Eigenanatomy: sparse dimensionality reduction for multi-modal medical image analysis.

Rigorous statistical analysis of multimodal imaging datasets is challenging. Mass-univariate methods for extracting correlations between image voxels and outcome measurements are not ideal for multimodal datasets, as they do not account for interactions between the different modalities. The extremely high dimensionality of medical images necessitates dimensionality reduction, such as principal component analysis (PCA) or independent component analysis (ICA). These dimensionality reduction techniques, however, consist of contributions from every region in the brain and are therefore difficult to interpret. Recent advances in sparse dimensionality reduction have enabled construction of a set of image regions that explain the variance of the images while still maintaining anatomical interpretability. The projections of the original data on the sparse eigenvectors, however, are highly collinear and therefore difficult to incorporate into multi-modal image analysis pipelines. We propose here a method for clustering sparse eigenvectors and selecting a subset of the eigenvectors to make interpretable predictions from a multi-modal dataset. Evaluation on a publicly available dataset shows that the proposed method outperforms PCA and ICA-based regressions while still maintaining anatomical meaning. To facilitate reproducibility, the complete dataset used and all source code is publicly available.

Decomposing cerebral blood flow MRI into functional and structural components: a non-local approach based on prediction.

We present RIPMMARC (Rotation Invariant Patch-based Multi-Modality Analysis aRChitecture), a flexible and widely applicable method for extracting information unique to a given modality from a multi-modal data set. We use RIPMMARC to improve the interpretation of arterial spin labeling (ASL) perfusion images by removing the component of perfusion that is predicted by the underlying anatomy. Using patch-based, rotation invariant descriptors derived from the anatomical image, we learn a predictive relationship between local neuroanatomical structure and the corresponding perfusion image. This relation allows us to produce an image of perfusion that would be predicted given only the underlying anatomy and a residual image that represents perfusion information that cannot be predicted by anatomical features. Our learned structural features are significantly better at predicting brain perfusion than tissue probability maps, which are the input to standard partial volume correction techniques. Studies in test-retest data show that both the anatomically predicted and residual perfusion signals are highly replicable for a given subject. In a pediatric population, both the raw perfusion and structurally predicted images are tightly linked to age throughout adolescence throughout the brain. Interestingly, the residual perfusion also shows a strong correlation with age in selected regions including the hippocampi (corr = 0.38, p-value <10(-6)), precuneus (corr = -0.44, p < 10(-5)), and combined default mode network regions (corr = -0.45, p < 10(-8)) that is independent of global anatomy-perfusion trends. This finding suggests that there is a regionally heterogeneous pattern of functional specialization that is distinct from that of cortical structural development.

Large-scale evaluation of ANTs and FreeSurfer cortical thickness measurements.

Many studies of the human brain have explored the relationship between cortical thickness and cognition, phenotype, or disease. Due to the subjectivity and time requirements in manual measurement of cortical thickness, scientists have relied on robust software tools for automation which facilitate the testing and refinement of neuroscientific hypotheses. The most widely used tool for cortical thickness studies is the publicly available, surface-based FreeSurfer package. Critical to the adoption of such tools is a demonstration of their reproducibility, validity, and the documentation of specific implementations that are robust across large, diverse imaging datasets. To this end, we have developed the automated, volume-based Advanced Normalization Tools (ANTs) cortical thickness pipeline comprising well-vetted components such as SyGN (multivariate template construction), SyN (image registration), N4 (bias correction), Atropos (n-tissue segmentation), and DiReCT (cortical thickness estimation). In this work, we have conducted the largest evaluation of automated cortical thickness measures in publicly available data, comparing FreeSurfer and ANTs measures computed on 1205 images from four open data sets (IXI, MMRR, NKI, and OASIS), with parcellation based on the recently proposed Desikan-Killiany-Tourville (DKT) cortical labeling protocol. We found good scan-rescan repeatability with both FreeSurfer and ANTs measures. Given that such assessments of precision do not necessarily reflect accuracy or an ability to make statistical inferences, we further tested the neurobiological validity of these approaches by evaluating thickness-based prediction of age and gender. ANTs is shown to have a higher predictive performance than FreeSurfer for both of these measures. In promotion of open science, we make all of our scripts, data, and results publicly available which complements the use of open image data sets and the open source availability of the proposed ANTs cortical thickness pipeline.

The relationship of head movements to semicircular canal size in cetaceans.

