Spatial Normalization Algorithm Research Articles

DR-TAMAS: Diffeomorphic Registration for Tensor Accurate Alignment of Anatomical Structures

In this work, we propose DR-TAMAS (Diffeomorphic Registration for Tensor Accurate alignMent of Anatomical Structures), a novel framework for intersubject registration of Diffusion Tensor Imaging (DTI) data sets. This framework is optimized for brain data and its main goal is to achieve an accurate alignment of all brain structures, including white matter (WM), gray matter (GM), and spaces containing cerebrospinal fluid (CSF). Currently most DTI-based spatial normalization algorithms emphasize alignment of anisotropic structures. While some diffusion-derived metrics, such as diffusion anisotropy and tensor eigenvector orientation, are highly informative for proper alignment of WM, other tensor metrics such as the trace or mean diffusivity (MD) are fundamental for a proper alignment of GM and CSF boundaries. Moreover, it is desirable to include information from structural MRI data, e.g., T1-weighted or T2-weighted images, which are usually available together with the diffusion data. The fundamental property of DR-TAMAS is to achieve global anatomical accuracy by incorporating in its cost function the most informative metrics locally. Another important feature of DR-TAMAS is a symmetric time-varying velocity-based transformation model, which enables it to account for potentially large anatomical variability in healthy subjects and patients. The performance of DR-TAMAS is evaluated with several data sets and compared with other widely-used diffeomorphic image registration techniques employing both full tensor information and/or DTI-derived scalar maps. Our results show that the proposed method has excellent overall performance in the entire brain, while being equivalent to the best existing methods in WM.

Characterization of disease‐related covariance topographies with SSMPCA toolbox: Effects of spatial normalization and PET scanners

To generate imaging biomarkers from disease-specific brain networks, we have implemented a general toolbox to rapidly perform scaled subprofile modeling (SSM) based on principal component analysis (PCA) on brain images of patients and normals. This SSMPCA toolbox can define spatial covariance patterns whose expression in individual subjects can discriminate patients from controls or predict behavioral measures. The technique may depend on differences in spatial normalization algorithms and brain imaging systems. We have evaluated the reproducibility of characteristic metabolic patterns generated by SSMPCA in patients with Parkinson's disease (PD). We used [(18) F]fluorodeoxyglucose PET scans from patients with PD and normal controls. Motor-related (PDRP) and cognition-related (PDCP) metabolic patterns were derived from images spatially normalized using four versions of SPM software (spm99, spm2, spm5, and spm8). Differences between these patterns and subject scores were compared across multiple independent groups of patients and control subjects. These patterns and subject scores were highly reproducible with different normalization programs in terms of disease discrimination and cognitive correlation. Subject scores were also comparable in patients with PD imaged across multiple PET scanners. Our findings confirm a very high degree of consistency among brain networks and their clinical correlates in PD using images normalized in four different SPM platforms. SSMPCA toolbox can be used reliably for generating disease-specific imaging biomarkers despite the continued evolution of image preprocessing software in the neuroimaging community. Network expressions can be quantified in individual patients independent of different physical characteristics of PET cameras.

The neural correlates of impaired collision avoidance in hemianopic patients

The aim of this study was to assess the brain regions associated with impaired performance in a virtual, dynamic collision avoidance task, in a group of patients with homonymous visual field defects (HVFDs) because of unilateral vascular brain lesions. Overall task performance was quantitatively assessed as the number of collisions while crossing an intersection at two levels of traffic density. Twenty-six patients were divided into two subgroups using the median split method: patients with 'performance above average' (HVFD(A), i.e. lower number of collisions) and patients with 'performance below average' (HVFD(B), i.e. higher number of collisions). In order to identify the anatomical structures that might be specifically affected in HVFD(B) patients but spared in HVFD(A) patients, overlap, subtraction and voxel-based lesion-symptom mapping analyses were performed using the MRIcron software. No significant difference in collision avoidance between patients with left- and right-hemispheric lesions was revealed. Separate lesion analysis in 12 patients with right- and 14 patients with left-hemispheric lesions showed that the cortical structures associated with impaired collision avoidance were the parieto-occipital region and posterior cingulate gyrus in the right hemisphere and the inferior occipital cortex and parts of the fusiform (occipito-temporal) gyrus in the left hemisphere. In the present collision avoidance paradigm, impaired performance of patients with right-hemispheric lesions is associated with damage in the dorsal processing stream and potential impact on the visual spatial working memory (WM), while impaired performance of patients with left-hemispheric lesions is associated with damage in the ventral stream and potential impact on the visual object WM.

