Image- versus histogram-based considerations in semantic segmentation of pulmonary hyperpolarized gas images.

PURPOSE: To characterize the differences between histogram-based and image-based algorithms for segmentation of hyperpolarized gas lung images. METHODS: Four previously published histogram-based segmentation algorithms (ie, linear binning, hierarchical k-means, fuzzy spatial c-means, and a Gaussian mixture model with a Markov random field prior) and an image-based convolutional neural network were used to segment 2 simulated data sets derived from a public (n = 29 subjects) and a retrospective collection (n = 51 subjects) of hyperpolarized 129Xe gas lung images transformed by common MRI artifacts (noise and nonlinear intensity distortion). The resulting ventilation-based segmentations were used to assess algorithmic performance and characterize optimization domain differences in terms of measurement bias and precision. RESULTS: Although facilitating computational processing and providing discriminating clinically relevant measures of interest, histogram-based segmentation methods discard important contextual spatial information and are consequently less robust in terms of measurement precision in the presence of common MRI artifacts relative to the image-based convolutional neural network. CONCLUSIONS: Direct optimization within the image domain using convolutional neural networks leverages spatial information, which mitigates problematic issues associated with histogram-based approaches and suggests a preferred future research direction. Further, the entire processing and evaluation framework, including the newly reported deep learning functionality, is available as open source through the well-known Advanced Normalization Tools ecosystem.

Neuroconductor: an R platform for medical imaging analysis.

Neuroconductor (https://neuroconductor.org) is an open-source platform for rapid testing and dissemination of reproducible computational imaging software. The goals of the project are to: (i) provide a centralized repository of R software dedicated to image analysis, (ii) disseminate software updates quickly, (iii) train a large, diverse community of scientists using detailed tutorials and short courses, (iv) increase software quality via automatic and manual quality controls, and (v) promote reproducibility of image data analysis. Based on the programming language R (https://www.r-project.org/), Neuroconductor starts with 51 inter-operable packages that cover multiple areas of imaging including visualization, data processing and storage, and statistical inference. Neuroconductor accepts new R package submissions, which are subject to a formal review and continuous automated testing. We provide a description of the purpose of Neuroconductor and the user and developer experience.

High-density three-dimensional localization microscopy across large volumes.

Extending three-dimensional (3D) single-molecule localization microscopy away from the coverslip and into thicker specimens will greatly broaden its biological utility. However, because of the limitations of both conventional imaging modalities and conventional labeling techniques, it is a challenge to localize molecules in three dimensions with high precision in such samples while simultaneously achieving the labeling densities required for high resolution of densely crowded structures. Here we combined lattice light-sheet microscopy with newly developed, freely diffusing, cell-permeable chemical probes with targeted affinity for DNA, intracellular membranes or the plasma membrane. We used this combination to perform high-localization precision, ultrahigh-labeling density, multicolor localization microscopy in samples up to 20 µm thick, including dividing cells and the neuromast organ of a zebrafish embryo. We also demonstrate super-resolution correlative imaging with protein-specific photoactivable fluorophores, providing a mutually compatible, single-platform alternative to correlative light-electron microscopy over large volumes.

Age-Related Effects and Sex Differences in Gray Matter Density, Volume, Mass, and Cortical Thickness from Childhood to Young Adulthood.