The semicircular canals measure head rotations, providing information critical for maintaining equilibrium. The canals of cetaceans (including whales, dolphins and porpoises) are extraordinarily small, making them unique exceptions to the allometric relationship shared by all other vertebrates between canal size and animal mass. Most modern cetaceans have shorter and less flexible necks than those of their ancestors, an adaptation hypothesized to have led to exaggerated head movements during locomotion. These movements are thought to have necessitated a decrease in the size and sensitivity of the canals, increasing their operating range to accommodate increased head motion. We tested whether the size of the semicircular canals in cetaceans is related to their head movements by comparing the rotational head velocities, frequencies and accelerations of the bottlenose dolphin (Tursiops truncatus) and a terrestrial relative, cattle (Bos taurus), using an array of three orthogonal head-fixed miniaturized rotational ratemeters. We collected data during typical locomotion (swimming; trotting) and during behaviors with enhanced head movements (rapid spiraling underwater; bucking). Cattle head movements always exceeded those of dolphins. Maximum head velocities were 528 deg. s(-1) in dolphins and 534 deg. s(-1) in cattle; maximum head frequencies were 2.86 Hz in dolphins and 3.45 Hz in cattle; and maximum head accelerations were 5253 deg. s(-2) in dolphins and 10,880 deg. s(-2) in cattle. These results indicate that accentuated head movements cannot explain the reduced size and sensitivity of cetacean semicircular canals. The evolutionary cause for their reduced canal size remains uncertain.

White matter hyperintensities are more highly associated with preclinical Alzheimer's disease than imaging and cognitive markers of neurodegeneration.

INTRODUCTION: Cognitive tests and nonamyloid imaging biomarkers do not consistently identify preclinical AD. The objective of this study was to evaluate whether white matter hyperintensity (WMH) volume, a cerebrovascular disease marker, is more associated with preclinical AD than conventional AD biomarkers and cognitive tests. METHODS: Elderly controls enrolled in the Alzheimer's Disease Neuroimaging Initiative (ADNI, n = 158) underwent florbetapir-PET scans, psychometric testing, neuroimaging with MRI and PET, and APOE genetic testing. Elderly controls the Parkinson's progression markers initiative (PPMI, n = 58) had WMH volume, cerebrospinal fluid (CSF) Aß1-42, and APOE status measured. RESULTS: In the ADNI cohort, only WMH volume and APOE ε4 status were associated with cerebral Aß (standardized ß = 0.44 and 1.25, P = .03 and .002). The association between WMH volume and APOE ε4 status with cerebral Aß (standardized ß = 1.12 and 0.26, P = .048 and .045) was confirmed in the PPMI cohort. DISCUSSION: WMH volume is more highly associated with preclinical AD than other AD biomarkers.

Arterial spin labeling perfusion predicts longitudinal decline in semantic variant primary progressive aphasia.

The objective of the study was to evaluate the prognostic value of regional cerebral blood flow (CBF) measured by arterial spin labeled (ASL) perfusion MRI in patients with semantic variant primary progressive aphasia (svPPA). We acquired pseudo-continuous ASL (pCASL) MRI and whole-brain T1-weighted structural MRI in svPPA patients (N = 13) with cerebrospinal fluid biomarkers consistent with frontotemporal lobar degeneration pathology. Follow-up T1-weighted MRI was available in a subset of patients (N = 8). We performed whole-brain comparisons of partial volume-corrected CBF and cortical thickness between svPPA and controls, and compared baseline and follow-up cortical thickness in regions of significant hypoperfusion and hyperperfusion. Patients with svPPA showed partial volume-corrected hypoperfusion relative to controls in left temporal lobe and insula. svPPA patients also had typical cortical thinning in anterior temporal, insula, and inferior frontal regions at baseline. Volume-corrected hypoperfusion was seen in areas of significant cortical thinning such as the left temporal lobe and insula. Additional regions of hypoperfusion corresponded to areas without cortical thinning. We also observed regions of hyperperfusion, some associated with cortical thinning and others without cortical thinning, including right superior temporal, inferior parietal, and orbitofrontal cortices. Regions of hypoperfusion and hyperperfusion near cortical thinning at baseline had significant longitudinal thinning between baseline and follow-up scans, but perfusion changes in distant areas did not show progressive thinning. Our findings suggest ASL MRI may be sensitive to functional changes not readily apparent in structural MRI, and specific changes in perfusion may be prognostic markers of disease progression in a manner consistent with cell-to-cell spreading pathology.

Neuropsychological Testing Predicts Cerebrospinal Fluid Amyloid-ß in Mild Cognitive Impairment.

BACKGROUND: Psychometric tests predict conversion of mild cognitive impairment (MCI) to probable Alzheimer's disease (AD). Because the definition of clinical AD relies on those same psychometric tests, the ability of these tests to identify underlying AD pathology remains unclear. OBJECTIVE: To determine the degree to which psychometric testing predicts molecular evidence of AD amyloid pathology, as indicated by cerebrospinal fluid (CSF) amyloid-ß (Aß)1 - 42, in patients with MCI, as compared to neuroimaging biomarkers. METHODS: We identified 408 MCI subjects with CSF Aß levels, psychometric test data, FDG-PET scans, and acceptable volumetric MR scans from the Alzheimer's Disease Neuroimaging Initiative (ADNI). We used psychometric tests and imaging biomarkers in univariate and multivariate models to predict Aß status. RESULTS: The 30-min delayed recall score of the Rey Auditory Verbal Learning Test was the best predictor of Aß status among the psychometric tests, achieving an AUC of 0.67 ± 0.02 and odds ratio of 2.5 ± 0.4. FDG-PET was the best imaging-based biomarker (AUC 0.67 ± 0.03, OR 3.2 ± 1.2), followed by hippocampal volume (AUC 0.64 ± 0.02, OR 2.4 ± 0.3). A multivariate analysis based on the psychometric tests improved on the univariate predictors, achieving an AUC of 0.68 ± 0.03 (OR 3.38 ± 1.2). Adding imaging biomarkers to the multivariate analysis did not improve the AUC. CONCLUSION: Psychometric tests perform as well as imaging biomarkers to predict presence of molecular markers of AD pathology in MCI patients and should be considered in the determination of the likelihood that MCI is due to AD.