Quantification of receptor–ligand binding potential in sub-striatal domains using probabilistic and template regions of interest

Sub-striatal regions of interest (ROIs) are widely used in PET studies to investigate the role of dopamine in the modulation of neural networks implicated in emotion, cognition and motor function. One common approach is that of Mawlawi et al. (2001) and Martinez et al. (2003), where each striatum is divided into five sub-regions. This study focuses on the use of two spatial normalization-based alternatives to manual sub-striatal ROI delineation per subject: manual ROI delineation on a template brain and the production of probabilistic ROIs from a set of subject-specific manually delineated ROIs. Two spatial normalization algorithms were compared: SPM5 unified segmentation and ART. The ability of these methods to quantify sub-striatal regional non-displaceable binding potential (BPND) and BPND % change (following methylphenidate) was tested on 32 subjects (16 controls and 16 ADHD patients) scanned with the dopamine D2/D3 ligand [18F]fallypride. Probabilistic ROIs produced by ART provided the best results, with similarity index values against subject-specific manual ROIs of 0.75–0.89 (mean 0.84) compared to 0.70–0.85 (mean 0.79) for template ROIs. Correlations (r) for BPND and BPND % change between subject-specific manual ROIs and these probabilistic ROIs of 0.90–0.98 (mean 0.95) and 0.98–1.00 (mean 0.99) respectively were superior overall to those obtained with template ROIs, although only marginally so for BPND % change. The significance of relationships between BPND measures and both behavioural tasks and methylphenidate plasma levels was preserved with ART combined with both probabilistic and template ROIs. SPM5 virtually matched the performance of ART for BPND % change estimation but was inferior for BPND estimation in caudate sub-regions. ART spatial normalization combined with probabilistic ROIs and to a lesser extent template ROIs provides an efficient and accurate alternative to time-consuming manual sub-striatal ROI delineation per subject, especially when the parameter of interest is BPND % change.

3D Gabor wavelets for evaluating SPM normalization algorithm

A Gabor wavelets based method is proposed in this paper for evaluating and tuning the parameters of image registration algorithms. We propose a 3D local anatomical structure descriptor, namely the Maximum Responded Gabor Wavelet (MRGW), for measuring registration quality based on anatomical variability of registered images. The effectiveness of the descriptor is demonstrated through a practical application, using the variance of MRGW response to tune parameters of a nonlinear spatial normalization algorithm, which is part of the popular software package for medical image processing - the Statistical Parametric Mapping (SPM).

