Detection and mapping of abnormal brain structure with a probabilistic atlas of cortical surfaces

High-resolution cortical parcellation based on conserved brain landmarks for localization of multimodal data to the nearest centimeter

Precise cortical brain localization presents an important challenge in the literature. Brain atlases provide data-guided parcellation based on functional and structural brain metrics, and each atlas has its own unique benefits for localization. We offer a parcellation guided by intracranial electroencephalography, a technique which has historically provided pioneering advances in our understanding of brain structure-function relationships. We used a consensus boundary mapping approach combining anatomical designations in Duvernoy's Atlas of the Human Brain, a widely recognized textbook of human brain anatomy, with the anatomy of the MNI152 template and the magnetic resonance imaging scans of an epilepsy surgery cohort. The Yale Brain Atlas consists of 690 one-square centimeter parcels based around conserved anatomical features and each with a unique identifier to communicate anatomically unambiguous localization. We report on the methodology we used to create the Atlas along with the findings of a neuroimaging study assessing the accuracy and clinical usefulness of cortical localization using the Atlas. We also share our vision for the Atlas as a tool in the clinical and research neurosciences, where it may facilitate precise localization of data on the cortex, accurate description of anatomical locations, and modern data science approaches using standardized brain regions.

Brain tissues have single-voxel signatures in multi-spectral MRI

Since the seminal works by Brodmann and contemporaries, it is well-known that different brain regions exhibit unique cytoarchitectonic and myeloarchitectonic features. Transferring the approach of classifying brain tissues - and other tissues - based on their intrinsic features to the realm of magnetic resonance (MR) is a longstanding endeavor. In the 1990s, atlas-based segmentation replaced earlier multi-spectral classification approaches because of the large overlap between the class distributions. Here, we explored the feasibility of performing global brain classification based on intrinsic MR features, and used several technological advances: ultra-high field MRI, q-space trajectory diffusion imaging revealing voxel-intrinsic diffusion properties, chemical exchange saturation transfer and semi-solid magnetization transfer imaging as a marker of myelination and neurochemistry, and current neural network architectures to analyze the data. In particular, we used the raw image data as well to increase the number of input features. We found that a global brain classification of roughly 97 brain regions was feasible with gross classification accuracy of 60%; and that mapping from voxel-intrinsic MR data to the brain region to which the data belongs is possible. This indicates the presence of unique MR signals of different brain regions, similar to their cytoarchitectonic and myeloarchitectonic fingerprints.

Combined Use of MRI, fMRIand Cognitive Data for Alzheimer’s Disease: Preliminary Results

MRI can favor clinical diagnosis providing morphological and functional information of several neurological disorders. This paper deals with the problem of exploiting both data, in a combined way, to develop a tool able to support clinicians in the study and diagnosis of Alzheimer’s Disease (AD). In this work, 69 subjects from the ADNI open database, 33 AD patients and 36 healthy controls, were analyzed. The possible existence of a relationship between brain structure modifications and altered functions between patients and healthy controls was investigated performing a correlation analysis on brain volume, calculated from the MRI image, the clustering coefficient, derived from fRMI acquisitions, and the Mini Mental Score Examination (MMSE). A statistically-significant correlation was found only in four ROIs after Bonferroni’s correction. The correlation analysis alone was still not sufficient to provide a reliable and powerful clinical tool in AD diagnosis however. Therefore, a machine learning strategy was studied by training a set of support vector machine classifiers comparing different features. The use of a unimodal approach led to unsatisfactory results, whereas the multimodal approach, i.e., the synergistic combination of MRI, fMRI, and MMSE features, resulted in an accuracy of 95.65%, a specificity of 97.22%, and a sensibility of 93.93%.

Long-Term Grey Matter Changes in First Episode Psychosis: A Systematic Review

OBJECTIVE

To determine possible progressive changes of the grey matter at the first stages of the schizophrenia spectrum disorders, and to determine what regions are involved in these changes.

METHODS

We searched the literature concerning studies on longitudinal changes in grey matter in first-episode psychosis using magnetic resonance imaging, especially studies with an interval between scans of more than a year. Only articles published before 2018 were searched. We selected 19 magnetic resonance imaging longitudinal studies that used different neuroimaging analysis techniques to study changes in cerebral grey matter in a group of patients with a first episode of psychosis.

RESULTS

Patients with first episode of psychosis showed a decrease over time in cortical grey matter compared with a group of control subjects in frontal, temporal (specifically in superior regions), parietal, and subcortical regions. In addition to the above, studies indicate that patients showed a grey matter decrease in cerebellum and lateral ventricles volume.

CONCLUSION

The results suggest a decrease in grey matter in the years after the first episode of psychosis. Furthermore, the results of the studies showed consistency, regardless of the methods used in their analyses, as well as the time intervals between image collections.

Accurate nonlinear mapping between MNI volumetric and FreeSurfer surface coordinate systems

The results of most neuroimaging studies are reported in volumetric (e.g., MNI152) or surface (e.g., fsaverage) coordinate systems. Accurate mappings between volumetric and surface coordinate systems can facilitate many applications, such as projecting fMRI group analyses from MNI152/Colin27 to fsaverage for visualization or projecting resting‐state fMRI parcellations from fsaverage to MNI152/Colin27 for volumetric analysis of new data. However, there has been surprisingly little research on this topic. Here, we evaluated three approaches for mapping data between MNI152/Colin27 and fsaverage coordinate systems by simulating the above applications: projection of group‐average data from MNI152/Colin27 to fsaverage and projection of fsaverage parcellations to MNI152/Colin27. Two of the approaches are currently widely used. A third approach (registration fusion) was previously proposed, but not widely adopted. Two implementations of the registration fusion (RF) approach were considered, with one implementation utilizing the Advanced Normalization Tools (ANTs). We found that RF‐ANTs performed the best for mapping between fsaverage and MNI152/Colin27, even for new subjects registered to MNI152/Colin27 using a different software tool (FSL FNIRT). This suggests that RF‐ANTs would be useful even for researchers not using ANTs. Finally, it is worth emphasizing that the most optimal approach for mapping data to a coordinate system (e.g., fsaverage) is to register individual subjects directly to the coordinate system, rather than via another coordinate system. Only in scenarios where the optimal approach is not possible (e.g., mapping previously published results from MNI152 to fsaverage), should the approaches evaluated in this manuscript be considered. In these scenarios, we recommend RF‐ANTs (https://github.com/ThomasYeoLab/CBIG/tree/master/stable_projects/registration/Wu2017_RegistrationFusion).

