Three-dimensional maximum probability atlas of the human brain, with particular reference to the temporal lobe

Multivariate classification provides a neural signature of Tourette disorder

BACKGROUND

Tourette disorder (TD), hallmarks of which are motor and vocal tics, has been related to functional abnormalities in large-scale brain networks. Using a fully data driven approach in a prospective, case-control study, we tested the hypothesis that functional connectivity of these networks carries a neural signature of TD. Our aim was to investigate (i) the brain networks that distinguish adult patients with TD from controls, and (ii) the effects of antipsychotic medication on these networks.

METHODS

Using a multivariate analysis based on support vector machine (SVM), we developed a predictive model of resting state functional connectivity in 48 patients and 51 controls, and identified brain networks that were most affected by disease and pharmacological treatments. We also performed standard univariate analyses to identify differences in specific connections across groups.

RESULTS

SVM was able to identify TD with 67% accuracy (p = 0.004), based on the connectivity in widespread networks involving the striatum, fronto-parietal cortical areas and the cerebellum. Medicated and unmedicated patients were discriminated with 69% accuracy (p = 0.019), based on the connectivity among striatum, insular and cerebellar networks. Univariate approaches revealed differences in functional connectivity within the striatum in patients v. controls, and between the caudate and insular cortex in medicated v. unmedicated TD.

CONCLUSIONS

SVM was able to identify a neuronal network that distinguishes patients with TD from control, as well as medicated and unmedicated patients with TD, holding a promise to identify imaging-based biomarkers of TD for clinical use and evaluation of the effects of treatment.

Human habit neural circuitry may be perturbed in eating disorders

Circuit-based mechanisms mediating the development and execution of habitual behaviors involve complex cortical-striatal interactions that have been investigated in animal models and more recently in humans. However, how human brain circuits implicated in habit formation may be perturbed in psychiatric disorders remains unclear. First, we identified the locations of the sensorimotor putamen and associative caudate in the human brain using probabilistic tractography from Human Connectome Project data. We found that multivariate connectivity of the sensorimotor putamen was altered in humans with binge eating disorder and bulimia nervosa and that the degree of alteration correlated with severity of disordered eating behavior. Furthermore, the extent of this circuit aberration correlated with mean diffusivity in the sensorimotor putamen and decreased basal dopamine D2/3 receptor binding potential in the striatum, consistent with previously reported microstructural changes and dopamine signaling mediating habit learning in animal models. Our findings suggest a neural circuit that links habit learning and binge eating behavior in humans, which could, in part, explain the treatment-resistant behavior common to eating disorders and other psychiatric conditions.

Case report: Multiple disconnection patterns revealed by a multi-modal analysis explained behavior after a focal frontal damage

Introduction

There is overwhelming evidence that focal lesions cause structural, metabolic, functional, and electrical disconnection of regions directly and indirectly connected with the site of injury. Unfortunately, methods to study disconnection (positron emission tomography, structural and functional magnetic resonance imaging, electroencephalography) have been applied primarily in isolation without capturing their interaction. Moreover, multi-modal imaging studies applied to focal lesions are rare.

Case report

We analyzed with a multi-modal approach the case of a patient presenting with borderline cognitive deficits across multiple domains and recurrent delirium. A post-surgical focal frontal lesion was evident based on the brain anatomical MRI. However, we were able to acquire also simultaneous MRI (structural and functional) and [18F]FDG using a hybrid PET/MRI scan along with EEG recordings. Despite the focality of the primary anatomical lesion, structural disconnection in the white matter bundles extended far beyond the lesion and showed a topographical match with the cortical glucose hypometabolism seen both locally and remotely, in posterior cortices. Similarly, a right frontal delta activity near/at the region of structural damage was associated with alterations of distant occipital alpha power. Moreover, functional MRI revealed even more widespread local and distant synchronization, involving also regions not affected by the structural/metabolic/electrical impairment.

Conclusion

Overall, this exemplary multi-modal case study illustrates how a focal brain lesion causes a multiplicity of disconnection and functional impairments that extend beyond the borders of the anatomical irrecoverable damage. These effects were relevant to explain patient’s behavior and may be potential targets of neuro-modulation strategies.

