When the single matters more than the group: very high false positive rates in single case Voxel Based Morphometry

Voxel-based morphometry in single subjects without a scanner-specific normal database using a convolutional neural network

OBJECTIVES

Reliable detection of disease-specific atrophy in individual T1w-MRI by voxel-based morphometry (VBM) requires scanner-specific normal databases (NDB), which often are not available. The aim of this retrospective study was to design, train, and test a deep convolutional neural network (CNN) for single-subject VBM without the need for a NDB (CNN-VBM).

MATERIALS AND METHODS

The training dataset comprised 8945 T1w scans from 65 different scanners. The gold standard VBM maps were obtained by conventional VBM with a scanner-specific NDB for each of the 65 scanners. CNN-VBM was tested in an independent dataset comprising healthy controls (n = 37) and subjects with Alzheimer's disease (AD, n = 51) or frontotemporal lobar degeneration (FTLD, n = 30). A scanner-specific NDB for the generation of the gold standard VBM maps was available also for the test set. The technical performance of CNN-VBM was characterized by the Dice coefficient of CNN-VBM maps relative to VBM maps from scanner-specific VBM. For clinical testing, VBM maps were categorized visually according to the clinical diagnoses in the test set by two independent readers, separately for both VBM methods.

RESULTS

The VBM maps from CNN-VBM were similar to the scanner-specific VBM maps (median Dice coefficient 0.85, interquartile range [0.81, 0.90]). Overall accuracy of the visual categorization of the VBM maps for the detection of AD or FTLD was 89.8% for CNN-VBM and 89.0% for scanner-specific VBM.

CONCLUSION

CNN-VBM without NDB provides a similar performance in the detection of AD- and FTLD-specific atrophy as conventional VBM.

CLINICAL RELEVANCE STATEMENT

A deep convolutional neural network for voxel-based morphometry eliminates the need of scanner-specific normal databases without relevant performance loss and, therefore, could pave the way for the widespread clinical use of voxel-based morphometry to support the diagnosis of neurodegenerative diseases.

KEY POINTS

• The need of normal databases is a barrier for widespread use of voxel-based brain morphometry. • A convolutional neural network achieved a similar performance for detection of atrophy than conventional voxel-based morphometry. • Convolutional neural networks can pave the way for widespread clinical use of voxel-based morphometry.

Brain structural changes in alternating hemiplegia of childhood using single-case voxel-based morphometry analysis

BACKGROUND AND PURPOSE

Alternating hemiplegia of childhood (AHC) is a rare neurodevelopmental disease caused by ATP1A3 mutations. Using voxel-based morphometry (VBM) analysis, we compared an AHC patient cohort with controls. Additionally, with single-case VBM analysis, we assessed the associations between clinical severity and brain volume in patients with AHC.

MATERIALS AND METHODS

To investigate structural brain changes in gray matter (GM) and white matter (WM) volumes between 9 patients with AHC and 20 age-matched controls, VBM analysis was performed using three-dimensional T1-weighted magnetic resonance imaging. Single-case VBM analysis was also performed on nine patients with AHC to investigate the associations between the respective volumes of GM/WM differences and the motor level, cognitive level, and status epilepticus severity in patients with AHC.

RESULTS

Compared with controls, patients with AHC showed significant GM volume reductions in both hippocampi and diffuse cerebellum, and there were WM reductions in both cerebral hemispheres. In patients with AHC, cases with more motor dysfunction, the less GM/WM volume of cerebellum was shown. Three of the six cases with cognitive dysfunction showed a clear GM volume reduction in the insulae. Five of the six cases with status epilepticus showed the GM volume reduction in hippocampi. One case had severe status epilepticus without motor dysfunction and showed no cerebellar atrophy.

CONCLUSION

With single-case VBM analysis, we could show the association between region-specific changes in brain volume and the severity of various clinical symptoms even in a small sample of subjects.

Common and distinct fronto-striatal volumetric changes in heroin and cocaine use disorders

Different drugs of abuse impact the morphology of fronto-striatal dopaminergic targets in both common and unique ways. While dorsal striatal volume tracks with addiction severity across drug classes, opiates impact ventromedial prefrontal cortex (vmPFC) and nucleus accumbens (NAcc) neuroplasticity in preclinical models, and psychostimulants alter inhibitory control, rooted in cortical regions such as the inferior frontal gyrus (IFG). We hypothesized parallel grey matter volume changes associated with human heroin or cocaine use disorder: lower grey matter volume of vmPFC/NAcc in heroin use disorder and IFG in cocaine use disorder, and putamen grey matter volume to be associated with addiction severity measures (including craving) across both. In this cross-sectional study, we quantified grey matter volume (P < 0.05-corrected) in age/sex/IQ-matched individuals with heroin use disorder (n = 32, seven females), cocaine use disorder (n = 32, six females) and healthy controls (n = 32, six females) and compared fronto-striatal volume between groups using voxel-wise general linear models and non-parametric permutation-based tests. Overall, individuals with heroin use disorder had smaller vmPFC and NAcc/putamen volumes than healthy controls. Bilateral lower IFG grey matter volume patterns were specifically evident in cocaine versus heroin use disorders. Correlations between addiction severity measures and putamen grey matter volume did not reach nominal significance level in this sample. These results indicate alterations in dopamine-innervated regions (in the vmPFC and NAcc) in heroin addiction. For the first time we demonstrate lower IFG grey matter volume specifically in cocaine compared with heroin use disorder, suggesting a signature of reduced inhibitory control, which remains to be tested directly using select behavioural measures. Overall, results suggest substance-specific volumetric changes in human psychostimulant or opiate addiction, with implications for fine-tuning biomarker and treatment identification by primary drug of abuse.

Submillimeter T1 atlas for subject-specific abnormality detection at 7T

PURPOSE

Studies at 3T have shown that T₁ relaxometry enables characterization of brain tissues at the single-subject level by comparing individual physical properties to a normative atlas. In this work, an atlas of normative T₁ values at 7T is introduced with 0.6 mm isotropic resolution and its clinical potential is explored in comparison to 3T.