Developmental structural neuroimaging studies in humans have long described decreases in gray matter volume (GMV) and cortical thickness (CT) during adolescence. Gray matter density (GMD), a measure often assumed to be highly related to volume, has not been systematically investigated in development. We used T1 imaging data collected on the Philadelphia Neurodevelopmental Cohort to study age-related effects and sex differences in four regional gray matter measures in 1189 youths ranging in age from 8 to 23 years. Custom T1 segmentation and a novel high-resolution gray matter parcellation were used to extract GMD, GMV, gray matter mass (GMM; defined as GMD × GMV), and CT from 1625 brain regions. Nonlinear models revealed that each modality exhibits unique age-related effects and sex differences. While GMV and CT generally decrease with age, GMD increases and shows the strongest age-related effects, while GMM shows a slight decline overall. Females have lower GMV but higher GMD than males throughout the brain. Our findings suggest that GMD is a prime phenotype for the assessment of brain development and likely cognition and that periadolescent gray matter loss may be less pronounced than previously thought. This work highlights the need for combined quantitative histological MRI studies.SIGNIFICANCE STATEMENT This study demonstrates that different MRI-derived gray matter measures show distinct age and sex effects and should not be considered equivalent but complementary. It is shown for the first time that gray matter density increases from childhood to young adulthood, in contrast with gray matter volume and cortical thickness, and that females, who are known to have lower gray matter volume than males, have higher density throughout the brain. A custom preprocessing pipeline and a novel high-resolution parcellation were created to analyze brain scans of 1189 youths collected as part of the Philadelphia Neurodevelopmental Cohort. A clear understanding of normal structural brain development is essential for the examination of brain-behavior relationships, the study of brain disease, and, ultimately, clinical applications of neuroimaging.

Multi-template analysis of human perirhinal cortex in brain MRI: Explicitly accounting for anatomical variability.

RATIONAL: The human perirhinal cortex (PRC) plays critical roles in episodic and semantic memory and visual perception. The PRC consists of Brodmann areas 35 and 36 (BA35, BA36). In Alzheimer's disease (AD), BA35 is the first cortical site affected by neurofibrillary tangle pathology, which is closely linked to neural injury in AD. Large anatomical variability, manifested in the form of different cortical folding and branching patterns, makes it difficult to segment the PRC in MRI scans. Pathology studies have found that in ~97% of specimens, the PRC falls into one of three discrete anatomical variants. However, current methods for PRC segmentation and morphometry in MRI are based on single-template approaches, which may not be able to accurately model these discrete variants METHODS: A multi-template analysis pipeline that explicitly accounts for anatomical variability is used to automatically label the PRC and measure its thickness in T2-weighted MRI scans. The pipeline uses multi-atlas segmentation to automatically label medial temporal lobe cortices including entorhinal cortex, PRC and the parahippocampal cortex. Pairwise registration between label maps and clustering based on residual dissimilarity after registration are used to construct separate templates for the anatomical variants of the PRC. An optimal path of deformations linking these templates is used to establish correspondences between all the subjects. Experimental evaluation focuses on the ability of single-template and multi-template analyses to detect differences in the thickness of medial temporal lobe cortices between patients with amnestic mild cognitive impairment (aMCI, n=41) and age-matched controls (n=44). RESULTS: The proposed technique is able to generate templates that recover the three dominant discrete variants of PRC and establish more meaningful correspondences between subjects than a single-template approach. The largest reduction in thickness associated with aMCI, in absolute terms, was found in left BA35 using both regional and summary thickness measures. Further, statistical maps of regional thickness difference between aMCI and controls revealed different patterns for the three anatomical variants.

Enhanced estimations of post-stroke aphasia severity using stacked multimodal predictions.

The severity of post-stroke aphasia and the potential for recovery are highly variable and difficult to predict. Evidence suggests that optimal estimation of aphasia severity requires the integration of multiple neuroimaging modalities and the adoption of new methods that can detect multivariate brain-behavior relationships. We created and tested a multimodal framework that relies on three information sources (lesion maps, structural connectivity, and functional connectivity) to create an array of unimodal predictions which are then fed into a final model that creates "stacked multimodal predictions" (STAMP). Crossvalidated predictions of four aphasia scores (picture naming, sentence repetition, sentence comprehension, and overall aphasia severity) were obtained from 53 left hemispheric chronic stroke patients (age: 57.1 ± 12.3 yrs, post-stroke interval: 20 months, 25 female). Results showed accurate predictions for all four aphasia scores (correlation true vs. predicted: r = 0.79-0.88). The accuracy was slightly smaller but yet significant (r = 0.66) in a full split crossvalidation with each patient considered as new. Critically, multimodal predictions produced more accurate results that any single modality alone. Topological maps of the brain regions involved in the prediction were recovered and compared with traditional voxel-based lesion-to-symptom maps, revealing high spatial congruency. These results suggest that neuroimaging modalities carry complementary information potentially useful for the prediction of aphasia scores. More broadly, this study shows that the translation of neuroimaging findings into clinically useful tools calls for a shift in perspective from unimodal to multimodal neuroimaging, from univariate to multivariate methods, from linear to nonlinear models, and, conceptually, from inferential to predictive brain mapping. Hum Brain Mapp 38:5603-5615, 2017. © 2017 Wiley Periodicals, Inc.