Image-derived input functions from dynamic 15O-water PET scans using penalised reconstruction

BACKGROUND

Quantitative positron emission tomography (PET) scans of the brain typically require arterial blood sampling but this is complicated and logistically challenging. One solution to remove the need for arterial blood sampling is the use of image-derived input functions (IDIFs). Obtaining accurate IDIFs, however, has proved to be challenging, mainly due to the limited resolution of PET. Here, we employ penalised reconstruction alongside iterative thresholding methods and simple partial volume correction methods to produce IDIFs from a single PET scan, and subsequently, compare these to blood-sampled input curves (BSIFs) as ground truth. Retrospectively we used data from sixteen subjects with two dynamic ¹⁵O-labelled water PET scans and continuous arterial blood sampling: one baseline scan and another post-administration of acetazolamide.

RESULTS

IDIFs and BSIFs agreed well in terms of the area under the curve of input curves when comparing peaks, tails and peak-to-tail ratios with R² values of 0.95, 0.70 and 0.76, respectively. Grey matter cerebral blood flow (CBF) values showed good agreement with an average difference between the BSIF and IDIF CBF values of 2% ± and a coefficient of variation (CoV) of 7.3%.

CONCLUSION

Our results show promising results that a robust IDIF can be produced for dynamic ¹⁵O-water PET scans using only the dynamic PET scan images with no need for a corresponding MRI or complex analytical techniques and thereby making routine clinical use of quantitative CBF measurements with ¹⁵O-water feasible.

Precision estimates of relative and absolute cerebral blood flow in Alzheimer's disease and cognitively normal individuals

Alzheimer's disease is characterized by regional reductions in cerebral blood flow (CBF). Although the gold standard for measuring CBF is [¹⁵O]H₂O PET, proxies of relative CBF, derived from the early distribution phase of amyloid and tau tracers, have gained attention. The present study assessed precision of [¹⁵O]H₂O derived relative and absolute CBF, and compared precision of these measures with that of (relative) CBF proxies. Dynamic [¹⁵O]H₂O, [¹⁸F]florbetapir and [¹⁸F]flortaucipir PET test-retest (TrT) datasets with eleven, nine and fourteen subjects, respectively, were included. Analyses were performed using an arterial input model and/or a simplified reference tissue model, depending on the data available. Relative CBF values (i.e. K₁/K₁' and/or R₁) were obtained using cerebellar cortex as reference tissue and TrT repeatability (i.e. precision) was calculated and compared between tracers, parameters and clinical groups. Relative CBF had significantly better TrT repeatability than absolute CBF derived from [¹⁵O]H₂O (r = -0.53), while best TrT repeatability was observed for [¹⁸F]florbetapir and [¹⁸F]flortaucipir R₁ (r = -0.23, r = -0.33). Furthermore, only R₁ showed, better TrT repeatability for cognitively normal individuals. High precision of CBF proxies could be due to a compensatory effect of the extraction fraction, although changes in extraction fraction could also bias these proxies, but not the gold standard.

Functional magnetic resonance imaging of headache: Issues, best-practices, and new directions, a narrative review

OBJECTIVE

To ensure readers are informed consumers of functional magnetic resonance imaging (fMRI) research in headache, to outline ongoing challenges in this area of research, and to describe potential considerations when asked to collaborate on fMRI research in headache, as well as to suggest future directions for improvement in the field.

BACKGROUND

Functional MRI has played a key role in understanding headache pathophysiology, and mapping networks involved with headache-related brain activity have the potential to identify intervention targets. Some investigators have also begun to explore its use for diagnosis.

METHODS/RESULTS

The manuscript is a narrative review of the current best practices in fMRI in headache research, including guidelines on transparency and reproducibility. It also contains an outline of the fundamentals of MRI theory, task-related study design, resting-state functional connectivity, relevant statistics and power analysis, image preprocessing, and other considerations essential to the field.

CONCLUSION

Best practices to increase reproducibility include methods transparency, eliminating error, using a priori hypotheses and power calculations, using standardized instruments and diagnostic criteria, and developing large-scale, publicly available datasets.

Exploring clinical outcomes in patients with idiopathic/inherited isolated generalized dystonia and stimulation of the subthalamic region

BACKGROUND

 Deep Brain Stimulation (DBS) is an established treatment option for refractory dystonia, but the improvement among the patients is variable.

