Semi-supervised learning in MCI-to-ad conversion prediction — When is unlabeled data useful?

In this paper, a new neural network model inspired by the biological immune system functions is presented. The model, termed Artificial Immune-Activated Neural Network (AIANN), extracts classification knowledge from a training data set, which is then used to classify input patterns or vectors. The AIANN is based on a neuron activation function whose behavior is conceptually modeled after the chemical bonds between the receptors and epitopes in the biological immune system. The bonding is controlled through an energy measure to ensure accurate recognition. The AIANN model was applied to the segmentation of 3-dimensional magnetic resonance imaging (MRI) data of the brain and a contextual basis was developed for the segmentation problem. Evaluation of the segmentation results was performed using both real MRI data obtained from the Center for Morphometric Analysis at Massachusetts General Hospital and simulated MRI data generated using the McGill University BrainWeb MRI simulator. Experimental results demonstrated that the AIANN model attained higher average results than those obtained using published methods for real MRI data and simulated MRI data, especially at low levels of noise.

The contrast enhancement of Magnetic Resonance Imaging (MRI) data is quite challenging as the noise present in this data also get amplified in this process. Dynamic Stochastic Resonance (DSR) is the technique that utilizes the noise to enhance the contrast of MRI data.The present study proposes the cascaded stochastic resonance, which exploits the properties of modified potential neuron model and quartic bistable model of DSR. The Multi-objective Particle Swarm Optimization (MOPSO) tunes the DSR parameters associated with the cascading of both the models. The MOPSO produces a set of the solution called Pareto front for the maximization of two image quality measures, i.e., contrast enhancement factor and universal image quality index. Further, the maximization of another image quality measure, i.e., anisotropy helps to obtain the optimum enhanced image from the Pareto fronts solution.The present study included the simulated and real MRI data. The results show that the proposed method obtained mean contrast enhancement factor, universal image quality index and anisotropy equal to 1.79, 0.78 and 0.021 respectively. These values are better than those obtained for classical bistable DSR and other conventional contrast enhancement techniques. The proposed algorithm has been tested on real MRI data as well and found valuable in the diagnosis of lacunar infarct and mesial temporal sclerosis.The cascaded DSR based on MOPSO has shown promising results and may be highly beneficial to the diagnosis of different brain pathology.

Fisher discriminant analysis (FDA) is a classic supervised dimensionality reduction method in statistical pattern recognition. FDA can maximize the scatter between different classes, while minimizing the scatter within each class. As it only utilizes the labeled data and ignores the unlabeled data in the analysis process of FDA, it cannot be used to solve the unsupervised learning problems. Its performance is also very poor in dealing with semi-supervised learning problems in some cases. Recently, several semi-supervised learning methods as an extension of FDA have proposed. Most of these methods solve the semi-supervised problem by using a tradeoff parameter that evaluates the ratio of the supervised and unsupervised methods. In this paper, we propose a general semi-supervised dimensionality learning idea for the partially labeled data, namely the reconstruction probability class of labeled and unlabeled data. Based on the probability class optimizes Fisher criterion function, we propose a novel Semi-Supervised Local Fisher Discriminant Analysis (S2LFDA) method. Experimental results on real-world datasets demonstrate its effectiveness compared to the existing similar correlation methods.

In this work, we introduce Permutation Disalignment Index (PDI) as a novel nonlinear, amplitude independent, robust to noise metric of coupling strength between time series, with the aim of applying it to electroencephalographic (EEG) signals recorded longitudinally from Alzheimer's Disease (AD) and Mild Cognitive Impaired (MCI) patients. The goal is to indirectly estimate the connectivity between the cortical areas, through the quantification of the coupling strength between the corresponding EEG signals, in order to find a possible matching with the disease's progression. PDI is first defined and tested on simulated interacting dynamic systems. PDI is then applied to real EEG recorded from 8 amnestic MCI subjects and 7 AD patients, who were longitudinally evaluated at time [Formula: see text]0 and 3 months later (time [Formula: see text]1). At time [Formula: see text]1, 5 out of 8 MCI patients were still diagnosed MCI (stable MCI) whereas the remaining 3 exhibited a conversion from MCI to AD (prodromal AD). PDI was compared to the Spectral Coherence and the Dissimilarity Index. Limited to the size of the analyzed dataset, both Coherence and PDI resulted sensitive to the conversion from MCI to AD, even though only PDI resulted specific. In particular, the intrasubject variability study showed that the three patients who converted to AD exhibited a significantly ([Formula: see text]) increased PDI (reduced coupling strength) in delta and theta bands. As regards Coherence, even though it significantly decreased in the three converted patients, in delta and theta bands, such a behavior was also detectable in one stable MCI patient, in delta band, thus making Coherence not specific. From the Dissimilarity Index point of view, the converted MCI showed no peculiar behavior. PDI significantly increased, in delta and theta bands, specifically in the MCI subjects who converted to AD. The increase of PDI reflects a reduced coupling strength among the brain areas, which is consistent with the expected connectivity reduction associated to AD progression.