Impairment of memory generalization in preclinical autosomal dominant Alzheimer's disease mutation carriers

Fast, inexpensive, and noninvasive identification of Alzheimer's disease (AD) before clinical symptoms emerge would augment our ability to intervene early in the disease. Individuals with fully penetrant genetic mutations causing autosomal dominant Alzheimer's disease (ADAD) are essentially certain to develop the disease, providing a unique opportunity to examine biomarkers during the preclinical stage. Using a generalization task that has previously shown to be sensitive to medial temporal lobe pathology, we compared preclinical individuals carrying ADAD mutations to noncarrying kin to determine whether generalization (the ability to transfer previous learning to novel but familiar recombinations) is vulnerable early, before overt cognitive decline. As predicted, results revealed that preclinical ADAD mutation carriers made significantly more errors during generalization than noncarrying kin, despite no differences between groups during learning or retention. This impairment correlated with the left hippocampal volume, particularly in mutation carriers. Such identification of generalization deficits in early ADAD may provide an easily implementable and potentially linguistically and culturally neutral way to identify and track cognition in ADAD.

CerebroMatic: A Versatile Toolbox for Spline-Based MRI Template Creation

Brain image spatial normalization and tissue segmentation rely on prior tissue probability maps. Appropriately selecting these tissue maps becomes particularly important when investigating "unusual" populations, such as young children or elderly subjects. When creating such priors, the disadvantage of applying more deformation must be weighed against the benefit of achieving a crisper image. We have previously suggested that statistically modeling demographic variables, instead of simply averaging images, is advantageous. Both aspects (more vs. less deformation and modeling vs. averaging) were explored here. We used imaging data from 1914 subjects, aged 13 months to 75 years, and employed multivariate adaptive regression splines to model the effects of age, field strength, gender, and data quality. Within the spm/cat12 framework, we compared an affine-only with a low- and a high-dimensional warping approach. As expected, more deformation on the individual level results in lower group dissimilarity. Consequently, effects of age in particular are less apparent in the resulting tissue maps when using a more extensive deformation scheme. Using statistically-described parameters, high-quality tissue probability maps could be generated for the whole age range; they are consistently closer to a gold standard than conventionally-generated priors based on 25, 50, or 100 subjects. Distinct effects of field strength, gender, and data quality were seen. We conclude that an extensive matching for generating tissue priors may model much of the variability inherent in the dataset which is then not contained in the resulting priors. Further, the statistical description of relevant parameters (using regression splines) allows for the generation of high-quality tissue probability maps while controlling for known confounds. The resulting CerebroMatic toolbox is available for download at http://irc.cchmc.org/software/cerebromatic.php.

Detecting Anatomical Landmarks for Fast Alzheimer's Disease Diagnosis

Structural magnetic resonance imaging (MRI) is a very popular and effective technique used to diagnose Alzheimer's disease (AD). The success of computer-aided diagnosis methods using structural MRI data is largely dependent on the two time-consuming steps: 1) nonlinear registration across subjects, and 2) brain tissue segmentation. To overcome this limitation, we propose a landmark-based feature extraction method that does not require nonlinear registration and tissue segmentation. In the training stage, in order to distinguish AD subjects from healthy controls (HCs), group comparisons, based on local morphological features, are first performed to identify brain regions that have significant group differences. In general, the centers of the identified regions become landmark locations (or AD landmarks for short) capable of differentiating AD subjects from HCs. In the testing stage, using the learned AD landmarks, the corresponding landmarks are detected in a testing image using an efficient technique based on a shape-constrained regression-forest algorithm. To improve detection accuracy, an additional set of salient and consistent landmarks are also identified to guide the AD landmark detection. Based on the identified AD landmarks, morphological features are extracted to train a support vector machine (SVM) classifier that is capable of predicting the AD condition. In the experiments, our method is evaluated on landmark detection and AD classification sequentially. Specifically, the landmark detection error (manually annotated versus automatically detected) of the proposed landmark detector is 2.41 mm , and our landmark-based AD classification accuracy is 83.7%. Lastly, the AD classification performance of our method is comparable to, or even better than, that achieved by existing region-based and voxel-based methods, while the proposed method is approximately 50 times faster.

3D Statistical Parametric Mapping of EEG Source Spectra by Means of Variable Resolution Electromagnetic Tomography (VARETA)

This article describes a new method for 3D QEEG tomography in the frequency domain. A variant of Statistical Parametric Mapping is presented for source log spectra. Sources are estimated by means of a Discrete Spline EEG inverse solution known as Variable Resolution Electromagnetic Tomography (VARETA). Anatomical constraints are incorporated by the use of the Montreal Neurological Institute (MNI) probabilistic brain atlas. Efficient methods are developed for frequency domain VARETA in order to estimate the source spectra for the set of 103–105 voxels that comprise an EEG/MEG inverse solution. High resolution source Z spectra are then defined with respect to the age dependent mean and standard deviations of each voxel, which are summarized as regression equations calculated from the Cuban EEG normative database. The statistical issues involved are addressed by the use of extreme value statistics. Examples are shown that illustrate the potential clinical utility of the methods herein developed.

Incorporating 3-dimensional models in online articles

INTRODUCTION

The aims of this article are to introduce the capability to view and interact with 3-dimensional (3D) surface models in online publications, and to describe how to prepare surface models for such online 3D visualizations.

METHODS

Three-dimensional image analysis methods include image acquisition, construction of surface models, registration in a common coordinate system, visualization of overlays, and quantification of changes. Cone-beam computed tomography scans were acquired as volumetric images that can be visualized as 3D projected images or used to construct polygonal meshes or surfaces of specific anatomic structures of interest. The anatomic structures of interest in the scans can be labeled with color (3D volumetric label maps), and then the scans are registered in a common coordinate system using a target region as the reference. The registered 3D volumetric label maps can be saved in .obj, .ply, .stl, or .vtk file formats and used for overlays, quantification of differences in each of the 3 planes of space, or color-coded graphic displays of 3D surface distances.

