state components as obtained with independent component analysis (ICA), ii) the dynamics of these FC matrices using a sliding window approach, iii) the graph properties (e.g., connection degree, and clustering coefficient) of the FC matrices, and iv) we distinguished five FC states and administered how long each subject resided in each of these five states. Furthermore, for each voxel we calculated v) FC with 10 resting state networks using dual regression, vi) FC with the hippocampus, vii) eigenvector centrality, and viii) the amplitude of low frequency fluctuations (ALFF). These eight measures were used separately as predictors in an elastic net logistic regression, and combined in a group lasso logistic regression model. We calculated the area under the receiver operating characteristic curve plots (AUC) to determine classification performance. The AUC values ranged between 0.51 and 0.84 and the highest were found for the FC matrices (0.82), FC dynamics (0.84) and ALFF (0.82). The combination of all measures resulted in an AUC of 0.85. We show that it is possible to obtain moderate to good AD classification using RSfMRI scans. FC matrices, FC dynamics and ALFF are most discriminative and the combination of all the resting state measures improves classification accuracy slightly.

In vivo staging of regional amyloid deposition

OBJECTIVES

To estimate a regional progression pattern of amyloid deposition from cross-sectional amyloid-sensitive PET data and evaluate its potential for in vivo staging of an individual's amyloid pathology.

METHODS

Multiregional analysis of florbetapir (18F-AV45)-PET data was used to determine individual amyloid distribution profiles in a sample of 667 participants from the Alzheimer's Disease Neuroimaging Initiative cohort, including cognitively normal older individuals (CN) as well as patients with mild cognitive impairment and Alzheimer disease (AD) dementia. The frequency of regional amyloid positivity across CN individuals was used to construct a 4-stage model of progressing amyloid pathology, and individual distribution profiles were used to evaluate the consistency of this hierarchical stage model across the full cohort.

RESULTS

According to a 4-stage model, amyloid deposition begins in temporobasal and frontomedial areas, and successively affects the remaining associative neocortex, primary sensory-motor areas and the medial temporal lobe, and finally the striatum. Amyloid deposition in these brain regions showed a highly consistent hierarchical nesting across participants, where only 2% exhibited distribution profiles that deviated from the staging scheme. The earliest in vivo amyloid stages were mostly missed by conventional dichotomous classification approaches based on global florbetapir-PET signal, but were associated with significantly reduced CSF Aβ42 levels. Advanced in vivo amyloid stages were most frequent in patients with AD and correlated with cognitive impairment in individuals without dementia.

CONCLUSIONS

The highly consistent regional hierarchy of PET-evidenced amyloid deposition across participants resembles neuropathologic observations and suggests a predictable regional sequence that may be used to stage an individual's progress of amyloid pathology in vivo.

Incremental value of biomarker combinations to predict progression of mild cognitive impairment to Alzheimer's dementia

Background

The progression of mild cognitive impairment (MCI) to Alzheimer’s disease (AD) dementia can be predicted by cognitive, neuroimaging, and cerebrospinal fluid (CSF) markers. Since most biomarkers reveal complementary information, a combination of biomarkers may increase the predictive power. We investigated which combination of the Mini-Mental State Examination (MMSE), Clinical Dementia Rating (CDR)-sum-of-boxes, the word list delayed free recall from the Consortium to Establish a Registry of Dementia (CERAD) test battery, hippocampal volume (HCV), amyloid-beta_1–42 (Aβ42), amyloid-beta_1–40 (Aβ40) levels, the ratio of Aβ42/Aβ40, phosphorylated tau, and total tau (t-Tau) levels in the CSF best predicted a short-term conversion from MCI to AD dementia.

Methods

We used 115 complete datasets from MCI patients of the “Dementia Competence Network”, a German multicenter cohort study with annual follow-up up to 3 years. MCI was broadly defined to include amnestic and nonamnestic syndromes. Variables known to predict progression in MCI patients were selected a priori. Nine individual predictors were compared by receiver operating characteristic (ROC) curve analysis. ROC curves of the five best two-, three-, and four-parameter combinations were analyzed for significant superiority by a bootstrapping wrapper around a support vector machine with linear kernel. The incremental value of combinations was tested for statistical significance by comparing the specificities of the different classifiers at a given sensitivity of 85%.

Results

Out of 115 subjects, 28 (24.3%) with MCI progressed to AD dementia within a mean follow-up period of 25.5 months. At baseline, MCI-AD patients were no different from stable MCI in age and gender distribution, but had lower educational attainment. All single biomarkers were significantly different between the two groups at baseline. ROC curves of the individual predictors gave areas under the curve (AUC) between 0.66 and 0.77, and all single predictors were statistically superior to Aβ40. The AUC of the two-parameter combinations ranged from 0.77 to 0.81. The three-parameter combinations ranged from AUC 0.80–0.83, and the four-parameter combination from AUC 0.81–0.82. None of the predictor combinations was significantly superior to the two best single predictors (HCV and t-Tau). When maximizing the AUC differences by fixing sensitivity at 85%, the two- to four-parameter combinations were superior to HCV alone.

