Machine Learning classification of MRI features of Alzheimer's disease and mild cognitive impairment subjects to reduce the sample size in clinical trials

Signal Processing and Multimedia Communications Research Group, School of Computing and Mathematics, University of Plymouth, Plymouth PL4 8AA, UK.
Conference proceedings: ... Annual International Conference of the IEEE Engineering in Medicine and Biology Society. IEEE Engineering in Medicine and Biology Society. Conference 08/2011; 2011:7957-60. DOI: 10.1109/IEMBS.2011.6091962
Source: PubMed


There is a need for objective tools to help clinicians to diagnose Alzheimer's Disease (AD) early and accurately and to conduct Clinical Trials (CTs) with fewer patients. Magnetic Resonance Imaging (MRI) is a promising AD biomarker but no single MRI feature is optimal for all disease stages. Machine Learning classification can address these challenges. In this study, we have investigated the classification of MRI features from AD, Mild Cognitive Impairment (MCI), and control subjects from ADNI with four techniques. The highest accuracy rates for the classification of controls against ADs and MCIs were 89.2% and 72.7%, respectively. Moreover, we used the classifiers to select AD and MCI subjects who are most likely to decline for inclusion in hypothetical CTs. Using the hippocampal volume as an outcome measure, we found that the required group sizes for the CTs were reduced from 197 to 117 AD patients and from 366 to 215 MCI subjects.

12 Reads
    • "This technique is based on the simulation of the human cognition process (McCulloch & Pitts, 1943 ). ANNs have been widely used for different problems, such as data analysis, non-invasive diagnosis in the medical field (Ding, Lu, Gao, & Peng, 2014; Resino et al., 2011 ), linking chemical knowledge (Gomez-Carracedo, Gestal, Dorado, & Andrade, 2007; Gómez-Carracedo, Gestal, Dorado, & Andrade, 2007) to protein function prediction ( FernandezLozano, Fernandez-Blanco, et al., 2014) or, as mentioned before, for AD classification with morphometric measures (Aguilar et al., 2013; Escudero et al., 2011 ). ANNs consist of many simple, interconnected computational units. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Several magnetic resonance techniques have been proposed as non-invasive imaging biomarkers for the evaluation of disease progression and early diagnosis of Alzheimer’s Disease (AD). This work is the first application of the Proton Magnetic Resonance Spectroscopy 1H-MRS data and machine-learning techniques to the classification of AD. A gender-matched cohort of 260 subjects aged between 57 and 99 years from the Alzheimer’s Disease Research Unit, of the Fundación CIEN-Fundación Reina Sofía has been used. A single-layer perceptron was found for AD prediction with only two spectroscopic voxel volumes (Tvol and CSFvol) in the left hippocampus, with an AUROC value of 0.866 (with TPR 0.812 and FPR 0.204) in a filter feature selection approach. These results suggest that knowing the composition of white and grey matter and cerebrospinal fluid of the spectroscopic voxel is essential in a 1H-MRS study to improve the accuracy of the quantifications and classifications, particularly in those studies involving elder patients and neurodegenerative diseases.
    No preview · Article · Mar 2015 · Expert Systems with Applications
  • Source
    • "(linear/non-linear) Effective in high-dimensional data Good generalization performance Deliver a unique solution Robust to noise/outliers data Computationally expensive Discriminant analysis 9 Well-known classical linear statistical methods [17, 29, 36, 60–66] (LDA/QDA) Available in a wide range of variations and extensions Simple to implement Suitable for dimensionality reduction in data with high dimensional features Enhance interpretation of between-group differences Optimal for data with Gaussian distribution Orthogonal Projection to Latent Structures 7 Beneficial for data with large number of dependent/correlated variables [11, 18, 48–52] Enhance model transparency and improve interpretation Robust to noise/missing data Provide a single predictor component for class separation Detect systematic variation in the data Decision Trees 3 Simple and fast method [18] [67] [68] Easy to understand and interpret High model transparency Capable of handling noise/outliers/missing data Capable of handling different type of attributes Requires little data preparation Artificial Neuronal Networks 2 Powerful nonlinear algorithm [18] [19] No requirement/assumptions on distribution/relationship of input data Capable of accurately handling non-linear and complex patterns Deals well with missing/incomplete data Black box nature (difficult to interpret) Excessive learning time for large neural networks Ensemble methods 4 Provide overall higher accuracy than individual classifier [30] [37] [69] [70] Flexible to use different learning algorithms Easy to implement Beneficial for data with high dimensionality and small sample size Robust to noise in data Black box nature (difficult to interpret) "
    [Show abstract] [Hide abstract]
    ABSTRACT: Machine learning algorithms and multivariate data analysis methods have been widely utilized in the field of Alzheimer's disease (AD) research in recent years. Advances in medical imaging and medical image analysis have provided a means to generate and extract valuable neuroimaging information. Automatic classification techniques provide tools to analyze this information and observe inherent disease-related patterns in the data. In particular, these classifiers have been used to discriminate AD patients from healthy control subjects and to predict conversion from mild cognitive impairment to AD. In this paper, recent studies are reviewed that have used machine learning and multivariate analysis in the field of AD research. The main focus is on studies that used structural magnetic resonance imaging (MRI), but studies that included positron emission tomography and cerebrospinal fluid biomarkers in addition to MRI are also considered. A wide variety of materials and methods has been employed in different studies, resulting in a range of different outcomes. Influential factors such as classifiers, feature extraction algorithms, feature selection methods, validation approaches, and cohort properties are reviewed, as well as key MRI-based and multi-modal based studies. Current and future trends are discussed.
    Full-text · Article · Apr 2014 · Journal of Alzheimer's disease: JAD
  • Source

    Full-text · Article · Dec 2011 · Trials
Show more