RESULTS

All registered 3D surface models in this study were saved in .vtk file format and loaded in the Elsevier 3D viewer. In this study, we describe possible ways to visualize the surface models constructed from cone-beam computed tomography images using 2D and 3D figures. The 3D surface models are available in the article's online version for viewing and downloading using the reader's software of choice. These 3D graphic displays are represented in the print version as 2D snapshots. Overlays and color-coded distance maps can be displayed using the reader's software of choice, allowing graphic assessment of the location and direction of changes or morphologic differences relative to the structure of reference. The interpretation of 3D overlays and quantitative color-coded maps requires basic knowledge of 3D image analysis.

CONCLUSIONS

When submitting manuscripts, authors can now upload 3D models that will allow readers to interact with or download them. Such interaction with 3D models in online articles now will give readers and authors better understanding and visualization of the results.

Medial temporal atrophy in early and late-onset Alzheimer's disease

Late-onset and early-onset Alzheimer's disease (LOAD, EOAD) affect different neural systems and may be separate nosographic entities. The most striking differences are in the medial temporal lobe, severely affected in LOAD and relatively spared in EOAD. We assessed amygdalar morphology and volume in 18 LOAD and 18 EOAD patients and 36 aged-matched controls and explored their relationship with the hippocampal volume. Three-dimensional amygdalar shape was reconstructed with the radial atrophy mapping technique, hippocampal volume was measured using a manual method. Atrophy was greater in LOAD than EOAD: 25% versus 17% in the amygdala and 20% versus 13% in the hippocampus. In the amygdala, LOAD showed significantly greater tissue loss than EOAD in the right dorsal central, lateral, and basolateral nuclei (20%-30% loss, p < 0.03), all known to be connected to limbic regions. In LOAD but not EOAD, greater hippocampal atrophy was associated with amygdalar atrophy in the left dorsal central and medial nuclei (r = 0.6, p < 0.05) also part of the limbic system. These findings support the notion that limbic involvement is a prominent feature of LOAD but not EOAD.

Multi-contrast multi-atlas parcellation of diffusion tensor imaging of the human brain

In this paper, we propose a novel method for parcellating the human brain into 193 anatomical structures based on diffusion tensor images (DTIs). This was accomplished in the setting of multi-contrast diffeomorphic likelihood fusion using multiple DTI atlases. DTI images are modeled as high dimensional fields, with each voxel exhibiting a vector valued feature comprising of mean diffusivity (MD), fractional anisotropy (FA), and fiber angle. For each structure, the probability distribution of each element in the feature vector is modeled as a mixture of Gaussians, the parameters of which are estimated from the labeled atlases. The structure-specific feature vector is then used to parcellate the test image. For each atlas, a likelihood is iteratively computed based on the structure-specific vector feature. The likelihoods from multiple atlases are then fused. The updating and fusing of the likelihoods is achieved based on the expectation-maximization (EM) algorithm for maximum a posteriori (MAP) estimation problems. We first demonstrate the performance of the algorithm by examining the parcellation accuracy of 18 structures from 25 subjects with a varying degree of structural abnormality. Dice values ranging 0.8-0.9 were obtained. In addition, strong correlation was found between the volume size of the automated and the manual parcellation. Then, we present scan-rescan reproducibility based on another dataset of 16 DTI images - an average of 3.73%, 1.91%, and 1.79% for volume, mean FA, and mean MD respectively. Finally, the range of anatomical variability in the normal population was quantified for each structure.

Atlas-based neuroinformatics via MRI: harnessing information from past clinical cases and quantitative image analysis for patient care

With the ever-increasing amount of anatomical information radiologists have to evaluate for routine diagnoses, computational support that facilitates more efficient education and clinical decision making is highly desired. Despite the rapid progress of image analysis technologies for magnetic resonance imaging of the human brain, these methods have not been widely adopted for clinical diagnoses. To bring computational support into the clinical arena, we need to understand the decision-making process employed by well-trained clinicians and develop tools to simulate that process. In this review, we discuss the potential of atlas-based clinical neuroinformatics, which consists of annotated databases of anatomical measurements grouped according to their morphometric phenotypes and coupled with the clinical informatics upon which their diagnostic groupings are based. As these are indexed via parametric representations, we can use image retrieval tools to search for phenotypes along with their clinical metadata. The review covers the current technology, preliminary data, and future directions of this field.

Dissecting the gene dose-effects of the APOE ε4 and ε2 alleles on hippocampal volumes in aging and Alzheimer's disease

OBJECTIVE

To investigate whether there is a specific dose-dependent effect of the Apolipoprotein E (APOE) ε4 and ε2 alleles on hippocampal volume, across the cognitive spectrum, from normal aging to Alzheimer's Disease (AD).

MATERIALS AND METHODS

We analyzed MR and genetic data on 662 patients from the Alzheimer's Disease Neuroimaging Initiative (ADNI) database-198 cognitively normal controls (CN), 321 mild-cognitive impairment (MCI) subjects, and 143 AD subjects-looking for dose-dependent effects of the ε4 and ε2 alleles on hippocampal volumes. Volumes were measured using a fully-automated algorithm applied to high resolution T1-weighted MR images. Statistical analysis consisted of a multivariate regression with repeated-measures model.

RESULTS

There was a dose-dependent effect of the ε4 allele on hippocampal volume in AD (p = 0.04) and MCI (p = 0.02)-in both cases, each allele accounted for loss of >150 mm(3) (approximately 4%) of hippocampal volume below the mean volume for AD and MCI subjects with no such alleles (Cohen's d = -0.16 and -0.19 for AD and MCI, respectively). There was also a dose-dependent, main effect of the ε2 allele (p<0.0001), suggestive of a moderate protective effect on hippocampal volume-an approximately 20% per allele volume increase as compared to CN with no ε2 alleles (Cohen's d = 0.23).

CONCLUSION

Though no effect of ε4 was seen in CN subjects, our findings confirm and extend prior data on the opposing effects of the APOE ε4 and ε2 alleles on hippocampal morphology across the spectrum of cognitive aging.