METHODS

T₁ maps were acquired in two separate healthy cohorts scanned at 3T and 7T. Using transfer learning, a template-based brain segmentation algorithm was adapted to ultra-high field imaging data. After segmenting brain tissues, volumes were normalized into a common space, and an atlas of normative T₁ values was established by modeling the T₁ inter-subject variability. A method for single-subject comparisons restricted to white matter and subcortical structures was developed by computing Z-scores. The comparison was applied to eight patients scanned at both field strengths for proof of concept.

RESULTS

The proposed method for morphometry delivered segmentation masks without statistically significant differences from those derived with the original pipeline at 3T and achieved accurate segmentation at 7T. The established normative atlas allowed characterizing tissue alterations in single-subject comparisons at 7T, and showed greater anatomical details compared with 3T results.

CONCLUSION

A high-resolution quantitative atlas with an adapted pipeline was introduced and validated. Several case studies on different clinical conditions showed the feasibility, potential and limitations of high-resolution single-subject comparisons based on quantitative MRI atlases. This method in conjunction with 7T higher resolution broadens the range of potential applications of quantitative MRI in clinical practice.

Personalized maps of T1 relaxometry abnormalities provide correlates of disability in multiple sclerosis patients

OBJECTIVES AND AIMS

Quantitative MRI (qMRI) has greatly improved the sensitivity and specificity of microstructural brain pathology in multiple sclerosis (MS) when compared to conventional MRI (cMRI). More than cMRI, qMRI also provides means to assess pathology within the normal-appearing and lesion tissue. In this work, we further developed a method providing personalized quantitative T1 (qT1) abnormality maps in individual MS patients by modeling the age dependence of qT1 alterations. In addition, we assessed the relationship between qT1 abnormality maps and patients' disability, in order to evaluate the potential value of this measurement in clinical practice.

METHODS

We included 119 MS patients (64 relapsing-remitting MS (RRMS), 34 secondary progressive MS (SPMS), 21 primary progressive MS (PPMS)), and 98 Healthy Controls (HC). All individuals underwent 3T MRI examinations, including Magnetization Prepared 2 Rapid Acquisition Gradient Echoes (MP2RAGE) for qT1 maps and High-Resolution 3D Fluid Attenuated Inversion Recovery (FLAIR) imaging. To calculate personalized qT1 abnormality maps, we compared qT1 in each brain voxel in MS patients to the average qT1 obtained in the same tissue (grey/white matter) and region of interest (ROI) in healthy controls, hereby providing individual voxel-based Z-score maps. The age dependence of qT1 in HC was modeled using linear polynomial regression. We computed the average qT1 Z-scores in white matter lesions (WMLs), normal-appearing white matter (NAWM), cortical grey matter lesions (GMcLs) and normal-appearing cortical grey matter (NAcGM). Lastly, a multiple linear regression (MLR) model with the backward selection including age, sex, disease duration, phenotype, lesion number, lesion volume and average Z-score (NAWM/NAcGM/WMLs/GMcLs) was used to assess the relationship between qT1 measures and clinical disability (evaluated with EDSS).

RESULTS

The average qT1 Z-score was higher in WMLs than in NAWM. (WMLs: 1.366 ± 0.409, NAWM: -0.133 ± 0.288, [mean ± SD], p < 0.001). The average Z-score in NAWM in RRMS patients was significantly lower than in PPMS patients (p = 0.010). The MLR model showed a strong association between average qT1 Z-scores in white matter lesions (WMLs) and EDSS (R² = 0.549, β = 0.178, 97.5 % CI = 0.030 to 0.326, p = 0.019). Specifically, we measured a 26.9 % increase in EDSS per unit of qT1 Z-score in WMLs in RRMS patients (R² = 0.099, β = 0.269, 97.5 % CI = 0.078 to 0.461, p = 0.007).

CONCLUSIONS

We showed that personalized qT1 abnormality maps in MS patients provide measures related to clinical disability, supporting the use of those maps in clinical practice.

Hyper-Adaptation in the Human Brain: Functional and Structural Changes in the Foot Section of the Primary Motor Cortex in a Top Wheelchair Racing Paralympian

The human brain has the capacity to drastically alter its somatotopic representations in response to congenital or acquired limb deficiencies and dysfunctions. The main purpose of the present study was to elucidate such extreme adaptability in the brain of an active top wheelchair racing Paralympian (participant P1) who has congenital paraplegia (dysfunction of bilateral lower limbs). Participant P1 has undergone long-term wheelchair racing training using bilateral upper limbs and has won a total of 19 medals in six consecutive summer Paralympic games as of 2021. We examined the functional and structural changes in the foot section of the primary motor cortex (M1) in participant P1 as compared to able-bodied control participants. We also examined the functional and structural changes in three other individuals (participants P2, P3, and P4) with acquired paraplegia, who also had long-term non-use period of the lower limbs and had undergone long-term training for wheelchair sports (but not top athletes at the level of participant P1). We measured brain activity in all the participants using functional magnetic resonance imaging (MRI) when bimanual wrist extension-flexion movement was performed, and the structural MRI images were collected. Compared to 37 control participants, participant P1 showed significantly greater activity in the M1 foot section during the bimanual task, and significant local GM expansion in this section. Significantly greater activity in the M1 foot section was also observed in participant P4, but not in P2 and P3, and the significant local GM expansion was observed in participant P2, but not in P3 and P4. Thus, functional or structural change was observed in an acquired paraplegic participant, but was not observed in all the paraplegic participants. The functional and structural changes typically observed in participant P1 may represent extreme adaptability of the human brain. We discuss the results in terms of a new idea of hyper-adaptation.

Translational Application of a Neuro-Scientific Multi-Modal Approach Into Forensic Psychiatric Evaluation: Why and How?