Relevancia del efecto de posición serial en el diagnóstico diferencial entre el deterioro cognitivo leve, la demencia de tipo Alzheimer y el envejecimiento normal

Cortical sources of resting EEG rhythms in mild cognitive impairment and subjective memory complaint

In recent years, pathological researches of mild cognitive impairment (MCI) subjects and the Alzheimer's disease (AD) patients have gained a great deal of attention. In this study, we used the voxel-based morphometry (VBM) method to analyze the Magnatic Resonance Imaging (MRI) data of gray matter volumes in 98 normal controls (NCs), 91 AD patients and 113 MCI subjects. The measurements of gray matter volumes were calculated for each of the three groups respectively. We found that compared with MCI subjects, AD patients had further atrophy in the following brain regions: right insula, left hippocampus, right fusiform and bilateral middle temporal gyrus. Compared with NCs, AD patients and MCI subjects shared some abnormal brain regions such as bilateral parahippocampal gyrus, bilateral hippocampus, left amygdala and left fusiform. The results provided additional evidences to support the viewpoint that MCI is the transitional stage between normal aging and AD.

FDG-PET POWER TO PREDICT MEMORY DECLINE IN ALZHEIMER’S DISEASE DEPENDS ON DISEASE PHASE AND AMYLOID AND TAU STATUS

Machine Learning is a modern and actively developing field of computer science, devoted to extracting and estimating dependencies from empirical data. It combines such fields as statistics, optimization theory and artificial intelligence. In practical tasks, the general aim of Machine Learning is to construct algorithms able to generalize and predict in previously unseen situations based on some set of examples. Given some finite information, Machine Learning provides ways to exract knowledge, describe, explain and predict from data. Kernel Methods are one of the most successful branches of Machine Learning. They allow applying linear algorithms with well-founded properties such as generalization ability, to non-linear real-life problems. Support Vector Machine is a well-known example of a kernel method, which has found a wide range of applications in data analysis nowadays. In many practical applications, some additional prior knowledge is often available. This can be the knowledge about the data domain, invariant transformations, inner geometrical structures in data, some properties of the underlying process, etc. If used smartly, this information can provide significant improvement to any data processing algorithm. Thus, it is important to develop methods for incorporating prior knowledge into data-dependent models. The main objective of this thesis is to investigate approaches towards learning with kernel methods using prior knowledge. Invariant learning with kernel methods is considered in more details. In the first part of the thesis, kernels are developed which incorporate prior knowledge on invariant transformations. They apply when the desired transformation produce an object around every example, assuming that all points in the given object share the same class. Different types of objects, including hard geometrical objects and distributions are considered. These kernels were then applied for images classification with Support Vector Machines. Next, algorithms which specifically include prior knowledge are considered. An algorithm which linearly classifies distributions by their domain was developed. It is constructed such that it allows to apply kernels to solve non-linear tasks. Thus, it combines the discriminative power of support vector machines and the well-developed framework of generative models. It can be applied to a number of real-life tasks which include data represented as distributions. In the last part of the thesis, the use of unlabelled data as a source of prior knowledge is considered. The technique of modelling the unlabelled data with a graph is taken as a baseline from semi-supervised manifold learning. For classification problems, we use this apporach for building graph models of invariant manifolds. For regression problems, we use unlabelled data to take into account the inner geometry of the input space. To conclude, in this thesis we developed a number of approaches for incorporating some prior knowledge into kernel methods. We proposed invariant kernels for existing algorithms, developed new algorithms and adapted a technique taken from semi-supervised learning for invariant learning. In all these cases, links with related state-of-the-art approaches were investigated. Several illustrative experiments were carried out on real data on optical character recognition, face image classification, brain-computer interfaces, and a number of benchmark and synthetic datasets.