Automatic statistical shape analysis of cerebral asymmetry in 3D T1-weighted magnetic resonance images at vertex-level: application to neuroleptic-naïve schizophrenia

The study of the structural asymmetries in the human brain can assist the early diagnosis and progression of various neuropsychiatric disorders, and give insights into the biological bases of several cognitive deficits. The high inter-subject variability in cortical morphology complicates the detection of abnormal asymmetries especially if only small samples are available. This work introduces a novel automatic method for the local (vertex-level) statistical shape analysis of gross cerebral hemispheric surface asymmetries which is robust to the individual cortical variations. After segmentation of the cerebral hemispheric volumes from three-dimensional (3D) T1-weighted magnetic resonance images (MRI) and their spatial normalization to a common space, the right hemispheric masks were reflected to match with the left ones. Cerebral hemispheric surfaces were extracted using a deformable model-based algorithm which extracted the salient morphological features while establishing the point correspondence between the surfaces. The interhemispheric asymmetry, quantified by customized measures of asymmetry, was evaluated in a few thousands of corresponding surface vertices and tested for statistical significance. The developed method was tested on scans obtained from a small sample of healthy volunteers and first-episode neuroleptic-naïve schizophrenics. A significant main effect of the disease on the local interhemispheric asymmetry was observed, both in females and males, at the frontal and temporal lobes, the latter being often linked to the cognitive, auditory, and memory deficits in schizophrenia. The findings of this study, although need further testing in larger samples, partially replicate previous studies supporting the hypothesis of schizophrenia as a neurodevelopmental disorder.

Automated VOI Analysis in FDDNP PET Using Structural Warping: Validation through Classification of Alzheimer's Disease Patients

We evaluate an automated approach to the cortical surface mapping (CSM) method of VOI analysis in PET. Although CSM has been previously shown to be successful, the process can be long and tedious. Here, we present an approach that removes these difficulties through the use of 3D image warping to a common space. We test this automated method using studies of FDDNP PET in Alzheimer's disease and mild cognitive impairment. For each subject, VOIs were created, through CSM, to extract regional PET data. After warping to the common space, a single set of CSM-generated VOIs was used to extract PET data from all subjects. The data extracted using a single set of VOIs outperformed the manual approach in classifying AD patients from MCIs and controls. This suggests that this automated method can remove variance in measurements of PET data and can facilitate accurate, high-throughput image analysis.

Prediction of cognitive decline based on hemispheric cortical surface maps of FDDNP PET

OBJECTIVES

A cross-sectional study to establish whether a subject's cognitive state can be predicted based on regional values obtained from brain cortical maps of FDDNP Distribution Volume Ratio (DVR), which shows the pattern of beta amyloid and neurofibrillary binding, along with those of early summed FDDNP PET images (reflecting the pattern of perfusion) was performed.

METHODS

Dynamic FDDNP PET studies were performed in a group of 23 subjects (8 control (NL), 8 Mild Cognitive Impairment (MCI) and 7 Alzheimer's Disease (AD) subjects). FDDNP DVR images were mapped to the MR derived hemispheric cortical surface map warped into a common space. A set of Regions of Interest (ROI) values of FDDNP DVR and early summed FDDNP PET (0-6 min post tracer injection), were thus calculated for each subject which along with the MMSE score were used to construct a linear mathematical model relating ROI values to MMSE. After the MMSE prediction models were developed, the models' predictive ability was tested in a non-overlapping set of 8 additional individuals, whose cognitive status was unknown to the investigators who constructed the predictive models.

RESULTS

Among all possible subsets of ROIs, we found that the standard deviation of the predicted MMSE was 1.8 by using only DVR values from medial and lateral temporal and prefrontal regions plus the early summed FDDNP value in the posterior cingulate gyrus. The root mean square prediction error for the eight new subjects was 1.6.

CONCLUSION

FDDNP scans reflect progressive neuropathology accumulation and can potentially be used to predict the cognitive state of an individual.

ICA-fNORM: Spatial Normalization of fMRI Data Using Intrinsic Group-ICA Networks

A common pre-processing challenge associated with group level fMRI analysis is spatial registration of multiple subjects to a standard space. Spatial normalization, using a reference image such as the Montreal Neurological Institute brain template, is the most common technique currently in use to achieve spatial congruence across multiple subjects. This method corrects for global shape differences preserving regional asymmetries, but does not account for functional differences. We propose a novel approach to co-register task-based fMRI data using resting state group-ICA networks. We posit that these intrinsic networks (INs) can provide to the spatial normalization process with important information about how each individual’s brain is organized functionally. The algorithm is initiated by the extraction of single subject representations of INs using group level independent component analysis (ICA) on resting state fMRI data. In this proof-of-concept work two of the robust, commonly identified, networks are chosen as functional templates. As an estimation step, the relevant INs are utilized to derive a set of normalization parameters for each subject. Finally, the normalization parameters are applied individually to a different set of fMRI data acquired while the subjects performed an auditory oddball task. These normalization parameters, although derived using rest data, generalize successfully to data obtained with a cognitive paradigm for each subject. The improvement in results is verified using two widely applied fMRI analysis methods: the general linear model and ICA. Resulting activation patterns from each analysis method show significant improvements in terms of detection sensitivity and statistical significance at the group level. The results presented in this article provide initial evidence to show that common functional domains from the resting state brain may be used to improve the group statistics of task-fMRI data.

Automated landmark identification for human cortical surface-based registration

Volume-based registration (VBR) is the predominant method used in human neuroimaging to compensate for individual variability. However, surface-based registration (SBR) techniques have an inherent advantage over VBR because they respect the topology of the convoluted cortical sheet. There is evidence that existing SBR methods indeed confer a registration advantage over affine VBR. Landmark-SBR constrains registration using explicit landmarks to represent corresponding geographical locations on individual and atlas surfaces. The need for manual landmark identification has been an impediment to the widespread adoption of Landmark-SBR. To circumvent this obstacle, we have implemented and evaluated an automated landmark identification (ALI) algorithm for registration to the human PALS-B12 atlas. We compared ALI performance with that from two trained human raters and one expert anatomical rater (ENR). We employed both quantitative and qualitative quality assurance metrics, including a biologically meaningful analysis of hemispheric asymmetry. ALI performed well across all quality assurance tests, indicating that it yields robust and largely accurate results that require only modest manual correction (<10 min per subject). ALI largely circumvents human error and bias and enables high throughput analysis of large neuroimaging datasets for inter-subject registration to an atlas.