A prominent body of literature indicates that insanity evaluations, which are intended to provide influential expert reports for judges to reach a decision "beyond any reasonable doubt," suffer from a low inter-rater reliability. This paper reviews the limitations of the classical approach to insanity evaluation and the criticisms to the introduction of neuro-scientific approach in court. Here, we explain why in our opinion these criticisms, that seriously hamper the translational implementation of neuroscience into the forensic setting, do not survive scientific scrutiny. Moreover, we discuss how the neuro-scientific multimodal approach may improve the inter-rater reliability in insanity evaluation. Critically, neuroscience does not aim to introduce a brain-based concept of insanity. Indeed, criteria for responsibility and insanity are and should remain clinical. Rather, following the falsificationist approach and the convergence of evidence principle, the neuro-scientific multimodal approach is being proposed as a way to improve reliability of insanity evaluation and to mitigate the influence of cognitive biases on the formulation of insanity opinions, with the final aim to reduce errors and controversies.

A multidisciplinary approach to insanity assessment as a way to reduce cognitive biases

Insanity assessment requires the evaluation of the psychopathological condition that underlies the mens rea. Psychopathological evaluation may be quite challenging due to (i) absence of biomarkers; (ii) low inter-rater reliability; (iii) presence of cognitive bias. This intrinsic low reliability of forensic psychiatric diagnosis does impact on insanity assessment, leading to arbitrary and unjust legal outcomes for the examinee. Thus, strategies to improve the reliability of insanity evaluation are strongly needed. A multidisciplinary approach has been proposed as a way to enrich clinical diagnosis with reliable and biologically founded data, thus minimizing subjectivity, reducing controversies and increasing inter-subject concordance in insanity assessment. By discussing a real case, here we show how the convergence of multiple indices can produce evidence that cannot be denied without introducing logical fallacies. Applying this approach, the forensic discussion will move from the presence/absence of psychopathology to the impact of psychopathology on insanity. This article illustrates how a multidisciplinary evaluation, which integrates neuroscientific methods with the classical insanity assessment, may lead to a more accurate approach in insanity evaluation. Critically, this approach will minimize the impact of cognitive bias on insanity opinion and thus result in an improvement of the whole criminal justice process.

Voxel-Based Morphometry-from Hype to Hope. A Study on Hippocampal Atrophy in Mesial Temporal Lobe Epilepsy

BACKGROUND AND PURPOSE

Automated volumetry of the hippocampus is considered useful to assist the diagnosis of hippocampal sclerosis in temporal lobe epilepsy. However, voxel-based morphometry is rarely used for individual subjects because of high rates of false-positives. We investigated whether an approach with high dimensional warping to the template and nonparametric statistics would be useful to detect hippocampal atrophy in patients with hippocampal sclerosis.

MATERIALS AND METHODS

We performed single-subject voxel-based morphometry with nonparametric statistics within the framework of Statistical Parametric Mapping to compare MRI from 26 well-characterized patients with temporal lobe epilepsy individually against a group of 110 healthy controls. The following statistical threshold was used: P < .05 corrected for multiple comparisons with family-wise error over the region of interest right and left hippocampus.

RESULTS

The sensitivity for the detection of atrophy related to hippocampal sclerosis was 0.92 (95% CI, 0.67-0.99) for the right hippocampus and 0.60 (0.31-0.83) for the left, and the specificity for volume changes was 0.98 (0.93-0.99). All clusters of decreased hippocampal volumes were correctly lateralized to the seizure focus. Hippocampal volume decrease was in accordance with neuronal cell loss on histology reports.

CONCLUSIONS

Nonparametric voxel-based morphometry is sensitive and specific for hippocampal atrophy in patients with mesial temporal lobe epilepsy and may be useful in clinical practice.

Machine learning for filtering out false positive grey matter atrophies in single subject voxel based morphometry: A simulation based study

Single subject VBM (SS-VBM), has been used as an alternative tool to standard VBM for single case studies. However, it has the disadvantage of producing an excessively large number of false positive detections. In this study we propose a machine learning technique widely used for automated data classification, namely Support Vector Machine (SVM), to refine the findings produced by SS-VBM. A controlled set of experiments was conducted to evaluate the proposed approach using three-dimensional T1 MRI scans from control subjects collected from the publicly available IXI dataset. The scans were artificially atrophied at different locations and with different sizes to mimic the behavior of neurological disorders. Results empirically demonstrated that the proposed method is able to significantly reduce the amount of false positive clusters (p < 0.05), with no statistical differences in the true positive findings (p > 0.05). This evidence was observed to be consistent for different atrophied areas and sizes of atrophies. This approach could be potentially be applied to alleviate the intensive manual analysis that radiologists and clinicians have to perform to filter out miss-detections of SS-VBM, increasing its usability for image reading.

Neuroharmony: A new tool for harmonizing volumetric MRI data from unseen scanners

•We present Neuroharmony, a harmonization tool for images from unseen scanners. •We developed Neuroharmony using a total of 15,026 sMRI images. •The tool was able to reduce scanner-related bias from unseen scans. •Neuroharmony represents a significant step towards imaging-based clinical tools. •Neuroharmony is available at https://github.com/garciadias/Neuroharmony .

Translating research findings into clinical practice: a systematic and critical review of neuroimaging-based clinical tools for brain disorders

A pivotal aim of psychiatric and neurological research is to promote the translation of the findings into clinical practice to improve diagnostic and prognostic assessment of individual patients. Structural neuroimaging holds much promise, with neuroanatomical measures accounting for up to 40% of the variance in clinical outcome. Building on these findings, a number of imaging-based clinical tools have been developed to make diagnostic and prognostic inferences about individual patients from their structural Magnetic Resonance Imaging scans. This systematic review describes and compares the technical characteristics of the available tools, with the aim to assess their translational potential into real-world clinical settings. The results reveal that a total of eight tools. All of these were specifically developed for neurological disorders, and as such are not suitable for application to psychiatric disorders. Furthermore, most of the tools were trained and validated in a single dataset, which can result in poor generalizability, or using a small number of individuals, which can cause overoptimistic results. In addition, all of the tools rely on two strategies to detect brain abnormalities in single individuals, one based on univariate comparison, and the other based on multivariate machine-learning algorithms. We discuss current barriers to the adoption of these tools in clinical practice and propose a checklist of pivotal characteristics that should be included in an "ideal" neuroimaging-based clinical tool for brain disorders.