OBJECTIVE

 To describe the outcomes of DBS of the subthalamic region (STN) in dystonic patients and to determine whether the volume of tissue activated (VTA) inside the STN or the structural connectivity between the area stimulated and different regions of the brain are associated with dystonia improvement.

METHODS

 The response to DBS was measured by the Burke-Fahn-Marsden Dystonia Rating Scale (BFM) before and 7 months after surgery in patients with generalized isolated dystonia of inherited/idiopathic etiology. The sum of the two overlapping STN volumes from both hemispheres was correlated with the change in BFM scores to assess whether the area stimulated inside the STN affects the clinical outcome. Structural connectivity estimates between the VTA (of each patient) and different brain regions were computed using a normative connectome taken from healthy subjects.

RESULTS

 Five patients were included. The baseline BFM motor and disability subscores were 78.30 ± 13.55 (62.00-98.00) and 20.60 ± 7.80 (13.00-32.00), respectively. Patients improved dystonic symptoms, though differently. No relationships were found between the VTA inside the STN and the BFM improvement after surgery (p = 0.463). However, the connectivity between the VTA and the cerebellum structurally correlated with dystonia improvement (p = 0.003).

CONCLUSIONS

 These data suggest that the volume of the stimulated STN does not explain the variance in outcomes in dystonia. Still, the connectivity pattern between the region stimulated and the cerebellum is linked to outcomes of patients.

Evaluation of eight registration algorithms applied to the insula and insular gyri

BACKGROUND AND PURPOSE

Spatial registration is crucial in establishing correspondence between anatomic brain regions for research and clinical purposes. The insular cortex (IC) and gyri (IG) are implicated in various functions and pathologies including epilepsy. Optimizing registration of the insula to a common atlas can improve the accuracy of group-level analyses. Here, we compared six nonlinear, one linear, and one semiautomated registration algorithms (RAs) for registering the IC and IG to the Montreal Neurologic Institute standard space (MNI152).

METHODS

3T images acquired from 20 controls and 20 temporal lobe epilepsy patients with mesial temporal sclerosis underwent automated segmentation of the insula. This was followed by manual segmentation of the entire IC and six individual IGs. Consensus segmentations were created at 75% agreement for IC and IG before undergoing registration to MNI152 space with eight RAs. Dice similarity coefficients (DSCs) were calculated between segmentations after registration and the IC and IG in MNI152 space. Statistical analysis involved the Kruskal-Wallace test with Dunn's test for IC and two-way analysis of variance with Tukey's honest significant difference test for IG.

RESULTS

DSCs were significantly different between RAs. Based on multiple pairwise comparisons, we report that certain RAs performed better than others across population groups. Additionally, registration performance differed according to specific IG.

CONCLUSION

We compared different methods for registering the IC and IG to MNI152 space. We found differences in performance between RAs, which suggests that algorithm choice is important factor in analyses involving the insula.

Impact of framing scheme optimization and smoking status on binding potential analysis in dynamic PET with [11C]ABP688

BACKGROUND

For positron emission tomography (PET) ligands, such as [¹¹C]ABP688, to be able to provide more evidence about the glutamatergic hypothesis in schizophrenia (SZ), quantification bias during dynamic PET studies and its propagation into the estimated values of non-displaceable binding potential (BP_ND) must be addressed. This would enable more accurate quantification during bolus + infusion (BI) neuroreceptor studies and further our understanding of neurological diseases. Previous studies have shown BP_ND-related biases can often occur due to overestimated cerebellum activity (reference region). This work investigates whether an alternative framing scheme can minimize quantification biases propagated into BP_ND, whether confounders, such as smoking status, need to be controlled for during the study, and what the consequences for the data interpretation following analysis are. A group of healthy controls (HC) and a group of SZ patients (balanced and unbalanced number of smokers) were investigated with [¹¹C]ABP688 and a BI protocol. Possible differences in BP_ND quantification as a function of smoking status were tested with constant 5 min ('Const 5 min') and constant true counts ('Const Trues') framing schemes. In order to find biomarkers for SZ, the differences in smoking effects were compared between groups. The normalized BP_ND and the balanced number of smokers and non-smokers for both framing schemes were evaluated.