Mouse phenotyping with MRI

The field of mouse phenotyping with magnetic resonance imaging (MRI) is rapidly growing, motivated by the need for improved tools for characterizing and evaluating mouse models of human disease. Image results can provide important comparisons of human conditions with mouse disease models, evaluations of treatment, development or disease progression, as well as direction for histological or other investigations. Effective mouse MRI studies require attention to many aspects of experiment design. In this chapter, we provide details and discussion of important practical considerations: hardware requirements, mouse handling for in vivo imaging, specimen preparation for ex vivo imaging, sequence and contrast agent selection, study size, and quantitative image analysis. We focus particularly on anatomical phenotyping, an important and accessible application that has shown a high potential for impact in many mouse models at our imaging center.

Algorithms to improve the reparameterization of spherical mappings of brain surface meshes

A spherical map of a cortical surface is often used for improved brain registration, for advanced morphometric analysis (eg, of brain shape), and for surface-based analysis of functional signals recorded from the cortex. Furthermore, for intersubject analysis, it is usually necessary to reparameterize the surface mesh into a common coordinate system. An isometric map conserves all angle and area information in the original cortical mesh; however, in practice, spherical maps contain some distortion. Here, we propose fast new algorithms to reduce the distortion of initial spherical mappings generated using one of three common spherical mapping methods. The algorithms iteratively solve a nonlinear optimization problem to reduce distortion. Our results demonstrate that our correction process is computationally inexpensive and the resulting spherical maps have improved distortion metrics. We show that our corrected spherical maps improve reparameterization of the cortical surface mesh, such that the distance error measures between the original and reparameterized surface are significantly decreased.

A nonconservative Lagrangian framework for statistical fluid registration-SAFIRA

In this paper, we used a nonconservative Lagrangian mechanics approach to formulate a new statistical algorithm for fluid registration of 3-D brain images. This algorithm is named SAFIRA, acronym for statistically-assisted fluid image registration algorithm. A nonstatistical version of this algorithm was implemented , where the deformation was regularized by penalizing deviations from a zero rate of strain. In , the terms regularizing the deformation included the covariance of the deformation matrices (Σ) and the vector fields (q) . Here, we used a Lagrangian framework to reformulate this algorithm, showing that the regularizing terms essentially allow nonconservative work to occur during the flow. Given 3-D brain images from a group of subjects, vector fields and their corresponding deformation matrices are computed in a first round of registrations using the nonstatistical implementation. Covariance matrices for both the deformation matrices and the vector fields are then obtained and incorporated (separately or jointly) in the nonconservative terms, creating four versions of SAFIRA. We evaluated and compared our algorithms' performance on 92 3-D brain scans from healthy monozygotic and dizygotic twins; 2-D validations are also shown for corpus callosum shapes delineated at midline in the same subjects. After preliminary tests to demonstrate each method, we compared their detection power using tensor-based morphometry (TBM), a technique to analyze local volumetric differences in brain structure. We compared the accuracy of each algorithm variant using various statistical metrics derived from the images and deformation fields. All these tests were also run with a traditional fluid method, which has been quite widely used in TBM studies. The versions incorporating vector-based empirical statistics on brain variation were consistently more accurate than their counterparts, when used for automated volumetric quantification in new brain images. This suggests the advantages of this approach for large-scale neuroimaging studies.

Clinical application of SPHARM-PDM to quantify temporomandibular joint osteoarthritis

The severe bone destruction and resorption that can occur in osteoarthritis of the temporomandibular joint (TMJ) is associated with significant pain and limited joint mobility. However, there is no validated method for the quantification of discrete changes in joint morphology in early diagnosis or assessment of disease progression or treatment effects. To achieve this, the objective of this cross-sectional study was to use simulated bone resorption on cone-beam CT (CBCT) to study condylar morphological variation in subjects with temporomandibular joint (TMJ) osteoarthritis (OA). The first part of this study assessed the hypothesis that the agreement between the simulated defects and the shape analysis measurements made of these defects would be within 0.5mm (the image's spatial resolution). One hundred seventy-nine discrete bony defects measuring 3mm and 6mm were simulated on the surfaces of 3D models derived from CBCT images of asymptomatic patients using ITK-Snap software. SPHARM shape correspondence was used to localize and quantify morphological differences of each resorption model with the original asymptomatic control. The size of each simulated defect was analyzed and the values obtained compared to the true defect size. The statistical analysis revealed very high probabilities that mean shape correspondence measured defects within 0.5mm of the true defect size. 95% confidence intervals (CI) were (2.67, 2.92) and (5.99, 6.36) and 95% prediction intervals (PI) were (2.22, 3.37) and (5.54, 6.82), respectively for 3mm and 6mm simulated defects. The second part of this study applied shape correspondence methods to a longitudinal sample of TMJ OA patients. The mapped longitudinal stages of TMJ OA progression identified morphological variants or subtypes, which may explain the heterogeneity of the clinical presentation. This study validated shape correspondence as a method to precisely and predictably quantify 3D condylar resorption.

General multivariate linear modeling of surface shapes using SurfStat

Although there are many imaging studies on traditional ROI-based amygdala volumetry, there are very few studies on modeling amygdala shape variations. This paper presents a unified computational and statistical framework for modeling amygdala shape variations in a clinical population. The weighted spherical harmonic representation is used to parameterize, smooth out, and normalize amygdala surfaces. The representation is subsequently used as an input for multivariate linear models accounting for nuisance covariates such as age and brain size difference using the SurfStat package that completely avoids the complexity of specifying design matrices. The methodology has been applied for quantifying abnormal local amygdala shape variations in 22 high functioning autistic subjects.

Uncovering and improving upon the inherent deficiencies of radiology reporting through data mining

Uncertainty has been the perceived Achilles heel of the radiology report since the inception of the free-text report. As a measure of diagnostic confidence (or lack thereof), uncertainty in reporting has the potential to lead to diagnostic errors, delayed clinical decision making, increased cost of healthcare delivery, and adverse outcomes. Recent developments in data mining technologies, such as natural language processing (NLP), have provided the medical informatics community with an opportunity to quantify report concepts, such as uncertainty. The challenge ahead lies in taking the next step from quantification to understanding, which requires combining standardized report content, data mining, and artificial intelligence; thereby creating Knowledge Discovery Databases (KDD). The development of this database technology will expand our ability to record, track, and analyze report data, along with the potential to create data-driven and automated decision support technologies at the point of care. For the radiologist community, this could improve report content through an objective and thorough understanding of uncertainty, identifying its causative factors, and providing data-driven analysis for enhanced diagnosis and clinical outcomes.