Neuroanatomy of dyslexia: An allometric approach

Despite evidence for a difference in total brain volume between dyslexic and good readers, no previous neuroimaging study examined differences in allometric scaling (i.e. differences in the relationship between regional and total brain volumes) between dyslexic and good readers. The present study aims to fill this gap by testing differences in allometric scaling and regional brain volume differences in dyslexic and good readers. Object-based morphometry analysis was used to determine grey and white matter volumes of the four lobes, the cerebellum and limbic structures in 130 dyslexic and 106 good readers aged 8-14 years. Data were collected across three countries (France, Poland and Germany). Three methodological approaches were used as follows: principal component analysis (PCA), linear regression and multiple-group confirmatory factor analysis (MGCFA). Difference in total brain volume between good and dyslexic readers was Cohen's d = 0.39. We found no difference in allometric scaling, nor in regional brain volume between dyslexic and good readers. Results of our three methodological approaches (PCA, linear regression and MGCFA) were consistent. This study provides evidence for total brain volume differences between dyslexic and control children, but no evidence for differences in the volumes of the four lobes, the cerebellum or limbic structures, once allometry is taken into account. It also finds no evidence for a difference in allometric relationships between the groups. We highlight the methodological interest of the MGCFA approach to investigate such research issues.

Quantitative brain relaxation atlases for personalized detection and characterization of brain pathology

PURPOSE

To exploit the improved comparability and hardware independency of quantitative MRI, databases of MR physical parameters in healthy tissue are required, to which tissue properties of patients can be compared. In this work, normative values for longitudinal and transverse relaxation times in the brain were established and tested in single-subject comparisons for detection of abnormal relaxation times.

METHODS

Relaxometry maps of the brain were acquired from 52 healthy volunteers. After spatially normalizing the volumes into a common space, T₁ and T₂ inter-subject variability within the healthy cohort was modeled voxel-wise. A method for a single-subject comparison against the atlases was developed by computing z-scores with respect to the established healthy norms. The comparison was applied to two multiple sclerosis and one clinically isolated syndrome cases for a proof of concept.

RESULTS

The established atlases exhibit a low variation in white matter structures (median RMSE of models equal to 32 ms for T₁ and 4 ms for T₂ ), indicating that relaxation times are in a narrow range for normal tissues. The proposed method for single-subject comparison detected relaxation time deviations from healthy norms in the example patient data sets. Relaxation times were found to be increased in brain lesions (mean z-scores >5). Moreover, subtle and confluent differences (z-scores ~2-4) were observed in clinically plausible regions (between lesions, corpus callosum).

CONCLUSIONS

Brain T₁ and T₂ quantitative norms were derived voxel-wise with low variability in healthy tissue. Example patient deviation maps demonstrated good sensitivity of the atlases for detecting relaxation time alterations.

Personalized pathology maps to quantify diffuse and focal brain damage

Background and objectives

Quantitative MRI (qMRI) permits the quantification of brain changes compatible with inflammation, degeneration and repair in multiple sclerosis (MS) patients. In this study, we propose a new method to provide personalized maps of tissue alterations and longitudinal brain changes based on different qMRI metrics, which provide complementary information about brain pathology.

Methods

We performed baseline and two-years follow-up on (i) 13 relapsing-remitting MS patients and (ii) four healthy controls. A group consisting of up to 65 healthy controls was used to compute the reference distribution of qMRI metrics in healthy tissue. All subjects underwent 3T MRI examinations including T1, T2, T2* relaxation and Magnetization Transfer Ratio (MTR) imaging. We used a recent partial volume estimation algorithm to estimate the concentration of different brain tissue types on T1 maps; then, we computed a deviation map (z-score map) for each contrast at both time-points. Finally, we subtracted those deviation maps only for voxels showing a significant difference with healthy tissue in one of the time points, to obtain a difference map for each subject.

Results and conclusion

Control subjects did not show any significant z-score deviations or longitudinal z-score changes. On the other hand, MS patients showed brain regions with cross-sectional and longitudinal concomitant increase in T1, T2, T2* z-scores and decrease of MTR z-scores, suggesting brain tissue degeneration/loss. In the lesion periphery, we observed areas with cross-sectional and longitudinal decreased T1/T2 and slight decrease in T2* most likely related to iron accumulation. Moreover, we measured longitudinal decrease in T1, T2 - and to a lesser extent in T2* - as well as a concomitant increase in MTR, suggesting remyelination/repair.

In summary, we have developed a method that provides whole-brain personalized maps of cross-sectional and longitudinal changes in MS patients, which are computed in patient space. These maps may open new perspectives to complement and support radiological evaluation of brain damage for a given patient.

Abnormal subcortical activity in congenital mirror movement disorder with RAD51 mutation

PURPOSE

Congenital mirror movement disorder (CMMD) is characterized by unintended, nonsuppressible, homologous mirroring activity contralateral to the movement on the intended side of the body. In healthy controls, unilateral movements are accompanied with predominantly contralateral cortical activity, whereas in CMMD, in line with the abnormal behavior, bilateral cortical activity is observed for unilateral motor tasks. However, task-related activities in subcortical structures, which are known to play critical roles in motor actions, have not been investigated in CMMD previously.

METHODS

We investigated the functional activation patterns of the motor components in CMMD patients. By using linkage analysis and exome sequencing, common mutations were revealed in seven affected individuals from the same family. Next, using functional magnetic resonance imaging (fMRI) we investigated cortical and subcortical activity during manual motor actions in two right-handed affected brothers and sex, age, education, and socioeconomically matched healthy individuals.