RESULTS

When applying F-tests to the 'Const 5 min' framing scheme, effect sizes (η²p) and brain regions which showed significant effects fluctuated considerably with F = 50.106 ± 54.948 (9.389 to 112.607), P-values 0.005 to < 0.001 and η²p = 0.514 ± 0.282 (0.238 to 0.801). Conversely, when the 'Const Trues' framing scheme was applied, the results showed much smaller fluctuations with F = 78.038 ± 8.975 (86.450 to 68.590), P < 0.001 for all conditions and η²p = 0.730 ± 0.017 (0.742 to 0.710), and regions with significant effects were more robustly reproduced. Further, differences, which would indicate false positive identifications between HC and SZ groups in five brain regions when using the 'Const 5 min' framing scheme, were not observed with the 'Const Trues' framing.

CONCLUSIONS

Based on an [¹¹C]ABP688 PET study in SZ patients, the results show that non-consistent BP_ND outcomes can be propagated by the framing scheme and that potential bias can be minimized using 'Const Trues' framing.

Odor imagery but not perception drives risk for food cue reactivity and increased adiposity

Mental imagery has been proposed to play a critical role in the amplification of cravings. Here we tested whether olfactory imagery drives food cue reactivity strength to promote adiposity in 45 healthy individuals. We measured odor perception, odor imagery ability, and food cue reactivity using self-report, perceptual testing, and neuroimaging. Adiposity was assessed at baseline and one year later. Brain responses to real and imagined odors were analyzed with univariate and multivariate decoding methods to identify pattern-based olfactory codes. We found that the accuracy of decoding imagined, but not real, odor quality correlated with a perceptual measure of odor imagery ability and with greater adiposity changes. This latter relationship was mediated by cue-potentiated craving and intake. Collectively, these findings establish odor imagery ability as a risk factor for weight gain and more specifically as a mechanism by which exposure to food cues promotes craving and overeating.

Underlying differences in resting-state activity metrics related to sensitivity to punishment

Reinforcement sensitivity theory (RST) of personality establishes the punishment sensitivity trait as a source of variation in defensive avoidance/approach behaviors. These individual differences reflect dissimilar sensitivity and reactivity of the fight-flight-freeze and behavioral inhibition systems (FFFS/BIS). The sensitivity to punishment (SP) scale has been widely used in personality research aimed at studying the activity of these systems. Structural and functional neuroimaging studies have confirmed the core biological correlates of FFFS/BIS in humans. Nonetheless, some brain functional features derived from resting-state blood-oxygen level-dependent (BOLD) activity and its association with the punishment sensitivity dimension remain unclear. This relationship would shed light on stable neural activity patterns linked to anxiety-like behaviors and anxiety predisposition. In this study, we analyzed functional activity metrics "at rest" [e.g., regional homogeneity (ReHo) and fractional amplitude of low-frequency fluctuation (fALFF)] and their relationship with SP in key FFFS/BIS regions (e.g., amygdala, hippocampus, and periaqueductal gray) in a sample of 127 healthy adults. Our results revealed a significant negative correlation between the fALFF within all these regions and the scores on SP. Our findings suggest aberrant neural activity (lower fALFF) within the brain's defense system in participants with high trait anxiety, which in turn could reflect lower FFFS/BIS activation thresholds. These neurally-located differences could lead to pathological fear/anxiety behaviors arising from the FFFS and BIS.

Racial differences in white matter hyperintensity burden in older adults

White matter hyperintensities (WMHs) may be one of the earliest pathological changes in aging. Race differences in WMH burden has been conflicting. This study examined if race influences WMHs and whether these differences are influenced by vascular risk factors. Alzheimer's Disease Neuroimaging Initiative participants were included if they had a baseline MRI, diagnosis, and WMH measurements. Ninety-one Blacks and 1937 Whites were included. Using bootstrap re-sampling, 91 Whites were randomly sampled and matched to Blacks based on age, sex, education, and diagnosis 1000 times. Linear models examined the influence of race on baseline WMHs, and change of WMHs over time, with and without vascular factors. Vascular risk factors had higher prevalence in Blacks than Whites. When not including vascular factors, Blacks had greater frontal, parietal, deep, and total WMH burden compared to Whites. There were no race differences in longitudinal progression of WMH accumulation. After controlling for vascular factors, only overall longitudinal parietal WMH group differences remained significant, suggesting that vascular factors contribute to racial group differences observed in WMHs.