Automated brain tissue segmentation based on fractional signal mapping from inversion recovery Look-Locker acquisition

Most current automated segmentation methods are performed on T(1)- or T(2)-weighted MR images, relying on relative image intensity that is dependent on other MR parameters and sensitive to B(1) magnetic field inhomogeneity. Here, we propose an image segmentation method based on quantitative longitudinal magnetization relaxation time (T(1)) of brain tissues. Considering the partial volume effect, fractional volume maps of brain tissues (white matter, gray matter, and cerebrospinal fluid) were obtained by fitting the observed signal in an inversion recovery procedure to a linear combination of three exponential functions, which represents the relaxations of each of the tissue types. A Look-Locker acquisition was employed to accelerate the acquisition process. The feasibility and efficacy of this proposed method were evaluated using simulations and experiments. The potential applications of this method in the study of neurological disease as well as normal brain development and aging are discussed.

Cranial base superimposition for 3-dimensional evaluation of soft-tissue changes

INTRODUCTION

The recent emphases on soft tissues as the limiting factor in treatment and on soft-tissue relationships in establishing the goals of treatment has made 3-dimensional (3D) analysis of soft tissues more important in diagnosis and treatment planning. It is equally important to be able to detect changes in the facial soft tissues produced by growth or treatment. This requires structures of reference for superimposition and a way to display the changes with quantitative information.

METHODS

In this study, we outlined a technique for quantifying facial soft-tissue changes viewed in cone-beam computed tomography data, using fully automated voxel-wise registrations of the cranial base surface. The assessment of soft-tissue changes is done by calculation of the Euclidean surface distances between the 3D models. Color maps are used for visual assessment of the location and the quantification of changes.

RESULTS

This methodology allows a detailed examination of soft-tissue changes with growth or treatment.

CONCLUSIONS

Because of the lack of stable references with 3D photogrammetry, 3D photography, and laser scanning, soft-tissue changes cannot be accurately quantified by these methods.

Study-specific EPI template improves group analysis in functional MRI of young and older adults

Spatial normalization to a common coordinate space, e.g. via the Montreal Neurological Institute (MNI) brain template, is an essential step of analyzing multi-subject functional MRI (fMRI) datasets. The imperfect compensation for individual regional discrepancies during spatial transformation, which could potentially introduce localization errors of the activation foci and/or reduce the detection sensitivity, may be minimized if a template specifically designed for the subjects of a study is applied. In this fMRI study, we proposed and evaluated the use of a study-specific template (SST) based on the mean of individually normalized echo-planar images for group data analysis. A hand flexion and a word generation tasks were performed on young volunteers in experiment 1. Comparing with the MNI template approach, greater t-values of local maxima and activated voxels were detected within volume-of-interests (VOIs) with the SST approach in both tasks. Moreover, the SST approach reduced Euclidean distances between activation foci of individuals and group by 1.52 mm in motor fMRI and 5.84 mm in language fMRI. Similar results were obtained with or without spatial smoothing of the echo-planar images. Experiment 2 further examined these two approaches in older adults, in which volumetric differences between subjects are of great concerns. With a working memory task, the SST approach showed greater t-values of local maxima and activated voxels within the VOI of prefrontal gyrus. This study demonstrated that the SST resulted in more focused activation patterns and effectively improved the fMRI sensitivity, which suggested potentials of reducing number of subjects required for group analysis.

Methods of MRI-based structural imaging in the aging monkey

Rhesus monkeys, whose typical lifespan can be as long as 30 years in the presence of veterinary care, undergo a cognitive decline as a function of age. While cortical neurons are largely preserved in the cerebral cortex, including primary motor and visual cortex as well as prefrontal association cortex there is marked breakdown of axonal myelin and an overall reduction in white matter predominantly in the frontal and temporal lobes. Whether the myelin breakdown is diffuse or specific to individual white matter fiber pathways is important to be known with certainty. To this end the delineation and quantification of specific frontotemporal fiber pathways within the frontal and temporal lobes is essential to determine which structures are altered and the extent to which these alterations correlate with behavioral findings. The capability of studying the living brain non-invasively with MRI opens up a new window in structural-functional and anatomic-clinical relationships allowing the integration of information derived from different scanning modalities in the same subject. For instance, for any particular voxel in the cerebrum we can obtain structural T1-, diffusion- and magnetization transfer- magnetic resonance imaging (MRI) based information. Moreover, it is thus possible to follow any observed changes longitudinally over time. These acquisitions of multidimensional data in the same individual within the same MRI experimental setting would enable the creation of a data base of integrated structural MRI-behavioral correlations for normal aging monkeys to elucidate the underlying neurobiological mechanisms of functional senescence in the aging non-human primate.

Multivariate tensor-based morphometry on surfaces: application to mapping ventricular abnormalities in HIV/AIDS

Here we developed a new method, called multivariate tensor-based surface morphometry (TBM), and applied it to study lateral ventricular surface differences associated with HIV/AIDS. Using concepts from differential geometry and the theory of differential forms, we created mathematical structures known as holomorphic one-forms, to obtain an efficient and accurate conformal parameterization of the lateral ventricular surfaces in the brain. The new meshing approach also provides a natural way to register anatomical surfaces across subjects, and improves on prior methods as it handles surfaces that branch and join at complex 3D junctions. To analyze anatomical differences, we computed new statistics from the Riemannian surface metrics-these retain multivariate information on local surface geometry. We applied this framework to analyze lateral ventricular surface morphometry in 3D MRI data from 11 subjects with HIV/AIDS and 8 healthy controls. Our method detected a 3D profile of surface abnormalities even in this small sample. Multivariate statistics on the local tensors gave better effect sizes for detecting group differences, relative to other TBM-based methods including analysis of the Jacobian determinant, the largest and smallest eigenvalues of the surface metric, and the pair of eigenvalues of the Jacobian matrix. The resulting analysis pipeline may improve the power of surface-based morphometry studies of the brain.