A population level atlas of Mus musculus craniofacial skeleton and automated image-based shape analysis.

Laboratory mice are staples for evo/devo and genetics studies. Inbred strains provide a uniform genetic background to manipulate and understand gene-environment interactions, while their crosses have been instrumental in studies of genetic architecture, integration and modularity, and mapping of complex biological traits. Recently, there have been multiple large-scale studies of laboratory mice to further our understanding of the developmental basis, evolution, and genetic control of shape variation in the craniofacial skeleton (i.e. skull and mandible). These experiments typically use micro-computed tomography (micro-CT) to capture the craniofacial phenotype in 3D and rely on manually annotated anatomical landmarks to conduct statistical shape analysis. Although the common choice for imaging modality and phenotyping provides the potential for collaborative research for even larger studies with more statistical power, the investigator (or lab-specific) nature of the data collection hampers these efforts. Investigators are rightly concerned that subtle differences in how anatomical landmarks were recorded will create systematic bias between studies that will eventually influence scientific findings. Even if researchers are willing to repeat landmark annotation on a combined dataset, different lab practices and software choices may create obstacles for standardization beyond the underlying imaging data. Here, we propose a freely available analysis system that could assist in the standardization of micro-CT studies in the mouse. Our proposal uses best practices developed in biomedical imaging and takes advantage of existing open-source software and imaging formats. Our first contribution is the creation of a synthetic template for the adult mouse craniofacial skeleton from 25 inbred strains and five F1 crosses that are widely used in biological research. The template contains a fully segmented cranium, left and right hemi-mandibles, endocranial space, and the first few cervical vertebrae. We have been using this template in our lab to segment and isolate cranial structures in an automated fashion from a mixed population of mice, including craniofacial mutants, aged 4-12.5 weeks. As a secondary contribution, we demonstrate an application of nearly automated shape analysis, using symmetric diffeomorphic image registration. This approach, which we call diGPA, closely approximates the popular generalized Procrustes analysis (GPA) but negates the collection of anatomical landmarks. We achieve our goals by using the open-source advanced normalization tools (ANT) image quantification library, as well as its associated R library (ANTsR) for statistical image analysis. Finally, we make a plea to investigators to commit to using open imaging standards and software in their labs to the extent possible to increase the potential for data exchange and improve the reproducibility of findings. Future work will incorporate more anatomical detail (such as individual cranial bones, turbinals, dentition, middle ear ossicles) and more diversity into the template.

Automated segmentation of chronic stroke lesions using LINDA: Lesion identification with neighborhood data analysis.

The gold standard for identifying stroke lesions is manual tracing, a method that is known to be observer dependent and time consuming, thus impractical for big data studies. We propose LINDA (Lesion Identification with Neighborhood Data Analysis), an automated segmentation algorithm capable of learning the relationship between existing manual segmentations and a single T1-weighted MRI. A dataset of 60 left hemispheric chronic stroke patients is used to build the method and test it with k-fold and leave-one-out procedures. With respect to manual tracings, predicted lesion maps showed a mean dice overlap of 0.696 ± 0.16, Hausdorff distance of 17.9 ± 9.8 mm, and average displacement of 2.54 ± 1.38 mm. The manual and predicted lesion volumes correlated at r = 0.961. An additional dataset of 45 patients was utilized to test LINDA with independent data, achieving high accuracy rates and confirming its cross-institutional applicability. To investigate the cost of moving from manual tracings to automated segmentation, we performed comparative lesion-to-symptom mapping (LSM) on five behavioral scores. Predicted and manual lesions produced similar neuro-cognitive maps, albeit with some discussed discrepancies. Of note, region-wise LSM was more robust to the prediction error than voxel-wise LSM. Our results show that, while several limitations exist, our current results compete with or exceed the state-of-the-art, producing consistent predictions, very low failure rates, and transferable knowledge between labs. This work also establishes a new viewpoint on evaluating automated methods not only with segmentation accuracy but also with brain-behavior relationships. LINDA is made available online with trained models from over 100 patients.