Hippocampal, basal ganglia and olfactory connectivity contribute to cognitive impairments in Parkinson's disease

Cognitive impairment is increasingly recognized as a characteristic feature of Parkinson's disease (PD), yet relatively little is known about its underlying neurobiology. Previous investigations suggest that dementia in PD is associated with subcortical atrophy, but similar studies in PD with mild cognitive impairment have been mixed. Variability in cognitive phenotypes and diversity of PD symptoms suggest that a common neuropathological origin results in a multitude of impacts within the brain. These direct and indirect impacts of disease pathology can be investigated using network analysis. Functional connectivity, for instance, may be more sensitive than atrophy to decline in specific cognitive domains in the PD population. Fifty-eight participants with PD underwent a neuropsychological test battery and scanning with structural and resting state functional MRI in a comprehensive whole-brain association analysis. To investigate atrophy as a potential marker of impairment, structural gray matter atrophy was associated with cognitive scores in each cognitive domain using voxel-based morphometry. To investigate connectivity, large-scale networks were correlated with voxel time series and associated with cognitive scores using distance covariance. Structural atrophy was not associated with any cognitive domain, with the exception of visuospatial measures in primary sensory and motor cortices. In contrast, functional connectivity was associated with attention, executive function, language, learning and memory, visuospatial, and global cognition in the bilateral hippocampus, left putamen, olfactory cortex, and bilateral anterior temporal poles. These preliminary results suggest that cognitive domain-specific networks in PD are distinct from each other and could provide a network signature for different cognitive phenotypes.

White matter hyperintensity load varies depending on subjective cognitive decline criteria

Increased age and cognitive impairment is associated with an increase in cerebrovascular pathology often measured as white matter hyperintensities (WMHs) on MRI. Whether WMH burden differs between cognitively unimpaired older adults with subjective cognitive decline (SCD +) and without subjective cognitive decline (SCD -) remains conflicting, and could be related to the methods used to identify SCD. Our goal was to examine if four common SCD classification methods are associated with different WMH accumulation patterns between SCD + and SCD - . A total of 535 cognitively unimpaired older adults with 1353 time points from the Alzheimer's Disease Neuroimaging Initiative were included in this study. SCD was operationalized using four different methods: Cognitive Change Index (CCI), Everyday Cognition Scale (ECog), ECog + Worry, and Worry. Linear mixed-effects models were used to investigate the associations between SCD and overall and regional WMH burden. Overall temporal WMH burden differences were only observed with the Worry questionnaire. Higher WMH burden change over time was observed in SCD + compared to SCD - in the temporal and parietal regions using the CCI (temporal, p = .01; parietal p = .02) and ECog (temporal, p = .02; parietal p = .01). For both the ECog + Worry and Worry questionnaire, change in WMH burden over time was increased in SCD + compared to SCD - for overall, frontal, temporal, and parietal WMH burden (p < .05). These results show that WMH burden differs between SCD + and SCD - depending on the questionnaire and the approach (regional/global) used to measure WMHs. The various methods used to define SCD may reflect different types of underlying pathologies.

Structural and metabolic correlates of neuropsychological profiles in multiple system atrophy and Parkinson's disease

BACKGROUND

Despite increased recognition of cognitive impairment in Multiple System Atrophy (MSA), its neuroanatomical correlates are not well defined. We aimed to explore cognitive profiles in MSA with predominant parkinsonism (MSA-P) and Parkinson's disease (PD) and their relationship to frontostriatal structural and metabolic changes.

METHODS

Detailed clinical and neuropsychological evaluation was performed together with diffusion tensor imaging (DTI) and [¹⁸F]-fluoro-deoxyglucose positron emission tomography ([¹⁸F]-FDG-PET) in patients with MSA-P (n = 11) and PD (n = 11). We compared clinical and neuropsychological data to healthy controls (n = 9) and correlated neuropsychological data with imaging findings in MSA-P and PD.

RESULTS

Patients with MSA-P showed deficits in executive function (Trail Making Test B-A) and scored higher in measures of depression and anxiety compared to those with PD and healthy controls. Widespread frontostriatal white matter tract reduction in fractional anisotropy was seen in MSA-P and PD compared to an imaging control group. Stroop Test interference performance correlated with [¹⁸F]-FDG uptake in the bilateral dorsolateral prefrontal cortex (DLPFC) and with white matter integrity between the striatum and left inferior frontal gyrus (IFG) in PD. Trail Making Test performance correlated with corticostriatal white matter integrity along tracts from the bilateral IFG in MSA-P and from the right DLPFC in both groups.