Optimal Symmetric Multimodal Templates and Concatenated Random Forests for Supervised Brain Tumor Segmentation (Simplified) with ANTsR.

Segmenting and quantifying gliomas from MRI is an important task for diagnosis, planning intervention, and for tracking tumor changes over time. However, this task is complicated by the lack of prior knowledge concerning tumor location, spatial extent, shape, possible displacement of normal tissue, and intensity signature. To accommodate such complications, we introduce a framework for supervised segmentation based on multiple modality intensity, geometry, and asymmetry feature sets. These features drive a supervised whole-brain and tumor segmentation approach based on random forest-derived probabilities. The asymmetry-related features (based on optimal symmetric multimodal templates) demonstrate excellent discriminative properties within this framework. We also gain performance by generating probability maps from random forest models and using these maps for a refining Markov random field regularized probabilistic segmentation. This strategy allows us to interface the supervised learning capabilities of the random forest model with regularized probabilistic segmentation using the recently developed ANTsR package--a comprehensive statistical and visualization interface between the popular Advanced Normalization Tools (ANTs) and the R statistical project. The reported algorithmic framework was the top-performing entry in the MICCAI 2013 Multimodal Brain Tumor Segmentation challenge. The challenge data were widely varying consisting of both high-grade and low-grade glioma tumor four-modality MRI from five different institutions. Average Dice overlap measures for the final algorithmic assessment were 0.87, 0.78, and 0.74 for "complete", "core", and "enhanced" tumor components, respectively.

Single-subject structural networks with closed-form rotation invariant matching mprove power in developmental studies of the cortex.

Although much attention has recently been focused on single-subject functional networks, using methods such as resting-state functional MRI, methods for constructing single-subject structural networks are in their infancy. Single-subject cortical networks aim to describe the self-similarity across the cortical structure, possibly signifying convergent developmental pathways. Previous methods for constructing single-subject cortical networks have used patch-based correlations and distance metrics based on curvature and thickness. We present here a method for constructing similarity-based cortical structural networks that utilizes a rotation-invariant representation of structure. The resulting graph metrics are closely linked to age and indicate an increasing degree of closeness throughout development in nearly all brain regions, perhaps corresponding to a more regular structure as the brain matures. The derived graph metrics demonstrate a four-fold increase in power for detecting age as compared to cortical thickness. This proof of concept study indicates that the proposed metric may be useful in identifying biologically relevant cortical patterns.

Predicting cognitive data from medical images using sparse linear regression.

We present a new framework for predicting cognitive or other continuous-variable data from medical images. Current methods of probing the connection between medical images and other clinical data typically use voxel-based mass univariate approaches. These approaches do not take into account the multivariate, network-based interactions between the various areas of the brain and do not give readily interpretable metrics that describe how strongly cognitive function is related to neuroanatomical structure. On the other hand, high-dimensional machine learning techniques do not typically provide a direct method for discovering which parts of the brain are used for making predictions. We present a framework, based on recent work in sparse linear regression, that addresses both drawbacks of mass univariate approaches, while preserving the direct spatial interpretability that they provide. In addition, we present a novel optimization algorithm that adapts the conjugate gradient method for sparse regression on medical imaging data. This algorithm produces coefficients that are more interpretable than existing sparse regression techniques.

Eigenanatomy improves detection power for longitudinal cortical change.

We contribute a novel and interpretable dimensionality reduction strategy, eigenanatomy, that is tuned for neuroimaging data. The method approximates the eigendecomposition of an image set with basis functions (the eigenanatomy vectors) that are sparse, unsigned and are anatomically clustered. We employ the eigenanatomy vectors as anatomical predictors to improve detection power in morphometry. Standard voxel-based morphometry (VBM) analyzes imaging data voxel-by-voxel--and follows this with cluster-based or voxel-wise multiple comparisons correction methods to determine significance. Eigenanatomy reverses the standard order of operations by first clustering the voxel data and then using standard linear regression in this reduced dimensionality space. As with traditional region-of-interest (ROI) analysis, this strategy can greatly improve detection power. Our results show that eigenanatomy provides a principled objective function that leads to localized, data-driven regions of interest. These regions improve our ability to quantify biologically plausible rates of cortical change in two distinct forms of neurodegeneration. We detail the algorithm and show experimental evidence of its efficacy.