Effect of socioeconomic status (SES) disparity on neural development in female African-American infants at age 1 month.

There is increasing interest in both the cumulative and long-term impact of early life adversity on brain structure and function, especially as the brain is both highly vulnerable and highly adaptive during childhood. Relationships between SES and neural development have been shown in children older than age 2 years. Less is known regarding the impact of SES on neural development in children before age 2. This paper examines the effect of SES, indexed by income-to-needs (ITN) and maternal education, on cortical gray, deep gray, and white matter volumes in term, healthy, appropriate for gestational age, African-American, female infants. At 5 weeks postnatal age, unsedated infants underwent MRI (3.0T Siemens Verio scanner, 32-channel head coil). Images were segmented based on a locally constructed template. Utilizing hierarchical linear regression, SES effects on MRI volumes were examined. In this cohort of healthy African-American female infants of varying SES, lower SES was associated with smaller cortical gray and deep gray matter volumes. These SES effects on neural outcome at such a young age build on similar studies of older children, suggesting that the biological embedding of adversity may occur very early in development.

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.

Regional and hemispheric variation in cortical thickness in chimpanzees (Pan troglodytes).

Recent advances in structural magnetic resonance imaging technology and analysis now allows for accurate in vivo measurement of cortical thickness, an important aspect of cortical organization that has historically only been conducted on postmortem brains. In this study, for the first time, we examined regional and lateralized cortical thickness in a sample of 71 chimpanzees for comparison with previously reported findings in humans. We also measured gray and white matter volumes for each subject. The results indicated that chimpanzees showed significant regional variation in cortical thickness with lower values in primary motor and sensory cortex compared with association cortex. Furthermore, chimpanzees showed significant rightward asymmetries in cortical thickness for a number of regions of interest throughout the cortex and leftward asymmetries in white but not gray matter volume. We also found that total and region-specific cortical thickness was significantly negatively correlated with white matter volume. Thus, chimpanzees with greater white matter volumes had thinner cortical thickness. The collective findings are discussed within the context of previous findings in humans and theories on the evolution of cortical organization and lateralization in primates.

Relating brain anatomy and cognitive ability using a multivariate multimodal framework.

Linking structural neuroimaging data from multiple modalities to cognitive performance is an important challenge for cognitive neuroscience. In this study we examined the relationship between verbal fluency performance and neuroanatomy in 54 patients with frontotemporal degeneration (FTD) and 15 age-matched controls, all of whom had T1- and diffusion-weighted imaging. Our goal was to incorporate measures of both gray matter (voxel-based cortical thickness) and white matter (fractional anisotropy) into a single statistical model that relates to behavioral performance. We first used eigenanatomy to define data-driven regions of interest (DD-ROIs) for both gray matter and white matter. Eigenanatomy is a multivariate dimensionality reduction approach that identifies spatially smooth, unsigned principal components that explain the maximal amount of variance across subjects. We then used a statistical model selection procedure to see which of these DD-ROIs best modeled performance on verbal fluency tasks hypothesized to rely on distinct components of a large-scale neural network that support language: category fluency requires a semantic-guided search and is hypothesized to rely primarily on temporal cortices that support lexical-semantic representations; letter-guided fluency requires a strategic mental search and is hypothesized to require executive resources to support a more demanding search process, which depends on prefrontal cortex in addition to temporal network components that support lexical representations. We observed that both types of verbal fluency performance are best described by a network that includes a combination of gray matter and white matter. For category fluency, the identified regions included bilateral temporal cortex and a white matter region including left inferior longitudinal fasciculus and frontal-occipital fasciculus. For letter fluency, a left temporal lobe region was also selected, and also regions of frontal cortex. These results are consistent with our hypothesized neuroanatomical models of language processing and its breakdown in FTD. We conclude that clustering the data with eigenanatomy before performing linear regression is a promising tool for multimodal data analysis.