Conclusion

A combination of two biomarkers of neurodegeneration (e.g., HCV and t-Tau) is not superior over the single parameters in identifying patients with MCI who are most likely to progress to AD dementia, although there is a gradual increase in the statistical measures across increasing biomarker combinations. This may have implications for clinical diagnosis and for selecting subjects for participation in clinical trials.

Multi-modal discriminative dictionary learning for Alzheimer's disease and mild cognitive impairment

BACKGROUND AND OBJECTIVE

The differentiation of mild cognitive impairment (MCI), which is the prodromal stage of Alzheimer's disease (AD), from normal control (NC) is important as the recent research emphasis on early pre-clinical stage for possible disease abnormality identification, intervention and even possible prevention.

METHODS

The current study puts forward a multi-modal supervised within-class-similarity discriminative dictionary learning algorithm (SCDDL) we introduced previously for distinguishing MCI from NC. The proposed new algorithm was based on weighted combination and named as multi-modality SCDDL (mSCDDL). Structural magnetic resonance imaging (sMRI), fluorodeoxyglucose (FDG) positron emission tomography (PET) and florbetapir PET data of 113 AD patients, 110 MCI patients and 117 NC subjects from the Alzheimer's disease Neuroimaging Initiative database were adopted for classification between MCI and NC, as well as between AD and NC.

RESULTS

Adopting mSCDDL, the classification accuracy achieved 98.5% for AD vs. NC and 82.8% for MCI vs. NC, which were superior to or comparable with the results of some other state-of-the-art approaches as reported in recent multi-modality publications.

CONCLUSIONS

The mSCDDL procedure was a promising tool in assisting early diseases diagnosis using neuroimaging data.

The diabetic brain and cognition

The prevalence of both Alzheimer's disease (AD) and vascular dementia (VaD) is increasing with the aging of the population. Studies from the last several years have shown that people with diabetes have an increased risk for dementia and cognitive impairment. Therefore, the authors of this consensus review tried to elaborate on the role of diabetes, especially diabetes type 2 (T2DM) in both AD and VaD. Based on the clinical and experimental work of scientists from 18 countries participating in the International Congress on Vascular Disorders and on literature search using PUBMED, it can be concluded that T2DM is a risk factor for both, AD and VaD, based on a pathology of glucose utilization. This pathology is the consequence of a disturbance of insulin-related mechanisms leading to brain insulin resistance. Although the underlying pathological mechanisms for AD and VaD are different in many aspects, the contribution of T2DM and insulin resistant brain state (IRBS) to cerebrovascular disturbances in both disorders cannot be neglected. Therefore, early diagnosis of metabolic parameters including those relevant for T2DM is required. Moreover, it is possible that therapeutic options utilized today for diabetes treatment may also have an effect on the risk for dementia. T2DM/IRBS contribute to pathological processes in AD and VaD.

Biomarkers for Alzheimer Disease: Classical and Novel Candidates' Review

The biomarkers may be useful for predictive diagnosis of Alzheimer's disease (AD). The current challenge is to diagnose it in its preclinical phase. The combination of cerebrospinal fluid (CSF) biomarkers and imaging has been investigated extensively for a number of years. It can provide an increased diagnostic accuracy. This review discusses the contribution of classical biomarkers to predict AD and highlights novel candidates identified as potential markers for AD. We referred to the electronic databases PubMed/Medline and Web of Science to search for articles that were published until February 2016. Sixty-two records were included in qualitative synthesis. In the first section, the results show the contribution of biomarkers to predict and track AD considered as classical biomarkers. In the second section, the results highlight the involvement of novel candidates that should be considered for future evaluation in the characterization of the AD progression. Reported findings open prospect to define noninvasive biomarkers to predict AD before symptoms onset.

Early identification of MCI converting to AD: a FDG PET study

PURPOSE

Mild cognitive impairment (MCI) is a transitional pathological stage between normal ageing (NA) and Alzheimer's disease (AD). Although subjects with MCI show a decline at different rates, some individuals remain stable or even show an improvement in their cognitive level after some years. We assessed the accuracy of FDG PET in discriminating MCI patients who converted to AD from those who did not.

METHODS

FDG PET was performed in 42 NA subjects, 27 MCI patients who had not converted to AD at 5 years (nc-MCI; mean follow-up time 7.5 ± 1.5 years), and 95 MCI patients who converted to AD within 5 years (MCI-AD; mean conversion time 1.8 ± 1.1 years). Relative FDG uptake values in 26 meta-volumes of interest were submitted to ANCOVA and support vector machine analyses to evaluate regional differences and discrimination accuracy.

RESULTS

The MCI-AD group showed significantly lower FDG uptake values in the temporoparietal cortex than the other two groups. FDG uptake values in the nc-MCI group were similar to those in the NA group. Support vector machine analysis discriminated nc-MCI from MCI-AD patients with an accuracy of 89% (AUC 0.91), correctly detecting 93% of the nc-MCI patients.

CONCLUSION

In MCI patients not converting to AD within a minimum follow-up time of 5 years and MCI patients converting within 5 years, baseline FDG PET and volume-based analysis identified those who converted with an accuracy of 89%. However, further analysis is needed in patients with amnestic MCI who convert to a dementia other than AD.