Maite García-Sebastián

Lattice independent component analysis for functional magnetic resonance imaging

We introduce a lattice independent component analysis (LICA) unsupervised scheme to functional magnetic resonance imaging (fMRI) data analysis. LICA is a non-linear alternative to independent component analysis (ICA), such that ICAs statistical independent sources correspond to LICAs lattice independent sources. In this paper, LICA uses an incremental lattice source induction algorithm (ILSIA) to induce the lattice independent sources from the input dataset. The ILSIA computes a set of Strongly Lattice Independent vectors using properties of lattice associative memories regarding Lattice Independence and Chebyshev best approximation. The lattice independent sources constitute a set of Affine Independent vectors that define a simplex covering the input data. LICA carries out data linear unmixing based on the lattice independent sources basis. Therefore, LICA is a hybrid combination of a non-linear lattice based component and a linear unmixing component. The principal advantage over ICA is that LICA does not impose any probabilistic model assumptions on the data sources. We compare LICA with ICA in two case studies. Firstly, including simulated fMRI data, LICA discovers the spatial location of meaningful sources with less ambiguity than ICA. Secondly, including real data from an auditory stimulation experiment, LICA improves over some state of the art ICA variants discovering the activation patterns detected by Statistical Parametric Mapping (SPM) on the same data.

Neurocognitive disorder detection based on feature vectors extracted from VBM analysis of structural MRI

Dementia is a growing concern due to the aging process of the western societies. Non-invasive detection is therefore a high priority research endeavor. In this paper we report results of classification systems applied to the feature vectors obtained by a feature extraction method computed on structural magnetic resonance imaging (sMRI) volumes for the detection of two neurological disorders with cognitive impairment: myotonic dystrophy of type 1 (MD1) and Alzheimer disease (AD). The feature extraction process is based on the voxel clusters detected by voxel-based morphometry (VBM) analysis of sMRI upon a set of patient and control subjects. This feature extraction process is specific for each kind of disease and is grounded on the findings obtained by medical experts. The 10-fold cross-validation results of several statistical and neural network based classification algorithms trained and tested on these features show high specificity and moderate sensitivity of the classifiers, suggesting that the approach is better suited for rejecting than for detecting early stages of the diseases.

Classification results of artificial neural networks for Alzheimer's disease detection

Detection of Alzheimers disease on brain Magnetic Resonance Imaging (MRI) is a highly sought goal in the Neurosciences. We used four different models of Artificial Neural Networks (ANN): Back-propagation (BP), Radial Basis Networks (RBF), Learning Vector Quantization Networks (LVQ) and Probabilistic Neural Networks (PNN) to perform classification of patients of mild Alzheimers disease vs. control subjects. Features are extracted from the brain volume data using Voxel-based Morphometry (VBM) detection clusters. The voxel location detection clusters given by the VBM were applied to select the voxel values upon which the classification features were computed. We have evaluated feature vectors computed from the GM segmentation volumes using the VBM clusters as voxel selection masks. The study has been performed on MRI volumes of 98 females, after careful demographic selection from the Open Access Series of Imaging Studies (OASIS) database, which is a large number of subjects compared to current reported studies.

On the Use of Morphometry Based Features for Alzheimer's Disease Detection on MRI

We have studied feature extraction processes for the detection of Alzheimers disease on brain Magnetic Resonance Imaging (MRI) based on Voxel-based morphometry (VBM). The clusters of voxel locations detected by the VBM were applied to select the voxel intensity values upon which the classification features were computed. We have explored the use of the data from the original MRI volumes and the GM segmentation volumes. In this paper, we apply the Support Vector Machine (SVM) algorithm to perform classification of patients with mild Alzheimers disease vs. control subjects. The study has been performed on MRI volumes of 98 females, after careful demographic selection from the Open Access Series of Imaging Studies (OASIS) database, which is a large number of subjects compared to current reported studies.

A lattice computing approach for on-line fMRI analysis

We introduce an approach to fMRI analysis based on the Endmember Induction Heuristic Algorithm (EIHA). This algorithm uses the Lattice Associative Memory (LAM) to detect Lattice Independent vectors, which can be assumed to be Affine Independent, and therefore candidates to be the endmembers of the data. Induced endmembers are used to compute the activation levels of voxels as result of an unmixing process. The endmembers correspond to diverse activation patterns, one of these activation patterns corresponds to the resting state of the neuronal tissue. The on-line working of the algorithm does not need neither a previous training process nor a priori models of the data. Results on a case study compare with the results given by the state of art SPM software.