Sparse canonical correlation analysis relates network-level atrophy to multivariate cognitive measures in a neurodegenerative population.

This study establishes that sparse canonical correlation analysis (SCCAN) identifies generalizable, structural MRI-derived cortical networks that relate to five distinct categories of cognition. We obtain multivariate psychometrics from the domain-specific sub-scales of the Philadelphia Brief Assessment of Cognition (PBAC). By using a training and separate testing stage, we find that PBAC-defined cognitive domains of language, visuospatial functioning, episodic memory, executive control, and social functioning correlate with unique and distributed areas of gray matter (GM). In contrast, a parallel univariate framework fails to identify, from the training data, regions that are also significant in the left-out test dataset. The cohort includes164 patients with Alzheimer's disease, behavioral-variant frontotemporal dementia, semantic variant primary progressive aphasia, non-fluent/agrammatic primary progressive aphasia, or corticobasal syndrome. The analysis is implemented with open-source software for which we provide examples in the text. In conclusion, we show that multivariate techniques identify biologically-plausible brain regions supporting specific cognitive domains. The findings are identified in training data and confirmed in test data.

Histology-derived volumetric annotation of the human hippocampal subfields in postmortem MRI.

Recently, there has been a growing effort to analyze the morphometry of hippocampal subfields using both in vivo and postmortem magnetic resonance imaging (MRI). However, given that boundaries between subregions of the hippocampal formation (HF) are conventionally defined on the basis of microscopic features that often lack discernible signature in MRI, subfield delineation in MRI literature has largely relied on heuristic geometric rules, the validity of which with respect to the underlying anatomy is largely unknown. The development and evaluation of such rules are challenged by the limited availability of data linking MRI appearance to microscopic hippocampal anatomy, particularly in three dimensions (3D). The present paper, for the first time, demonstrates the feasibility of labeling hippocampal subfields in a high resolution volumetric MRI dataset based directly on microscopic features extracted from histology. It uses a combination of computational techniques and manual post-processing to map subfield boundaries from a stack of histology images (obtained with 200µm spacing and 5µm slice thickness; stained using the Kluver-Barrera method) onto a postmortem 9.4Tesla MRI scan of the intact, whole hippocampal formation acquired with 160µm isotropic resolution. The histology reconstruction procedure consists of sequential application of a graph-theoretic slice stacking algorithm that mitigates the effects of distorted slices, followed by iterative affine and diffeomorphic co-registration to postmortem MRI scans of approximately 1cm-thick tissue sub-blocks acquired with 200µm isotropic resolution. These 1cm blocks are subsequently co-registered to the MRI of the whole HF. Reconstruction accuracy is evaluated as the average displacement error between boundaries manually delineated in both the histology and MRI following the sequential stages of reconstruction. The methods presented and evaluated in this single-subject study can potentially be applied to multiple hippocampal tissue samples in order to construct a histologically informed MRI atlas of the hippocampal formation.

Subject-specific functional parcellation via prior based eigenanatomy.

