Tohka J, Zijdenbos A, Evans A. Fast and robust parameter estimation for statistical partial volume models in brain MRI. Neuroimage 23: 84-97

Digital Media Institute/Signal Processing, Tampere University of Technology, FIN-33101, Finland.
NeuroImage (Impact Factor: 6.36). 10/2004; 23(1):84-97. DOI: 10.1016/j.neuroimage.2004.05.007
Source: PubMed

ABSTRACT Due to the finite spatial resolution of imaging devices, a single voxel in a medical image may be composed of mixture of tissue types, an effect known as partial volume effect (PVE). Partial volume estimation, that is, the estimation of the amount of each tissue type within each voxel, has received considerable interest in recent years. Much of this work has been focused on the mixel model, a statistical model of PVE. We propose a novel trimmed minimum covariance determinant (TMCD) method for the estimation of the parameters of the mixel PVE model. In this method, each voxel is first labeled according to the most dominant tissue type. Voxels that are prone to PVE are removed from this labeled set, following which robust location estimators with high breakdown points are used to estimate the mean and the covariance of each tissue class. Comparisons between different methods for parameter estimation based on classified images as well as expectation--maximization-like (EM-like) procedure for simultaneous parameter and partial volume estimation are reported. The robust estimators based on a pruned classification as presented here are shown to perform well even if the initial classification is of poor quality. The results obtained are comparable to those obtained using the EM-like procedure, but require considerably less computation time. Segmentation results of real data based on partial volume estimation are also reported. In addition to considering the parameter estimation problem, we discuss differences between different approximations to the complete mixel model. In summary, the proposed TMCD method allows for the accurate, robust, and efficient estimation of partial volume model parameters, which is crucial to a variety of brain MRI data analysis procedures such as the accurate estimation of tissue volumes and the accurate delineation of the cortical surface.

Download full-text


Available from: Jussi Tohka, Aug 19, 2015
    • "All T1-weighted images were corrected for bias-field inhomogeneities , then spatially normalised and segmented into grey (GM), white matter (WM), and cerebrospinal fluid (CSF) within the same generative model (Ashburner and Friston, 2005). As described previously (Gaser, 2009), the segmentation procedure was further extended by accounting for partial volume effects (Tohka et al., 2004), applying adaptive maximum a posteriori estimations (Rajapakse et al., 1997), and using a hidden Markov Random Field model (Cuadra et al., 2005). For exclusion of artefacts on the grey–white-matter border (i.e. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Schizotypal traits are phenotypic risk factors for schizophrenia, associated with biological changes across a putative schizophrenia spectrum. In this study, we tested the hypothesis that brain structural changes in key brain areas relevant to this spectrum (esp. medial and lateral prefrontal cortex) would vary across different degrees of schizotypal trait expression and/or phenotypic markers of psychosis proneness in healthy non-clinical volunteers. We analysed high-resolution 3Tesla magnetic resonance images (MRI) of 59 healthy volunteers using voxel-based morphometry (VBM), correlating grey matter values to the positive and negative symptom factors of the schizotypal personality questionnaire (SPQ, German version) and a measure of psychosis proneness (community assessment of psychic experiences, CAPE). We found positive correlations between positive SPQ dimension and bilateral inferior and right superior frontal cortices, and positive CAPE dimension and left inferior frontal cortex, as well as CAPE negative dimension and right supplementary motor area (SMA) and left inferior parietal cortex. However, only the positive correlation of the right precuneus with negative schizotypy scores was significant after FWE correction for multiple comparisons. Our findings confirm an effect of schizotypal traits and psychosis proneness on brain structure in healthy subjects, providing further support to a biological continuum model. Copyright © 2015 Elsevier B.V. All rights reserved.
    Schizophrenia Research 07/2015; DOI:10.1016/j.schres.2015.06.017 · 4.43 Impact Factor
  • Source
    • ") using posteriori rating ( Rajapakse et al . , 1997 ) to get partial volume effects ( Tohka et al . , 2004 ) . Modulation by nonlinear effects ensured the correction for differences in brain size . Smoothing was conducted using a 8 mm full width at half maximum ( FWHM ) Gaussian kernel ."
    [Show abstract] [Hide abstract]
    ABSTRACT: Time-stable personality traits, such as impulsivity and its relationship with functional and structural brain alterations, have gained much attention in the recent literature. Evidence from functional neuroimaging data implies an association between impulsivity and cortical as well as subcortical areas of the reward system. Discounting future rewards during impulsive decisions can be related to activation in the orbitofrontal cortex and striatum. Cortical structural changes in prefrontal regions have been found for introspective impulsivity measures. The present study focuses on brain regions associated with delay discounting to investigate structural manifestations of trait impulsivity. To test this, seventy subjects underwent structural magnetic resonance imaging (MRI) followed by a behavioral delay discounting task outside of the scanner to measure impulsivity with questions like: "Would you like to have 3€ immediately or 10€ in 5 days?". The amount of smaller-but-sooner decisions was calculated and used as a measure of behavioral impulsivity. Furthermore, we estimated subject's individual delay discounting parameter K reflecting the tendency to discount future rewards. Behaviorally, we found strong evidence in favor of a discounting utility model compared to a standard hyperbolic model of choice valuation. Neuronally, we focused on cortical and subcortical brain structure and investigated the association of behavioral impulsivity with delay discounting tendencies and gray matter volume. Voxel-based morphometric analyses showed positive correlations between delay discounting and gray matter volume in the striatum. Additional analyses using Freesurfer provided evidence for a positive correlation between delay discounting and gray matter volume of the caudate. Taken together, our study provides strong evidence for a structural manifestation of time-stable trait impulsivity in the human brain.
    Frontiers in Human Neuroscience 07/2015; 9:384. DOI:10.3389/fnhum.2015.00384 · 2.90 Impact Factor
  • Source
    • " stereotaxic space using affine transformation ( Collins et al . , 1994 ) . The corrected and normalized volumes were classified as WM , GM , CSF , or background using an advanced neural network classifier ( Zijdenbos et al . , 2002 ) . The partial volume effect classification was performed using the trimmed minimum covariance determinant method ( Tohka et al . , 2004 ) . The partial volume fractions of each tissue class fell between 0 and 1 per voxel ."
    [Show abstract] [Hide abstract]
    ABSTRACT: The mean diffusivity (MD) value has been used to describe microstructural properties in Diffusion Tensor Imaging (DTI) in cortical gray matter (GM). Recently, researchers have applied a cortical surface generated from the T1-weighted volume. When the DTI data are analyzed using the cortical surface, it is important to assign an accurate MD value from the volume space to the vertex of the cortical surface, considering the anatomical correspondence between the DTI and the T1-weighted image. Previous studies usually sampled the MD value using the nearest-neighbor (NN) method or Linear method, even though there are geometric distortions in diffusion-weighted volumes. Here we introduce a Surface Guided Diffusion Mapping (SGDM) method to compensate for such geometric distortions. We compared our SGDM method with results using NN and Linear methods by investigating differences in the sampled MD value. We also projected the tissue classification results of non-diffusion-weighted volumes to the cortical midsurface. The CSF probability values provided by the SGDM method were lower than those produced by the NN and Linear methods. The MD values provided by the NN and Linear methods were significantly greater than those of the SGDM method in regions suffering from geometric distortion. These results indicate that the NN and Linear methods assigned the MD value in the CSF region to the cortical midsurface (GM region). Our results suggest that the SGDM method is an effective way to correct such mapping errors.
    Frontiers in Neuroscience 07/2015; 9:236. DOI:10.3389/fnins.2015.00236 · 3.70 Impact Factor
Show more