CONCLUSION

Executive dysfunction was more prominent in patients with MSA-P compared to PD. DLPFC metabolism and frontostriatal white matter integrity seem to be a driver of executive function in PD, whereas alterations in corticostriatal white matter integrity may contribute more to executive dysfunction in MSA-P.

Relationships between serotonin availability and frontolimbic response to fearful and threatening faces

Serotonin is a critical neurotransmitter in the regulation of emotional behavior. Although emotion processing is known to engage a corticolimbic circuit, including the amygdala and prefrontal cortex, exactly how this brain system is modulated by serotonin remains unclear. Here, we hypothesized that serotonin modulates variability in excitability and functional connectivity within this circuit. We tested whether this modulation contributes to inter-individual differences in emotion processing. Using a multimodal neuroimaging approach with a simultaneous PET-3T fMRI scanner, we simultaneously acquired BOLD signal while participants viewed emotional faces depicting fear and anger, while also measuring serotonin transporter (SERT) levels, regulating serotonin functions. Individuals with higher activity of the medial amygdala BOLD in response to fearful or angry facial expressions, who were temperamentally more anxious, also exhibited lower SERT availability in the dorsal raphe nucleus (DRN). Moreover, higher connectivity of the medial amygdala with the left dorsolateral prefrontal and the anterior cingulate cortex was associated with lower levels of SERT availability in the DRN. These results demonstrate the association between the serotonin transporter level and emotion processing through changes in functional interactions between the amygdala and the prefrontal areas in healthy humans.

[18F]F13640: a selective agonist PET radiopharmaceutical for imaging functional 5-HT1A receptors in humans

PURPOSE

F13640 (a.k.a. befiradol, NLX-112) is a highly selective 5-HT_1A receptor ligand that was selected as a PET radiopharmaceutical-candidate based on animal studies. Due to its high efficacy agonist properties, [¹⁸F]F13640 binds preferentially to functional 5-HT_1A receptors, which are coupled to intracellular G-proteins. Here, we characterize brain labeling of 5-HT_1A receptors by [¹⁸F]F13640 in humans and describe a simplified model for its quantification.

METHODS

PET/CT and PET-MRI scans were conducted in a total of 13 healthy male volunteers (29 ± 9 years old), with arterial input functions (AIF) (n = 9) and test-retest protocol (n = 8). Several kinetic models were compared (one tissue compartment model, two-tissue compartment model, and Logan); two models with reference region were also evaluated: simplified reference tissue model (SRTM) and the logan reference model (LREF).

RESULTS

[¹⁸F]F13640 showed high uptake values in raphe nuclei and cortical regions. SRTM and LREF models showed a very high correlation with kinetic models using AIF. As concerns test-retest parameters and the prolonged binding kinetics of [¹⁸F]F13640, better reproducibility, and reliability were found with the LREF method. Cerebellum white matter and frontal lobe white matter stand out as suitable reference regions.

CONCLUSION

The favorable brain labeling and kinetic profile of [¹⁸F]F13640, its high receptor specificity and its high efficacy agonist properties open new perspectives for studying functionally active 5-HT_1A receptors, unlike previous radiopharmaceuticals that act as antagonists. [¹⁸F]F13640's kinetic properties allow injection outside of the PET scanner with delayed acquisitions, facilitating the design of innovative longitudinal protocols in neurology and psychiatry.

TRIAL REGISTRATION

Trial Registration EudraCT 2017-002,722-21.

Cerebral blood flow, amyloid burden, and cognition in cognitively normal individuals

PURPOSE

The role of cerebral blood flow (CBF) in the early stages of Alzheimer's disease is complex and largely unknown. We investigated cross-sectional and longitudinal associations between CBF, amyloid burden, and cognition, in cognitively normal individuals with subjective cognitive decline (SCD).

METHODS

We included 187 cognitively normal individuals with SCD from the SCIENCe project (65 ± 8 years, 39% F, MMSE 29 ± 1). Each underwent a dynamic (0-70 min) [18F]florbetapir PET and T1-weighted MRI scan, enabling calculation of mean binding potential (BPND; specific amyloid binding) and R1 (measure of relative (r)CBF). Eighty-three individuals underwent a second [18F]florbetapir PET (2.6 ± 0.7 years). Participants annually underwent neuropsychological assessment (follow-up time 3.8 ± 3.1 years; number of observations n = 774).