RESULTS

Genetic analyses revealed heterozygous RAD51 c.401C>T mutation which cosegregated with the phenotype in two affected members of the family. Consistent with previous literature, our fMRI results on these two affected individuals showed that mirror movements were closely related to abnormal cortical activity in M1 and SMA during unimanual movements. Furthermore, we have found previously unknown abnormal task-related activity in subcortical structures. Specifically, we have found increased and bilateral activity during unimanual movements in thalamus, striatum, and globus pallidus in CMMD patients.

CONCLUSION

These findings reveal further neural correlates of CMMD, and may guide our understanding of the critical roles of subcortical structures for unimanual movements in healthy individuals.

The charm of structural neuroimaging in insanity evaluations: guidelines to avoid misinterpretation of the findings

Despite the popularity of structural neuroimaging techniques in twenty-first-century research, its results have had limited translational impact in real-world settings, where inferences need to be made at the individual level. Structural neuroimaging methods are now introduced frequently to aid in assessing defendants for insanity in criminal forensic evaluations, with the aim of providing "convergence" of evidence on the mens rea of the defendant. This approach may provide pivotal support for judges' decisions. Although neuroimaging aims to reduce uncertainty and controversies in legal settings and to increase the objectivity of criminal rulings, the application of structural neuroimaging in forensic settings is hampered by cognitive biases in the evaluation of evidence that lead to misinterpretation of the imaging results. It is thus increasingly important to have clear guidelines on the correct ways to apply and interpret neuroimaging evidence. In the current paper, we review the literature concerning structural neuroimaging in court settings with the aim of identifying rules for its correct application and interpretation. These rules, which aim to decrease the risk of biases, focus on the importance of (i) descriptive diagnoses, (ii) anatomo-clinical correlation, (iii) brain plasticity and (iv) avoiding logical fallacies, such as reverse inference. In addition, through the analysis of real forensic cases, we describe errors frequently observed due to incorrect interpretations of imaging. Clear guidelines for both the correct circumstances for introducing neuroimaging and its eventual interpretation are defined.

BrainMap VBM: An environment for structural meta-analysis

The BrainMap database is a community resource that curates peer-reviewed, coordinate-based human neuroimaging literature. By pairing the results of neuroimaging studies with their relevant meta-data, BrainMap facilitates coordinate-based meta-analysis (CBMA) of the neuroimaging literature en masse or at the level of experimental paradigm, clinical disease, or anatomic location. Initially dedicated to the functional, task-activation literature, BrainMap is now expanding to include voxel-based morphometry (VBM) studies in a separate sector, titled: BrainMap VBM. VBM is a whole-brain, voxel-wise method that measures significant structural differences between or within groups which are reported as standardized, peak x-y-z coordinates. Here we describe BrainMap VBM, including the meta-data structure, current data volume, and automated reverse inference functions (region-to-disease profile) of this new community resource. CBMA offers a robust methodology for retaining true-positive and excluding false-positive findings across studies in the VBM literature. As with BrainMap's functional database, BrainMap VBM may be synthesized en masse or at the level of clinical disease or anatomic location. As a use-case scenario for BrainMap VBM, we illustrate a trans-diagnostic data-mining procedure wherein we explore the underlying network structure of 2,002 experiments representing over 53,000 subjects through independent components analysis (ICA). To reduce data-redundancy effects inherent to any database, we demonstrate two data-filtering approaches that proved helpful to ICA. Finally, we apply hierarchical clustering analysis (HCA) to measure network- and disease-specificity. This procedure distinguished psychiatric from neurological diseases. We invite the neuroscientific community to further exploit BrainMap VBM with other modeling approaches.

False positive rates in surface-based anatomical analysis

The false positive rates (FPR) for surface-based group analysis of cortical thickness, surface area, and volume were evaluated for parametric and non-parametric clusterwise correction for multiple comparisons for a range of smoothing levels and cluster-forming thresholds (CFT) using real data under group assignments that should not yield significant results. For whole cortical surface analysis, thickness showed modest inflation in parametric FPRs above the nominal level (10% versus 5%). Surface area and volume FPRs were much higher (20-30%). In the analysis of interhemispheric thickness asymmetries, FPRs were well controlled by parametric correction, but FPRs for surface area and volume asymmetries were still inflated. In all cases, non-parametric permutation adequately controlled the FPRs. It was found that inflated parametric FPRs were caused by violations in the parametric assumptions, namely a heavier-than-Gaussian spatial correlation. The non-Gaussian spatial correlation originates from anatomical features unique to individuals (e.g., a patch of cortex slightly thicker or thinner than average) and is not a by-product of scanning or processing. Thickness performed better than surface area and volume because thickness does not require a Jacobian correction.

Common and distinct patterns of grey-matter volume alteration in major depression and bipolar disorder: evidence from voxel-based meta-analysis

Finding robust brain substrates of mood disorders is an important target for research. The degree to which major depression (MDD) and bipolar disorder (BD) are associated with common and/or distinct patterns of volumetric changes is nevertheless unclear. Furthermore, the extant literature is heterogeneous with respect to the nature of these changes. We report a meta-analysis of voxel-based morphometry (VBM) studies in MDD and BD. We identified studies published up to January 2015 that compared grey matter in MDD (50 data sets including 4101 individuals) and BD (36 data sets including 2407 individuals) using whole-brain VBM. We used statistical maps from the studies included where available and reported peak coordinates otherwise. Group comparisons and conjunction analyses identified regions in which the disorders showed common and distinct patterns of volumetric alteration. Both disorders were associated with lower grey-matter volume relative to healthy individuals in a number of areas. Conjunction analysis showed smaller volumes in both disorders in clusters in the dorsomedial and ventromedial prefrontal cortex, including the anterior cingulate cortex and bilateral insula. Group comparisons indicated that findings of smaller grey-matter volumes relative to controls in the right dorsolateral prefrontal cortex and left hippocampus, along with cerebellar, temporal and parietal regions were more substantial in major depression. These results suggest that MDD and BD are characterised by both common and distinct patterns of grey-matter volume changes. This combination of differences and similarities has the potential to inform the development of diagnostic biomarkers for these conditions.