Sulcal set optimization for cortical surface registration

Flat mapping based cortical surface registration constrained by manually traced sulcal curves has been widely used for inter subject comparisons of neuroanatomical data. Even for an experienced neuroanatomist, manual sulcal tracing can be quite time consuming, with the cost increasing with the number of sulcal curves used for registration. We present a method for estimation of an optimal subset of size N(C) from N possible candidate sulcal curves that minimizes a mean squared error metric over all combinations of N(C) curves. The resulting procedure allows us to estimate a subset with a reduced number of curves to be traced as part of the registration procedure leading to optimal use of manual labeling effort for registration. To minimize the error metric we analyze the correlation structure of the errors in the sulcal curves by modeling them as a multivariate Gaussian distribution. For a given subset of sulci used as constraints in surface registration, the proposed model estimates registration error based on the correlation structure of the sulcal errors. The optimal subset of constraint curves consists of the N(C) sulci that jointly minimize the estimated error variance for the subset of unconstrained curves conditioned on the N(C) constraint curves. The optimal subsets of sulci are presented and the estimated and actual registration errors for these subsets are computed.

Multimodal surface-based morphometry reveals diffuse cortical atrophy in traumatic brain injury

BACKGROUND

Patients with traumatic brain injury (TBI) often present with significant cognitive deficits without corresponding evidence of cortical damage on neuroradiological examinations. One explanation for this puzzling observation is that the diffuse cortical abnormalities that characterize TBI are difficult to detect with standard imaging procedures. Here we investigated a patient with severe TBI-related cognitive impairments whose scan was interpreted as normal by a board-certified radiologist in order to determine if quantitative neuroimaging could detect cortical abnormalities not evident with standard neuroimaging procedures.

METHODS

Cortical abnormalities were quantified using multimodal surfaced-based morphometry (MSBM) that statistically combined information from high-resolution structural MRI and diffusion tensor imaging (DTI). Normal values of cortical anatomy and cortical and pericortical DTI properties were quantified in a population of 43 healthy control subjects. Corresponding measures from the patient were obtained in two independent imaging sessions. These data were quantified using both the average values for each lobe and the measurements from each point on the cortical surface. The results were statistically analyzed as z-scores from the mean with a p < 0.05 criterion, corrected for multiple comparisons. False positive rates were verified by comparing the data from each control subject with the data from the remaining control population using identical statistical procedures.

RESULTS

The TBI patient showed significant regional abnormalities in cortical thickness, gray matter diffusivity and pericortical white matter integrity that replicated across imaging sessions. Consistent with the patient's impaired performance on neuropsychological tests of executive function, cortical abnormalities were most pronounced in the frontal lobes.

CONCLUSIONS

MSBM is a promising tool for detecting subtle cortical abnormalities with high sensitivity and selectivity. MSBM may be particularly useful in evaluating cortical structure in TBI and other neurological conditions that produce diffuse abnormalities in both cortical structure and tissue properties.

FDDNP binding using MR derived cortical surface maps

OBJECTIVES

To assess quantitatively the cortical pattern profile of regional FDDNP binding to beta-amyloid and neurofibrillary tangles on MR derived cortical maps, FDDNP PET images were corrected for movement and partial volume (PV), and optimized for kernel size.

METHODS

FDDNP DVR PET images from 23 subjects (7 with Alzheimer's disease (AD), 6 with mild cognitive impairment and 10 controls) were obtained from Logan analysis using cerebellum as reference. A hemispheric cortical surface model for each subject was extracted from the MRI. The same transformations were applied to the FDDNP DVR PET images to map them into the same space. The cortical map with PV correction was calculated as the ratio of the DVR cortical surface and that of the simulated map, created from the mask derived from MRI and smoothed to the PET resolution. Discriminant analysis was used to order the FDDNP DVR cortical surfaces based on subjects' disease state. Linear regression was used to assess the rate of change of DVR vs. MMSE for each hemispheric cortical surface point.

RESULTS

The FDDNP DVR cortical surface corrected for movement and PV had less hemispheric asymmetry. Optimal kernel size was determined to be 9 mm. The corrected cortical surface map of FDDNP DVR showed clear spatial pattern that was consistent with the known pathological progression of AD.

CONCLUSION

Correcting for movement, PV as well as optimizing kernel size provide sensitive statistical analysis of FDDNP distribution which confirms in the living brain known pathology patterns earlier observed with cognitive decline with brain specimens.

Probabilistic atlas can improve reconstruction from optical imaging of the neonatal brain

Diffuse optical imaging is an emerging medical imaging modality based on near-infrared and visible red light. The method can be used for imaging activations in the human brain. In this study, a deformable probabilistic atlas of the distribution of tissue types within the term neonatal head was created based on MR images. The use of anatomical prior information provided by such atlas in reconstructing brain activations from optical imaging measurements was studied using Monte Carlo simulations. The results suggest that use of generic anatomical information can greatly improve the spatial accuracy and robustness of the reconstruction when noise is present in the data.

Callosal atrophy in mild cognitive impairment and Alzheimer's disease: different effects in different stages

Alzheimer's Disease (AD) is a neurodegenerative disorder that mainly affects grey matter (GM). Nevertheless, a number of investigations have documented white matter (WM) pathology associated with AD. The corpus callosum (CC) is the largest WM fiber bundle in the human brain. It has been shown to be susceptible to atrophy in AD mainly as a correlate of Wallerian degeneration of commissural nerve fibers of the neocortex. The aim of this study was to investigate which callosal regions are affected and whether callosal degeneration is associated with the stage of the disease. For this purpose, we analyzed high-resolution MRI data of patients with amnesic mild cognitive impairment (MCI) (n=20), mild AD (n=20), severe AD (n=10), and of healthy controls (n=20). Callosal morphology was investigated applying two different structural techniques: mesh-based geometrical modeling methods and whole-brain voxel-based analyses. Our findings indicate significant reductions in severe AD patients compared to healthy controls in anterior (genu and anterior body) and posterior (splenium) sections. In contrast, differences between healthy controls and mild AD patients or amnesic MCI patients were less pronounced and did not survive corrections for multiple comparisons. When correlating anterior and posterior WM density of the CC with GM density of the cortex in the severe AD group, we detected significant positive relationships between posterior sections of the CC and the cortex. We conclude that callosal atrophy is present predominantly in the latest stage of AD, where two mechanisms might contribute to WM alterations in severe AD: the Wallerian degeneration in posterior subregions and the myelin breakdown process in anterior subregions.