Results of an Adaboost Approach on Alzheimer's Disease Detection on MRI

In this paper we explore the use of the Voxel-based Morphometry (VBM) detection clusters to guide the feature extraction processes for the detection of Alzheimers disease on brain Magnetic Resonance Imaging (MRI). The voxel location detection clusters given by the VBM were applied to select the voxel values upon which the classification features were computed. We have evaluated feature vectors computed over the data from the original MRI volumes and from the GM segmentation volumes, using the VBM clusters as voxel selection masks. We use the Support Vector Machine (SVM) algorithm to perform classification of patients with mild Alzheimers disease vs. control subjects. We have also considered combinations of isolated cluster based classifiers and an Adaboost strategy applied to the SVM built on the feature vectors. The study has been performed on MRI volumes of 98 females, after careful demographic selection from the Open Access Series of Imaging Studies (OASIS) database, which is a large number of subjects compared to current reported studies. Results are moderately encouraging, as we can obtain up to 85% accuracy with the Adaboost strategy in a 10-fold cross-validation.

A parametric gradient descent MRI intensity inhomogeneity correction algorithm

Given an appropriate imaging resolution, a common Magnetic Resonance Imaging (MRI) model assumes that the object under study is composed of homogeneous tissues whose imaging intensity is constant, so that MRI produces piecewise constant images. The intensity inhomogeneity (IIH) is modeled by a multiplicative inhomogeneity field. It is due to the spatial inhomogeneity in the excitatory Radio Frequency (RF) signal and other effects. It has been acknowledged as a greater source of error for automatic segmentation algorithms than additive noise. We propose a parametric IIH correction algorithm for MRI that consists of the gradient descent of an error function related to the classification error of the IIH corrected image. The inhomogeneity field is modeled as a linear combination of 3D products of Legendre polynomials. In this letter we test both the image restoration capabilities and the classification accuracy of the algorithm. In restoration processes the adaptive algorithm is used only to estimate the inhomogeneity field. Test images to be restored are IIH corrupted versions of the BrainWeb site simulations. The algorithm image restoration is evaluated by the correlation of the restored image with the known clean image. In classification processes the algorithm is used to estimate both the inhomogeneity field and the intensity class means. The algorithm classification accuracy is tested over the images from the IBSR site. The proposed algorithm is compared with Maximum A Posteriori (MAP) and Fuzzy Clustering algorithms.

An adaptive field rule for non-parametric MRI intensity inhomogeneity estimation algorithm

A widely accepted magnetic resonance imaging (MRI) model states that the observed voxel intensity is a piecewise constant signal intensity function corresponding to the tissue spatial distribution, corrupted with multiplicative and additive noise. The multiplicative noise is assumed to be a smooth bias field, it is called intensity inhomogeneity (IIH) field. Our approach to IIH correction is based on the definition of an energy function that incorporates some smoothness constraints into the conventional classification error function of the IIH corrected image. The IIH field estimation algorithm is a gradient descent of this energy function relative to the IIH field. We call it adaptive field rule (AFR). We comment on the likeness of our approach to the self-organizing map (SOM) learning rule, on the basis of the neighboring function that controls the influence of the neighborhood on each voxels IIH estimation. We discuss the convergence properties of the algorithm. Experimental results show that AFR compares well with state of the art algorithms. Moreover, the mean signal intensity corresponding to each class of tissue can be estimated from the image data applying the gradient descent of the proposed energy function relative to the intensity class means. We test several variations of this gradient descent approach, which embody diverse assumptions about available a priori information.

Cortical microstructural changes along the Alzheimer's disease continuum

Cortical mean diffusivity (MD) and free water fraction (FW) changes are proposed biomarkers for Alzheimers disease (AD).

Robustness of an adaptive MRI segmentation algorithm parametric intensity inhomogeneity modeling

We propose an unsupervised segmentation algorithm for magnetic resonance images (MRI) endowed with a parametric intensity inhomogeneity (IIH) correction schema and the on-line estimation of the image model intensity class means. The paper includes an extensive experimentation that shows that the algorithm is robust in the sense that it converges to good image segmentations despite the initial estimation of the image model intensity class means. The algorithm is, therefore, highly automatic requiring no interactive tuning to obtain good image segmentations, an appealing property in clinical environments. The IIH field and intensity class means estimation consists of the gradient descent of the restoration error of the intensity corrected image. Our algorithm does not work on the logarithmic transformation of the image, thus allowing for the explicit distinction between the smooth multiplicative field and the independent and identically distributed additive noise at each image voxel.