Comparison of florbetapir positron emission tomography scans with cerebrospinal fluid biomarkers in healthy individuals and people with mild cognitive impairment or Alzheimer's disease dementia

It has been shown that a new procedure (implicit function as squashing time, IFAST) based on artificial neural networks (ANNs) is able to compress eyes-closed resting electroencephalographic (EEG) data into spatial invariants of the instant voltage distributions for an automatic classification of mild cognitive impairment (MCI) and Alzheimer's disease (AD) subjects with classification accuracy of individual subjects higher than 92%. Here we tested the hypothesis that this is the case also for the classification of individual normal elderly (Nold) vs. MCI subjects, an important issue for the screening of large populations at high risk of AD. Eyes-closed resting EEG data (10- 20 electrode montage) were recorded in 171 Nold and in 115 amnesic MCI subjects. The data inputs for the classification by IFAST were the weights of the connections within a non linear auto-associative ANN trained to generate the instant voltage distributions of 60-s artifact free EEG data. The most relevant features were selected and coincidently the dataset was split into two halves for the final binary classification (training and testing) performed by a supervised ANN. The classification of the individual Nold and MCI subjects reached 95.87% of sensitivity and 91.06% of specificity (93.46% of accuracy). These results indicate that IFAST can reliably distinguish eyes-closed resting EEG in individual Nold and MCI subjects, and may be used for large-scale periodic screening of large populations at risk of AD and personalized care.

Introduction Recent work has characterized polysomnographic (PSG) changes in Alzheimer’s Dementia (AD) that correlate with disease severity and precede clinical diagnosis, although these findings are limited by relatively small sample sizes and a lack of studies examining sleep microarchitecture. In this study we examined sleep macro- and microarchitectural features in large cohorts of patients with AD and mild cognitive impairment (MCI). Methods Using a database of over 15,000 clinical PSGs with associated clinical metadata, 125 and 173 subjects were identified with mild AD and MCI, respectively. 260 age and sex matched subjects without neurological disease were identified as healthy controls (HC). Sleep staging results were based on AASM criteria, and micro-architectural features were assessed using an algorithm (Wake EEG Similarity Index, WESI) trained to predict state using spectral data from a unique healthy adult cohort. Results Sleep fragmentation was apparent in both the MCI and AD cohorts (relative to healthy controls), with AD subjects demonstrating substantially more severe PSG disturbances. Healthy controls demonstrated 400 minutes of total sleep time, in contrast to 374 minutes for MCI subjects and 362 minutes for AD subjects. Healthy subjects demonstrated 33 minutes of wake after sleep onset (WASO), compared to 60 and 67 minutes for MCI and AD subjects, respectively. As has been previously reported in smaller cohorts, wakeful bouts and N1 durations were substantially prolonged for AD and (to a lesser extent) MCI subjects, while N2, N3, and REM durations were reduced (with a pronounced reduction in AD N3 percentage). Sleep microarchitectural features assessed using WESI demonstrated that AD and MCI subjects had markedly higher relative wakefulness in all sleep stages except for N3, which showed the same sleep depth as HC. Conclusion We confirm increased sleep fragmentation and loss of deeper sleep stages that correlate to disease severity. Spectral features of all sleep stages except N3 appear more wakeful in AD and MCI subjects, suggesting microarousals or impaired sleep maintenance contribute to sleep pathology. While N3 duration was markedly reduced, depth of sleep as measured by the WESI algorithm was comparable between populations. PSG findings offer promise as early diagnostic and therapeutic targets. Support (if any)