Volumetric and shape analysis of the thalamus in first-episode schizophrenia

Thalamic abnormalities have been implicated in the pathogenesis of schizophrenia, although the majority of studies used chronic samples treated extensively with antipsychotics. Moreover, the clinical and neuropsychological correlates of these abnormalities remain largely unknown. Using high-resolution MR imaging and novel methods for shape analysis, we investigated thalamic subregions in 35 (25 M/10 F) first-episode schizophrenia patients compared with 33 (23 M/10 F) healthy volunteers. The right and left thalami were traced bilaterally on coronal brain slices and volumes were compared between groups. In addition, regional abnormalities were identified by comparing distances, measured from homologous thalamic surface points to the central core of each individual's surface model, between groups in 3D space. Patients had significantly less total thalamic volume compared with healthy volunteers. Statistical mapping demonstrated most pronounced shape abnormalities in the pulvinar; however, estimated false discovery rates in these regions were sizable. Smaller thalamus volume was significantly correlated with worse overall neuropsychological functioning and specific deficits were observed in the language, motor, and executive domains. There were no significant associations between thalamus volume and positive or negative symptoms. Our findings suggest that thalamic abnormalities are evident at the onset of a first episode of schizophrenia prior to extensive pharmacologic intervention and that these abnormalities have neuropsychological correlates.

Neuroanatomical Correlates of Intelligence

With the advancement of image acquisition and analysis methods in recent decades, unique opportunities have emerged to study the neuroanatomical correlates of intelligence. Traditional approaches examining global measures have been complemented by insights from more regional analyses based on pre-defined areas. Newer state-of-the-art approaches have further enhanced our ability to localize the presence of correlations between cerebral characteristics and intelligence with high anatomic precision. These in vivo assessments have confirmed mainly positive correlations, suggesting that optimally increased brain regions are associated with better cognitive performance. Findings further suggest that the models proposed to explain the anatomical substrates of intelligence should address contributions from not only (pre)frontal regions, but also widely distributed networks throughout the whole brain.

Semi-automated method for delineation of landmarks on models of the cerebral cortex

Sulcal and gyral landmarks on the human cerebral cortex are required for various studies of the human brain. Whether used directly to examine sulcal geometry, or indirectly to drive cortical surface registration methods, the accuracy of these landmarks is essential. While several methods have been developed to automatically identify sulci and gyri, their accuracy may be insufficient for certain neuroanatomical studies. We describe a semi-automated procedure that delineates a sulcus or gyrus given a limited number of user-selected points. The method uses a graph theory approach to identify the lowest-cost path between the points, where the cost is a combination of local curvature features and the distance between vertices on the surface representation. We implemented the algorithm in an interface that guides the user through a cortical surface delineation protocol, and we incorporated this tool into our BrainSuite software. We performed a study to compare the results produced using our method with results produced using Display, a popular tool that has been used extensively for manual delineation of sulcal landmarks. Six raters were trained on the delineation protocol. They performed delineations on 12 brains using both software packages. We performed a statistical analysis of 3 aspects of the delineation task: time required to delineate the surface, registration accuracy achieved compared to an expert-delineated gold-standard, and variation among raters. Our new method was shown to be faster to use, to provide reduced inter-rater variability, and to provide results that were at least as accurate as those produced using Display.

Fully automatic segmentation of the brain from T1-weighted MRI using Bridge Burner algorithm

PURPOSE

To validate Bridge Burner, a new brain segmentation algorithm based on thresholding, connectivity, surface detection, and a new operator of constrained growing.

MATERIALS AND METHODS

T1-weighted MR images were selected at random from three previous neuroimaging studies to represent a spectrum of system manufacturers, pulse sequences, subject ages, genders, and neurological conditions. The ground truth consisted of brain masks generated manually by a consensus of expert observers. All cases were segmented using a common set of parameters.

RESULTS

Bridge Burner segmentation errors were 3.4% +/- 1.3% (volume mismatch) and 0.34 +/- 0.17 mm (surface mismatch). The disagreement among experts was 3.8% +/- 2.0% (volume mismatch) and 0.48 +/- 0.49 mm (surface mismatch). The error obtained using the brain extraction tool (BET), a widely used brain segmentation program, was 8.3% +/- 9.1%. Bridge Burner brain masks are visually similar to the masks generated by human experts. Areas affected by signal intensity nonuniformity artifacts were occasionally undersegmented, and meninges and large sinuses were often falsely classified as the brain tissue. Segmentation of one MRI dataset takes seven seconds.

CONCLUSION

The new fully automatic algorithm appears to provide accurate brain segmentation from high-resolution T1-weighted MR images.

Symmetric abnormalities in sulcal patterning in schizophrenia

To compare the morphology of the cerebral cortex and its characteristic pattern of gyri and sulci in individuals with and without schizophrenia, T1-weighted magnetic resonance scans were collected, along with clinical and cognitive information, from 33 individuals with schizophrenia and 30 healthy individuals group-matched for age, gender, race and parental socioeconomic status. Sulcal depth was measured across the entire cerebral cortex by reconstructing surfaces of cortical mid-thickness (layer 4) in each hemisphere and registering them to the human PALS cortical atlas. Group differences in sulcal depth were tested using methods for cluster size analysis and interhemispheric symmetry analysis. A significant group difference was found bilaterally in the parietal operculum, where the average sulcal depth was shallower in individuals with schizophrenia. In addition, group differences in sulcal depth showed significant bilateral symmetry across much of the occipital, parietal, and temporal cortices. In individuals with schizophrenia, sulcal depth in the left hemisphere was correlated with the severity of impaired performance on tests of working memory and executive function.

Brain volumetrics to investigate aging and the principal forms of degenerative cognitive decline: a brief review

The volume of the brain and of some of its structures can provide insight into the pathological process of several diseases. For this reason, in the recent years we saw a tremendous progress in the development of automated techniques for gaining information about global and regional atrophy. This paper reviews the main methods of analysis to quantify brain volume, and their application to the study of normal aging and the principal forms of degenerative dementias.

A review of geometric transformations for nonrigid body registration

This paper provides a comprehensive and quantitative review of spatial transformations models for nonrigid image registration. It explains the theoretical foundation of the models and classifies them according to this basis. This results in two categories, physically based models described by partial differential equations of continuum mechanics (e.g., linear elasticity and fluid flow) and basis function expansions derived from interpolation and approximation theory (e.g., radial basis functions, B-splines and wavelets). Recent work on constraining the transformation so that it preserves the topology or is diffeomorphic is also described. The final section reviews some recent evaluation studies. The paper concludes by explaining under what conditions a particular transformation model is appropriate.