RESULTS

A low baseline R1 was associated with steeper decline on tests addressing memory, attention, and global cognition (range betas 0.01 to 0.27, p < 0.05). High BPND was associated with steeper decline on tests covering all domains (range betas - 0.004 to - 0.70, p < 0.05). When both predictors were simultaneously added to the model, associations remained essentially unchanged. Additionally, we found longitudinal associations between R1 and BPND. High baseline BPND predicted decline over time in R1 (all regions, range betasBP×time - 0.09 to - 0.14, p < 0.05). Vice versa, low baseline R1 predicted increase in BPND in frontal, temporal, and composite ROIs over time (range betasR1×time - 0.03 to - 0.08, p < 0.05).

CONCLUSION

Our results suggest that amyloid accumulation and decrease in rCBF are two parallel disease processes without a fixed order, both providing unique predictive information for cognitive decline and each process enhancing the other longitudinally.

Cortical Amyloid Burden Relates to Basal Forebrain Volume in Subjective Cognitive Decline

BACKGROUND

Atrophy of cholinergic basal forebrain (BF) nuclei is a frequent finding in magnetic resonance imaging (MRI) volumetry studies that examined patients with prodromal or clinical Alzheimer's disease (AD), but less clear for individuals in earlier stages of the clinical AD continuum.

OBJECTIVE

To examine BF volume reductions in subjective cognitive decline (SCD) participants with AD pathologic changes.

METHODS

The present study compared MRI-based BF volume measurements in age- and sex-matched samples of N = 24 amyloid-positive and N = 24 amyloid-negative SCD individuals, based on binary visual ratings of Florbetaben positron emission tomography (PET) measurements. Additionally, we assessed associations of BF volume with cortical amyloid burden, based on semiquantitative Centiloid (CL) analyses.

RESULTS

Group differences approached significance for BF total volume (p = 0.061) and the Ch4 subregion (p = 0.059) only, showing the expected relative volume reductions for the amyloid-positive subgroup. There were also significant inverse correlations between BF volumes and CL values, which again were most robust for BF total volume and the Ch4 subregion.

CONCLUSIONS

The results are consistent with the hypothesis that amyloid-positive SCD individuals, which are considered to represent a transitional stage on the clinical AD continuum, already show incipient alterations of BF integrity. The negative association with a continuous measure of cortical amyloid burden also suggests that this may reflect an incremental process. Yet, further research is needed to evaluate whether BF changes already emerge at "grey zone" levels of amyloid accumulation, before amyloidosis is reliably detected by PET visual readings.

Performance scaling for structural MRI surface parcellations: a machine learning analysis in the ABCD Study

The use of predefined parcellations on surface-based representations of the brain as a method for data reduction is common across neuroimaging studies. In particular, prediction-based studies typically employ parcellation-driven summaries of brain measures as input to predictive algorithms, but the choice of parcellation and its influence on performance is often ignored. Here we employed preprocessed structural magnetic resonance imaging (sMRI) data from the Adolescent Brain Cognitive Development Study® to examine the relationship between 220 parcellations and out-of-sample predictive performance across 45 phenotypic measures in a large sample of 9- to 10-year-old children (N = 9,432). Choice of machine learning (ML) pipeline and use of alternative multiple parcellation-based strategies were also assessed. Relative parcellation performance was dependent on the spatial resolution of the parcellation, with larger number of parcels (up to ~4,000) outperforming coarser parcellations, according to a power-law scaling of between 1/4 and 1/3. Performance was further influenced by the type of parcellation, ML pipeline, and general strategy, with existing literature-based parcellations, a support vector-based pipeline, and ensembling across multiple parcellations, respectively, as the highest performing. These findings highlight the choice of parcellation as an important influence on downstream predictive performance, showing in some cases that switching to a higher resolution parcellation can yield a relatively large boost to performance.