Patients with mild cognitive impairment (MCI) have increased risk to develop Alzheimer disease (AD). In AD increased brain amyloid burden has been demonstrated in vivo with PET using N-methyl-[(11)C]2-(4'-methylaminophenyl)-6-hydroxybenzothiazole ([(11)C]PIB) as a tracer. To investigate whether patients with amnestic MCI would show increased [(11)C]PIB uptake, indicating early AD process. We studied 13 patients with amnestic MCI and 14 control subjects with PET using [(11)C]PIB as tracer. Parametric images were computed by calculating the region-to-cerebellum ratio in each voxel over 60 to 90 minutes. Group differences in [(11)C]PIB uptake were analyzed with statistical parametric mapping (SPM) and automated region-of-interest (ROI) analysis. The SPM analysis showed that patients with MCI had significantly higher [(11)C]PIB uptake vs control subjects in the frontal, parietal, and lateral temporal cortices as well as in the posterior cingulate showing the most prominent differences. These results were supported by the automated ROI analysis in which MCI patients showed in comparison with healthy control subjects increased [(11)C]PIB uptake in the frontal cortex (39% increase from the control mean, p < 0.01), the posterior cingulate (39%, p < 0.01), the parietal (31%, p < 0.01) and lateral temporal (28%, p < 0.001) cortices, putamen (17%, p < 0.05), and caudate (25%, p < 0.05). Individually, in the frontal cortex and posterior cingulate, 8 of 13 patients with MCI had [(11)C]PIB uptake values above 2 SD from the control mean. MCI subjects having at least one APOE epsilon4 allele tended to have higher [(11)C]PIB uptake than MCI subjects without APOE epsilon4. At group level the elevated N-methyl-[(11)C]2-(4'-methylaminophenyl)-6-hydroxybenzothiazole ([(11)C]PIB) uptake in patients with mild cognitive impairment (MCI) resembled that seen in Alzheimer disease (AD). At the individual level, about half of the MCI patients had [(11)C]PIB uptake in the AD range, suggestive of early AD process.

Joint analysis of amyloid and metabolic PET patterns across healthy, mild cognitive impairment (MCI), and Alzheimer's disease (AD) subjects was performed using baseline 18F-florbetapir and 18F-FDG PET of 684 subjects from the ADNI (251 normal, 204 stable MCI, 85 AD converters, and 144 AD). Correlation between regional amyloid and metabolic uptake was measured and predictive value of PET profile regarding AD conversion in cognitively impaired subjects was assessed using survival analysis and support vector machine classification (SVM). The highest correlations were found in the temporal cortex, precuneus, and posterior cingulum. With respect to normal controls, amyloid load increase was diffuse and early in MCI subjects, whereas metabolism decrease occurred later and predominated in temporo-parietal, precuneus, and cingulate cortices. Five-year AD conversion rates in cognitively impaired subjects were 5%, 22%, 42%, and 78% in amyloid-/FDG-, amyloid-/FDG+, amyloid+/FDG-, and amyloid+/FDG+ subjects respectively (mean follow-up 37±14 months). Using SVM, the combination of ADAS-cog score, amyloid PET, and FDG PET yielded better performance in predicting AD conversion (77% accuracy; 58% positive predictive value; 88% negative predictive value) than ADAS-cog (72%; 52%; 86%), amyloid PET (72%; 52%; 87%), and FDG PET (67%; 47%; 84%). This study attests the complementary value of amyloid and FDG PET in MCI assessment and the efficiency of combined cognitive, amyloid, and metabolic scores to predict AD conversion.

Aims: To describe personality traits and their changes in mild cognitive impairment (MCI) and control subjects. Methods: Sixty-three MCI and 90 control subjects were asked to describe their current personality traits by the Structured Interview for the Five-Factor Model (SIFFM). For each subject, a close relative retrospectively assessed these descriptions both as to the previous and current personality traits, using the Revised NEO Personality Inventory, Form R (NEO-PI-R). Results: Self-assessed MCI subjects reported significantly lower scores in the openness dimension than control subjects [F(1, 150) = 9.84, p = 0.002, η_p² = 0.06]. In current observer ratings, MCI subjects had higher scores on neuroticism [F(1, 137) = 7.55, p = 0.007, η_p² = 0.05] and lower ones on extraversion [F(1, 137) = 6.40, p = 0.013, η_p² = 0.04], openness [F(1, 137) = 9.93, p = 0.002, η_p² = 0.07], agreeableness [F(1, 137) = 10.18, p = 0.002, η_p² = 0.07] and conscientiousness [F(1, 137) = 25.96, p < 0.001, η_p² = 0.16]. Previous personality traits discriminated the groups as previous openness [odds ratio (OR) = 0.97, 95% confidence interval (CI) = 0.95-0.99, p = 0.014] and conscientiousness (OR = 0.96, 95% CI 0.94-0.98, p = 0.001) were negatively related to MCI group membership. In MCI subjects, conscientiousness [F(1, 137) = 19.20, p < 0.001, η_p² = 0.12] and extraversion [F(1, 137) = 22.27, p < 0.001, η_p² = 0.14] decreased between previous and current evaluations and neuroticism increased [F(1, 137) = 22.23, p < 0.001, η_p² = 0.14], whereas no significant change was found in control subjects. Conclusions: MCI subjects undergo significant personality changes. Thus, personality assessment may aid the early detection of dementia.