We present a new framework for prior-constrained sparse decomposition of matrices derived from the neuroimaging data and apply this method to functional network analysis of a clinically relevant population. Matrix decomposition methods are powerful dimensionality reduction tools that have found widespread use in neuroimaging. However, the unconstrained nature of these totally data-driven techniques makes it difficult to interpret the results in a domain where network-specific hypotheses may exist. We propose a novel approach, Prior Based Eigenanatomy (p-Eigen), which seeks to identify a data-driven matrix decomposition but at the same time constrains the individual components by spatial anatomical priors (probabilistic ROIs). We formulate our novel solution in terms of prior-constrained ℓ1 penalized (sparse) principal component analysis. p-Eigen starts with a common functional parcellation for all the subjects and refines it with subject-specific information. This enables modeling of the inter-subject variability in the functional parcel boundaries and allows us to construct subject-specific networks with reduced sensitivity to ROI placement. We show that while still maintaining correspondence across subjects, p-Eigen extracts biologically-relevant and patient-specific functional parcels that facilitate hypothesis-driven network analysis. We construct default mode network (DMN) connectivity graphs using p-Eigen refined ROIs and use them in a classification paradigm. Our results show that the functional connectivity graphs derived from p-Eigen significantly aid classification of mild cognitive impairment (MCI) as well as the prediction of scores in a Delayed Recall memory task when compared to graph metrics derived from 1) standard registration-based seed ROI definitions, 2) totally data-driven ROIs, 3) a model based on standard demographics plus hippocampal volume as covariates, and 4) Ward Clustering based data-driven ROIs. In summary, p-Eigen incarnates a new class of prior-constrained dimensionality reduction tools that may improve our understanding of the relationship between MCI and functional connectivity.

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 power of neuroimaging biomarkers for screening frontotemporal dementia.

Frontotemporal dementia (FTD) is a clinically and pathologically heterogeneous neurodegenerative disease that can result from either frontotemporal lobar degeneration (FTLD) or Alzheimer's disease (AD) pathology. It is critical to establish statistically powerful biomarkers that can achieve substantial cost-savings and increase the feasibility of clinical trials. We assessed three broad categories of neuroimaging methods to screen underlying FTLD and AD pathology in a clinical FTD series: global measures (e.g., ventricular volume), anatomical volumes of interest (VOIs) (e.g., hippocampus) using a standard atlas, and data-driven VOIs using Eigenanatomy. We evaluated clinical FTD patients (N = 93) with cerebrospinal fluid, gray matter (GM) magnetic resonance imaging (MRI), and diffusion tensor imaging (DTI) to assess whether they had underlying FTLD or AD pathology. Linear regression was performed to identify the optimal VOIs for each method in a training dataset and then we evaluated classification sensitivity and specificity in an independent test cohort. Power was evaluated by calculating minimum sample sizes required in the test classification analyses for each model. The data-driven VOI analysis using a multimodal combination of GM MRI and DTI achieved the greatest classification accuracy (89% sensitive and 89% specific) and required a lower minimum sample size (N = 26) relative to anatomical VOI and global measures. We conclude that a data-driven VOI approach using Eigenanatomy provides more accurate classification, benefits from increased statistical power in unseen datasets, and therefore provides a robust method for screening underlying pathology in FTD patients for entry into clinical trials.

Logical circularity in voxel-based analysis: normalization strategy may induce statistical bias.

Recent discussions within the neuroimaging community have highlighted the problematic presence of selection bias in experimental design. Although initially centering on the selection of voxels during the course of fMRI studies, we demonstrate how this bias can potentially corrupt voxel-based analyses. For such studies, template-based registration plays a critical role in which a representative template serves as the normalized space for group alignment. A standard approach maps each subject's image to a representative template before performing statistical comparisons between different groups. We analytically demonstrate that in these scenarios the popular sum of squared difference (SSD) intensity metric, implicitly surrogating as a quantification of anatomical alignment, instead explicitly maximizes effect size--an experimental design flaw referred to as "circularity bias." We illustrate how this selection bias varies in strength with the similarity metric used during registration under the hypothesis that while SSD-related metrics, such as Demons, will manifest similar effects, other metrics which are not formulated based on absolute intensity differences will produce less of an effect. Consequently, given the variability in voxel-based analysis outcomes with similarity metric choice, we caution researchers specifically in the use of SSD and SSD-related measures where normalization and statistical analysis involve the same image set. Instead, we advocate a more cautious approach where normalization of the individual subject images to the reference space occurs through corresponding image sets which are independent of statistical testing. Alternatively, one can use similarity terms that are less sensitive to this bias.