Volume gain of periaqueductal gray in medication-overuse headache

BACKGROUND

Periaqueductal gray (PAG) is a substantial descending pain modulatory center, and previous voxel-based morphometry study confirmed the clusters with increased volume in PAG region in medication-overuse headache (MOH). The aim of this study is to investigate altered PAG volume in MOH using an automated PAG segment method to measure the true PAG volume.

METHODS

High resolution three-dimensional T1-weighted fast spoiled gradient recalled echo MR images were obtained from 22 patients with MOH and 22 normal controls (NC). PAG template was created based on ICBM 152 gray template, and the individual PAG was generated by applying the deformation field from structural image segment to the PAG template, and individual PAG volume was calculated.

RESULTS

There was a significant increased volume of PAG in MOH (0.366 ± 0.005 ml) than that in NC (0.341 ± 0.005 ml)(P < 0.05). There was no significant correlation between the PAG volume and the clinical variables in MOH patients (P > 0.05). The area of receiver operating characteristic (ROC) curve was 0.845, and the cut-off of PAG volume was 0.341 ml with sensitivity 95.5% and specificity 63.6%.

CONCLUSION

The present study demonstrated that the PAG volume gain was confirmed in MOH patients, and the automated individual PAG volume measure may be considered as a simple and effective imaging biomarker in MOH diagnosis.

Imaging the neuroplastic effects of ketamine with VBM and the necessity of placebo control

In the last years a plethora of studies have investigated morphological changes induced by behavioural or pharmacological interventions using structural T1-weighted MRI and voxel-based morphometry (VBM). Ketamine is thought to exert its antidepressant action by restoring neuroplasticity. In order to test for acute impact of a single ketamine infusion on grey matter volume we performed a placebo-controlled, double-blind investigation in healthy volunteers using VBM. 28 healthy individuals underwent two MRI sessions within a timeframe of 2 weeks, each consisting of two structural T1-weighted MRIs within a single session, one before and one 45min after infusion of S-ketamine (bolus of 0.11mg/kg, followed by an maintenance infusion of 0.12mg/kg) or placebo (0.9% NaCl infusion) using a crossover design. In the repeated-measures ANOVA with time (post-infusion/pre-infusion) and medication (placebo/ketamine) as factors, no significant effect of interaction and no effect of medication was found (FWE-corrected). Importantly, further post-hoc t-tests revealed a strong "decrease" of grey matter both in the placebo and the ketamine condition over time. This effect was evident mainly in frontal and temporal regions bilaterally with t-values ranging from 4.95 to 5.31 (FWE-corrected at p<0.05 voxel level). The vulnerabilities of VBM have been repeatedly demonstrated, with reports of influence of blood flow, tissue water and direct effects of pharmacological compounds on the MRI signal. Here again, we highlight that the relationship between intervention and VBM results is apparently subject to a number of physiological influences, which are partly unknown. Future studies focusing on the effects of ketamine on grey matter should try to integrate known influential factors such as blood flow into analysis. Furthermore, the results of this study highlight the importance of a carefully performed placebo condition in pharmacological fMRI studies.

Cluster failure: Why fMRI inferences for spatial extent have inflated false-positive rates

The most widely used task functional magnetic resonance imaging (fMRI) analyses use parametric statistical methods that depend on a variety of assumptions. In this work, we use real resting-state data and a total of 3 million random task group analyses to compute empirical familywise error rates for the fMRI software packages SPM, FSL, and AFNI, as well as a nonparametric permutation method. For a nominal familywise error rate of 5%, the parametric statistical methods are shown to be conservative for voxelwise inference and invalid for clusterwise inference. Our results suggest that the principal cause of the invalid cluster inferences is spatial autocorrelation functions that do not follow the assumed Gaussian shape. By comparison, the nonparametric permutation test is found to produce nominal results for voxelwise as well as clusterwise inference. These findings speak to the need of validating the statistical methods being used in the field of neuroimaging.

An a contrario approach for the detection of patient-specific brain perfusion abnormalities with arterial spin labelling

In this paper, we introduce a new locally multivariate procedure to quantitatively extract voxel-wise patterns of abnormal perfusion in individual patients. This a contrario approach uses a multivariate metric from the computer vision community that is suitable to detect abnormalities even in the presence of closeby hypo- and hyper-perfusions. This method takes into account local information without applying Gaussian smoothing to the data. Furthermore, to improve on the standard a contrario approach, which assumes white noise, we introduce an updated a contrario approach that takes into account the spatial coherency of the noise in the probability estimation. Validation is undertaken on a dataset of 25 patients diagnosed with brain tumours and 61 healthy volunteers. We show how the a contrario approach outperforms the massively univariate general linear model usually employed for this type of analysis.

When the Single Matters more than the Group (II): Addressing the Problem of High False Positive Rates in Single Case Voxel Based Morphometry Using Non-parametric Statistics

In recent years, an increasing number of studies have used Voxel Based Morphometry (VBM) to compare a single patient with a psychiatric or neurological condition of interest against a group of healthy controls. However, the validity of this approach critically relies on the assumption that the single patient is drawn from a hypothetical population with a normal distribution and variance equal to that of the control group. In a previous investigation, we demonstrated that family-wise false positive error rate (i.e., the proportion of statistical comparisons yielding at least one false positive) in single case VBM are much higher than expected (Scarpazza et al., 2013). Here, we examine whether the use of non-parametric statistics, which does not rely on the assumptions of normal distribution and equal variance, would enable the investigation of single subjects with good control of false positive risk. We empirically estimated false positive rates (FPRs) in single case non-parametric VBM, by performing 400 statistical comparisons between a single disease-free individual and a group of 100 disease-free controls. The impact of smoothing (4, 8, and 12 mm) and type of pre-processing (Modulated, Unmodulated) was also examined, as these factors have been found to influence FPRs in previous investigations using parametric statistics. The 400 statistical comparisons were repeated using two independent, freely available data sets in order to maximize the generalizability of the results. We found that the family-wise error rate was 5% for increases and 3.6% for decreases in one data set; and 5.6% for increases and 6.3% for decreases in the other data set (5% nominal). Further, these results were not dependent on the level of smoothing and modulation. Therefore, the present study provides empirical evidence that single case VBM studies with non-parametric statistics are not susceptible to high false positive rates. The critical implication of this finding is that VBM can be used to characterize neuroanatomical alterations in individual subjects as long as non-parametric statistics are employed.

