Search

We present a comparative split-half resampling analysis of various data driven feature selection and classification methods for the whole brain voxel-based classification analysis of anatomical magnetic resonance images. We compared support vector machines (SVMs), with or without filter based feature selection, several embedded feature selection methods and stability selection. While comparisons of the accuracy of various classification methods have been reported previously, the variability of the out-of-training sample classification accuracy and the set of selected features due to independent training and test sets have not been previously addressed in a brain imaging context. We studied two classification problems: 1) Alzheimer's disease (AD) vs. normal control (NC) and 2) mild cognitive impairment (MCI) vs. NC classification. In AD vs. NC classification, the variability in the test accuracy due to the subject sample did not vary between different methods and exceeded the variability due to different classifiers. In MCI vs. NC classification, particularly with a large training set, embedded feature selection methods outperformed SVM-based ones with the difference in the test accuracy exceeding the test accuracy variability due to the subject sample. The filter and embedded methods produced divergent feature patterns for MCI vs. NC classification that suggests the utility of the embedded feature selection for this problem when linked with the good generalization performance. The stability of the feature sets was strongly correlated with the number of features selected, weakly correlated with the stability of classification accuracy, and uncorrelated with the average classification accuracy.

Early identification and treatment of Alzheimer's disease (AD) is hampered by the lack of easily accessible biomarkers. Currently available fluid biomarkers of AD provide indications of the disease stage; however, these are measured in the cerebrospinal fluid, requiring invasive procedures, which are not applicable at the population level. Thus, gene expression profiling of blood provides a viable alternative as a way to screen individuals at risk of AD. Previous studies have shown that despite the limited permeability of the blood-brain barriers, expression profiles of blood genes can be used for the diagnosis and prognosis of several brain disorders. Here, we propose a new approach to pathway analysis of blood gene expression profiles to classify healthy (control [CTL]), mildly cognitively impaired (mild cognitive impairment [MCI]; preclinical stage of AD), and AD subjects. In the pathway analysis, gene expression data are mapped to pathway scores according to a predefined gene set instead of considering each gene separately. The robustness of the analysis enables detection of weak differences between groups owing to the inherent dimension reduction. Our proposed method for pathway analysis takes advantage of linear discriminant analysis for identifying a linear combination of features best separating groups of subjects within each gene set. The gene expression data were retrieved from Gene Expression Omnibus (batch 1: GSE63060; batch 2: GSE63061). Predefined gene sets for pathway analysis were obtained from the Broad Institute Collection of Curated Pathways. The method achieved a 10-fold cross-validated area under receiver operating characteristic curve of 0.84 for classification of AD versus CTL and 0.80 for classification of mild cognitive impairment versus CTL. These results reveal the good potential of blood-based biomarkers for assisting early diagnosis and disease monitoring of AD.

Machine learning approaches have been widely used for the identification of neuropathology from neuroimaging data. However, these approaches require large samples and suffer from the challenges associated with multi-site, multi-protocol data. We propose a novel approach to address these challenges, and demonstrate its usefulness with the Autism Brain Imaging Data Exchange (ABIDE) database. We predict symptom severity based on cortical thickness measurements from 156 individuals with autism spectrum disorder (ASD) from four different sites. The proposed approach consists of two main stages: a domain adaptation stage using partial least squares regression to maximize the consistency of imaging data across sites; and a learning stage combining support vector regression for regional prediction of severity with elastic-net penalized linear regression for integrating regional predictions into a whole-brain severity prediction. The proposed method performed markedly better than simpler alternatives, better with multi-site than single-site data, and resulted in a considerably higher cross-validated correlation score than has previously been reported in the literature for multi-site data. This demonstration of the utility of the proposed approach for detecting structural brain abnormalities in ASD from the multi-site, multi-protocol ABIDE dataset indicates the potential of designing machine learning methods to meet the challenges of agglomerative data.

Algorithms for computer-aided diagnosis of dementia based on structural MRI have demonstrated high performance in the literature, but are difficult to compare as different data sets and methodology were used for evaluation. In addition, it is unclear how the algorithms would perform on previously unseen data, and thus, how they would perform in clinical practice when there is no real opportunity to adapt the algorithm to the data at hand. To address these comparability, generalizability and clinical applicability issues, we organized a grand challenge that aimed to objectively compare algorithms based on a clinically representative multi-center data set. Using clinical practice as the starting point, the goal was to reproduce the clinical diagnosis. Therefore, we evaluated algorithms for multi-class classification of three diagnostic groups: patients with probable Alzheimer's disease, patients with mild cognitive impairment and healthy controls. The diagnosis based on clinical criteria was used as reference standard, as it was the best available reference despite its known limitations. For evaluation, a previously unseen test set was used consisting of 354 T1-weighted MRI scans with the diagnoses blinded. Fifteen research teams participated with a total of 29 algorithms. The algorithms were trained on a small training set (n = 30) and optionally on data from other sources (e.g., the Alzheimer's Disease Neuroimaging Initiative, the Australian Imaging Biomarkers and Lifestyle flagship study of aging). The best performing algorithm yielded an accuracy of 63.0% and an area under the receiver-operating-characteristic curve (AUC) of 78.8%. In general, the best performances were achieved using feature extraction based on voxel-based morphometry or a combination of features that included volume, cortical thickness, shape and intensity. The challenge is open for new submissions via the web-based framework: http://caddementia.grand-challenge.org. (C) 2015 Elsevier Inc. All rights reserved.

This paper investigates the use of semi-supervised learning (SSL) for predicting Alzheimers Disease (AD) conversion in Mild Cognitive Impairment (MCI) patients based on Magnetic Resonance Imaging (MRI). SSL methods differ from standard supervised learning methods in that they make use of unlabeled data - in this case data from MCI subjects whose final diagnosis is not yet known. We compare two widely used semi-supervised methods (low density separation (LDS) and semi-supervised discriminant analysis (SDA)) to the corresponding supervised methods using real and synthetic MRI data of MCI subjects. With simulated data, using SSL instead of supervised learning led to higher classification performance in certain cases, however, the applicability of semi-supervised methods depended strongly on the data distributions. With real MRI data, the SSL methods achieved significantly better classification performances over supervised methods. Moreover, even using a small number of unlabeled samples improved the AD conversion predictions.