Neural bases of personal and extrapersonal neglect in humans

Human awareness of left space may be disrupted by cerebral lesions to the right hemisphere (hemispatial neglect). Current knowledge on the anatomical bases of this complex syndrome is based on the results of group studies that investigated primarily the best known aspect of the syndrome, which is visual neglect for near extrapersonal (or peripersonal) space. However, another component-neglect for personal space-is more often associated with, than double-dissociated from, extrapersonal neglect, especially, in chronic patients. The present investigation aimed at exploring the anatomical substrate of both extrapersonal and personal neglect by using different advanced methodological approaches to lesion-function correlation. Fifty-two right ischaemic patients were submitted to neuropsychological assessment and in-depth MRI evaluation. The borders of each patient's lesion were delimited onto its own high-resolution anatomical image and then submitted to an automated spatial normalization algorithm. Besides conventional lesion density plots and subtraction analysis, region-based statistical analyses were performed on percentage values of the lesioned tissue also using a new parcellation of the white matter (WM). Data were finally submitted to voxelwise statistical analysis using a recently proposed method (voxel-based lesion-symptom mapping). Results converged in showing that awareness of extrapersonal space is based on the integrity of a circuit of right frontal (ventral premotor cortex and middle frontal gyrus) and superior temporal regions, whereas awareness of personal space is rooted in right inferior parietal regions (supramarginal gyrus, post-central gyrus and especially the WM medial to them). Common but less crucial regions for both neglect sub-types were located in the temporo-peri-Sylvian cortex. We suggest that extrapersonal space awareness critically involves a ventral circuit recently described for the exogenous allocation and reorienting of attention in space. Disruption of personal space awareness, instead, seems to be due to a functional disconnection between regions important for coding proprioceptive and somatosensory inputs, and regions coding more abstract egocentric representations of the body in space. In conclusion, present data strongly support a segregation of personal and extrapersonal spatial awareness in humans, both from a functional and an anatomical point of view.

Tuning and comparing spatial normalization methods

Spatial normalization is a key process in cross-sectional studies of brain structure and function using MRI, fMRI, PET and other imaging techniques. A wide range of 2D surface and 3D image deformation algorithms have been developed, all of which involve design choices that are subject to debate. Moreover, most have numerical parameters whose value must be specified by the user. This paper proposes a principled method for evaluating design choices and choosing parameter values. This method can also be used to compare competing spatial normalization algorithms. We demonstrate the method through a performance analysis of a nonaffine registration algorithm for 3D images and a registration algorithm for 2D cortical surfaces.

“Voxel-Based Morphometry” Should Not Be Used with Imperfectly Registered Images

John Ashburner and Karl Friston (2000) introduced a standardized method of “voxel-based morphometry” (VBM) for comparisons of local concentrations of gray matter between two groups of subjects. Segmented images of gray matter from grossly normalized high-resolution images are smoothed and their group differences analyzed by the now-conventional voxelwise Worsley approach to Gaussian random fields of differences. This comment concerns an unfortunate interaction between the algorithm's spatial normalization and voxelwise comparison steps, whereby several obvious quantitative confounds are injected at the core of the inference engine the authors put forward. Specifically, the statistics of the resulting voxelwise comparisons are uninformative about group differences wherever the spatial normalization algorithm has failed to register on any robustly appearing image gradient. The method of Ashburner and Friston is defensible only far from all image gradients.

Evaluation of octree regional spatial normalization method for regional anatomical matching

The goal of regional spatial normalization is to remove anatomical differences between individual three-dimensional (3D) brain images by warping them to match features of a standard brain atlas. Processing to fit features at the limiting resolution of a 3D MR image volume is computationally intensive, limiting the broad use of full-resolution regional spatial normalization. In Kochunov et al. (1999: Neuro-Image 10:724-737), we proposed a regional spatial normalization algorithm called octree spatial normalization (OSN) that reduces processing time to minutes while targeting the accuracy of previous methods. In the current study, modifications of the OSN algorithm for use in human brain images are described and tested. An automated brain tissue segmentation procedure was adopted to create anatomical templates to drive feature matching in white matter, gray matter, and cerebral-spinal fluid. Three similarity measurement functions (fast-cross correlation (CC), sum-square error, and centroid) were evaluated in a group of six subjects. A combination of fast-CC and centroid was found to provide the best feature matching and speed. Multiple iterations and multiple applications of the OSN algorithm were evaluated to improve fit quality. Two applications of the OSN algorithm with two iterations per application were found to significantly reduce volumetric mismatch (up to six times for lateral ventricle) while keeping processing time under 30 min. The refined version of OSN was tested with anatomical landmarks from several major sulci in a group of nine subjects. Anatomical variability was appreciably reduced for every sulcus investigated, and mean sulcal tracings accurately followed sulcal tracings in the target brain.