Investigating Neuroanatomical Features in Top Athletes at the Single Subject Level

In sport events like Olympic Games or World Championships competitive athletes keep pushing the boundaries of human performance. Compared to team sports, high achievements in many athletic disciplines depend solely on the individual's performance. Contrasting previous research looking for expertise-related differences in brain anatomy at the group level, we aim to demonstrate changes in individual top athlete's brain, which would be averaged out in a group analysis. We compared structural magnetic resonance images (MRI) of three professional track-and-field athletes to age-, gender- and education-matched control subjects. To determine brain features specific to these top athletes, we tested for significant deviations in structural grey matter density between each of the three top athletes and a carefully matched control sample. While total brain volumes were comparable between athletes and controls, we show regional grey matter differences in striatum and thalamus. The demonstrated brain anatomy patterns remained stable and were detected after 2 years with Olympic Games in between. We also found differences in the fusiform gyrus in two top long jumpers. We interpret our findings in reward-related areas as correlates of top athletes' persistency to reach top-level skill performance over years.

False positive rates in Voxel-based Morphometry studies of the human brain: should we be worried?

Voxel-based Morphometry (VBM) is a widely used automated technique for the analysis of neuroanatomical images. Despite its popularity within the neuroimaging community, there are outstanding concerns about its potential susceptibility to false positive findings. Here we review the main methodological factors that are known to influence the results of VBM studies comparing two groups of subjects. We then use two large, open-access data sets to empirically estimate false positive rates and how these depend on sample size, degree of smoothing and modulation. Our review and investigation provide three main results: (i) when groups of equal size are compared false positive rate is not higher than expected, i.e. about 5%; (ii) the sample size, degree of smoothing and modulation do not appear to influence false positive rate; (iii) when they exist, false positive findings are randomly distributed across the brain. These results provide reassurance that VBM studies comparing groups are not vulnerable to the higher than expected false positive rates that are evident in single case VBM.

Neural correlates of cognitive intervention in persons at risk of developing Alzheimer's disease

Cognitive training is an emergent approach that has begun to receive increased attention in recent years as a non-pharmacological, cost-effective intervention for Alzheimer’s disease (AD). There has been increasing behavioral evidence regarding training-related improvement in cognitive performance in early stages of AD. Although these studies provide important insight about the efficacy of cognitive training, neuroimaging studies are crucial to pinpoint changes in brain structure and function associated with training and to examine their overlap with pathology in AD. In this study, we reviewed the existing neuroimaging studies on cognitive training in persons at risk of developing AD to provide an overview of the overlap between neural networks rehabilitated by the current training methods and those affected in AD. The data suggest a consistent training-related increase in brain activity in medial temporal, prefrontal, and posterior default mode networks, as well as increase in gray matter structure in frontoparietal and entorhinal regions. This pattern differs from the observed pattern in healthy older adults that shows a combination of increased and decreased activity in response to training. Detailed investigation of the data suggests that training in persons at risk of developing AD mainly improves compensatory mechanisms and partly restores the affected functions. While current neuroimaging studies are quite helpful in identifying the mechanisms underlying cognitive training, the data calls for future multi-modal neuroimaging studies with focus on multi-domain cognitive training, network level connectivity, and individual differences in response to training.

Increasing the reliability of data analysis of functional magnetic resonance imaging by applying a new blockwise permutation method

A recent paper by Eklund et al. (2012) showed that up to 70% false positive results may occur when analyzing functional magnetic resonance imaging (fMRI) data using the statistical parametric mapping (SPM) software, which may mainly be caused by insufficient compensation for the temporal correlation between successive scans. Here, we show that a blockwise permutation method can be an effective alternative to the standard correction method for the correlated residuals in the general linear model, assuming an AR(1)-model as used in SPM for analyzing fMRI data. The blockwise permutation approach including a random shift developed by our group (Adolf et al., 2011) accounts for the temporal correlation structure of the data without having to provide a specific definition of the underlying autocorrelation model. 1465 publicly accessible resting-state data sets were re-analyzed, and the results were compared with those of Eklund et al. (2012). It was found that with the new permutation method the nominal familywise error rate for the detection of activated voxels could be maintained approximately under even the most critical conditions in which Eklund et al. found the largest deviations from the nominal error level. Thus, the method presented here can serve as a tool to ameliorate the quality and reliability of fMRI data analyses.

Multidimensional morphometric 3D MRI analyses for detecting brain abnormalities in children: impact of control population

Automated morphometric approaches are used to detect epileptogenic structural abnormalities in 3D MR images in adults, using the variance of a control population to obtain z-score maps in an individual patient. Due to the substantial changes the developing human brain undergoes, performing such analyses in children is challenging. This study investigated six features derived from high-resolution T1 datasets in four groups: normal children (1.5T or 3T data), normal clinical scans (3T data), and patients with structural brain lesions (3T data), with each n = 10. Normative control data were obtained from the NIH study on normal brain development (n = 401). We show that control group size substantially influences the captured variance, directly impacting the patient's z-scores. Interestingly, matching on gender does not seem to be beneficial, which was unexpected. Using data obtained at higher field scanners produces slightly different base rates of suprathreshold voxels, as does using clinically derived normal studies, suggesting a subtle but systematic effect of both factors. Two approaches for controlling suprathreshold voxels in a multidimensional approach (combining features and requiring a minimum cluster size) were shown to be substantial and effective in reducing this number. Finally, specific strengths and limitations of such an approach could be demonstrated in individual cases.

Harnessing graphics processing units for improved neuroimaging statistics

Simple models and algorithms based on restrictive assumptions are often used in the field of neuroimaging for studies involving functional magnetic resonance imaging, voxel based morphometry, and diffusion tensor imaging. Nonparametric statistical methods or flexible Bayesian models can be applied rather easily to yield more trustworthy results. The spatial normalization step required for multisubject studies can also be improved by taking advantage of more robust algorithms for image registration. A common drawback of algorithms based on weaker assumptions, however, is the increase in computational complexity. In this short overview, we will therefore present some examples of how inexpensive PC graphics hardware, normally used for demanding computer games, can be used to enable practical use of more realistic models and accurate algorithms, such that the outcome of neuroimaging studies really can be trusted.

Neuroanatomical abnormalities in chronic tinnitus in the human brain

•

We review brain anatomical studies of tinnitus.

•

We evaluate the “gating mechanism” in light of the evidence from these studies.

•

We discuss the results and the possible causes of disparity between findings.

•

Overall, the evidence for structural abnormalities in tinnitus is unconvincing.

•

We identify methodological concerns and suggest strategies for future research.

In this paper, we review studies that have investigated brain morphology in chronic tinnitus in order to better understand the underlying pathophysiology of the disorder. Current consensus is that tinnitus is a disorder involving a distributed network of peripheral and central pathways in the nervous system. However, the precise mechanism remains elusive and it is unclear which structures are involved. Given that brain structure and function are highly related, identification of anatomical differences may shed light upon the mechanism of tinnitus generation and maintenance. We discuss anatomical changes in the auditory cortex, the limbic system, and prefrontal cortex, among others. Specifically, we discuss the gating mechanism of tinnitus and evaluate the evidence in support of the model from studies of brain anatomy. Although individual studies claim significant effects related to tinnitus, outcomes are divergent and even contradictory across studies. Moreover, results are often confounded by the presence of hearing loss. We conclude that, at present, the overall evidence for structural abnormalities specifically related to tinnitus is poor. As this area of research is expanding, we identify some key considerations for research design and propose strategies for future research.

Individualized Gaussian process-based prediction and detection of local and global gray matter abnormalities in elderly subjects

Structural imaging based on MRI is an integral component of the clinical assessment of patients with potential dementia. We here propose an individualized Gaussian process-based inference scheme for clinical decision support in healthy and pathological aging elderly subjects using MRI. The approach aims at quantitative and transparent support for clinicians who aim to detect structural abnormalities in patients at risk of Alzheimer's disease or other types of dementia. Firstly, we introduce a generative model incorporating our knowledge about normative decline of local and global gray matter volume across the brain in elderly. By supposing smooth structural trajectories the models account for the general course of age-related structural decline as well as late-life accelerated loss. Considering healthy subjects' demography and global brain parameters as informative about normal brain aging variability affords individualized predictions in single cases. Using Gaussian process models as a normative reference, we predict new subjects' brain scans and quantify the local gray matter abnormalities in terms of Normative Probability Maps (NPM) and global z-scores. By integrating the observed expectation error and the predictive uncertainty, the local maps and global scores exploit the advantages of Bayesian inference for clinical decisions and provide a valuable extension of diagnostic information about pathological aging. We validate the approach in simulated data and real MRI data. We train the GP framework using 1238 healthy subjects with ages 18–94 years, and predict in 415 independent test subjects diagnosed as healthy controls, Mild Cognitive Impairment and Alzheimer's disease.

•

We propose an approach to support individualized clinical decisions in elderlies.

•

Gaussian process models are used to build a normative generative model of aging.

•

It affords probabilistic predictions of local gray matter volume in subjects.

•

We validate the model using simulated and large real MRI data samples.

Evidence of reporting biases in voxel-based morphometry (VBM) studies of psychiatric and neurological disorders

OBJECTIVES

To evaluate whether biases may influence the findings of whole-brain structural imaging literature.

METHODS

Forty-seven whole-brain voxel-based meta-analyses including voxel-based morphometry (VBM) studies in neuropsychiatric conditions were included, for a total of 324 individual VBM studies. The total sample size, the overall number of foci, and different moderators were extracted both at the level of the individual studies and at the level of the meta-analyses.

RESULTS

Sample size ranged from 12 to 545 (median n = 47) per VBM study. The median number of reported foci per study was six. VBM studies with larger sample sizes reported only slightly more abnormalities than smaller studies (2% increase in the number of foci per 10-patients increase in sample size). A similar pattern was seen in several analyses according to different moderator variables with some possible modulating evidence for the statistical threshold employed, publication year and number of coauthors. Whole-brain meta-analyses (median sample size n = 534) found fewer foci (median = 3) than single studies and overall they showed no significant increase in the number of foci with increasing sample size. Meta-analyses with ≥10 VBM studies reported a median of three foci and showed a significant increase with increasing sample size, while there was no relationship between sample size and number of foci (median = 5) in meta-analyses with <10 VBM studies.

CONCLUSIONS

The number of foci reported in small VBM studies and even in meta-analyses with few studies may often be inflated. This picture is consistent with reporting biases affecting small studies.

Quantitative analysis of structural neuroimaging of mesial temporal lobe epilepsy

Mesial temporal lobe epilepsy (MTLE) is the most common of the surgically remediable drug-resistant epilepsies. MRI is the primary diagnostic tool to detect anatomical abnormalities and, when combined with EEG, can more accurately identify an epileptogenic lesion, which is often hippocampal sclerosis in cases of MTLE. As structural imaging technology has advanced the surgical treatment of MTLE and other lesional epilepsies, so too have the analysis techniques that are used to measure different structural attributes of the brain. These techniques, which are reviewed here and have been used chiefly in basic research of epilepsy and in studies of MTLE, have identified different types and the extent of anatomical abnormalities that can extend beyond the affected hippocampus. These results suggest that structural imaging and sophisticated imaging analysis could provide important information to identify networks capable of generating spontaneous seizures and ultimately help guide surgical therapy that improves postsurgical seizure-freedom outcomes.