Investigation of image-derived input functions for non-invasive quantification of myelin density using [11C]MeDAS PET

Multiple sclerosis (MS) is an inflammatory demyelinating disease. Current treatments are focussed on immune suppression to modulate pathogenic activity that causes myelin damage. New treatment strategies are needed to prevent demyelination and promote remyelination. Development of such myelin repair therapies require a sensitive and specific biomarker for efficacy evaluation. Recently, it has been shown that quantification of myelin density is possible using [¹¹C]MeDAS PET. This method, however, requires arterial blood sampling to generate an arterial input function (AIF). As the invasive nature of arterial sampling will reduce clinical applicability, the purpose of this study was to assess whether an image-derived input function (IDIF) can be used as an alternative way to facilitate its routine clinical use. Six healthy controls and 11 MS patients underwent MRI and [¹¹C]MeDAS PET with arterial blood sampling. The application of both population-based whole blood-to-plasma conversion and metabolite corrections were assessed for the AIF. Next, summed images of the early time frames (0-70 s) and the frame with the highest blood-brain contrast were used to generate IDIFs. IDIFs were created using either the hottest 2, 4, 6 or 12 voxels, or an isocontour of the hottest 10% voxels of the carotid artery. This was followed by blood-to-plasma conversion and metabolite correction of the IDIF. The application of a population-based metabolite correction of the AIF resulted in high correlations of tracer binding (K_i) within subjects, but variable bias across subjects. All IDIFs had a sharper and higher peak in the blood curves than the AIF, most likely due to dispersion during blood sampling. All IDIF methods resulted in similar high correlations within subjects (r = 0.95-0.98), but highly variable bias across subjects (mean slope=0.90-1.09). Therefore, both the use of population based blood-plasma and metabolite corrections and the generation of the image-derived whole-blood curve resulted in substantial bias in [¹¹C]MeDAS PET quantification, due to high inter-subject variability. Consequently, when unbiased quantification of [¹¹C]MeDAS PET data is required, individual AIF needs to be used.

Sample size requirement for achieving multisite harmonization using structural brain MRI features

When data is pooled across multiple sites, the extracted features are confounded by site effects. Harmonization methods attempt to correct these site effects while preserving the biological variability within the features. However, little is known about the sample size requirement for effectively learning the harmonization parameters and their relationship with the increasing number of sites. In this study, we performed experiments to find the minimum sample size required to achieve multisite harmonization (using neuroHarmonize) using volumetric and surface features by leveraging the concept of learning curves. Our first two experiments show that site-effects are effectively removed in a univariate and multivariate manner; however, it is essential to regress the effect of covariates from the harmonized data additionally. Our following two experiments with actual and simulated data showed that the minimum sample size required for achieving harmonization grows with the increasing average Mahalanobis distances between the sites and their reference distribution. We conclude by positing a general framework to understand the site effects using the Mahalanobis distance. Further, we provide insights on the various factors in a cross-validation design to achieve optimal inter-site harmonization.

Advanced Diagnostic Technique for Alzheimer's Disease using MRI Top-Ranked Volume and Surface-based Features

Background

Alzheimer's disease (AD) is the most dominant type of dementia that has not been treated completely yet. Few Alzheimer's patients are correctly diagnosed on time. Therefore, diagnostic tools are needed for better and more efficient diagnoses.

Objective

This study aimed to develop an efficient automated method to differentiate Alzheimer's patients from normal elderly and present the essential features with accurate Alzheimer's diagnosis.

Material and Methods

In this analytical study, 154 Magnetic Resonance Imaging (MRI) scans were obtained from the Alzheimer's Disease Neuroimaging Initiative (ADNI) database, preprocessed, and normalized by the head size for extracting features (volume, cortical thickness, Sulci depth, and Gyrification Index Features (GIF). Relief-F algorithm, t-test, and one way-ANOVA were used for feature ranking to obtain the most effective features representing the AD for the classification process. Finally, in the classification step, four classifiers were used with 10 folds cross-validation as follows: Gaussian Support Vector Machine (GSVM), Linear Support Vector Machine (LSVM), Weighted K-Nearest Neighbors (W-KNN), and Decision Tree algorithm.

Results

The LSVM classifier and W-KNN produce a testing accuracy of 100% with only seven features. Additionally, GSVM and decision tree produce a testing accuracy of 97.83% and 93.48%, respectively.

Conclusion

The proposed system represents an automatic and highly accurate AD detection with a few reliable and effective features and minimum time.

Reducing instability of inter-subject covariance of FDG uptake networks using structure-weighted sparse estimation approach

PURPOSE

Sparse inverse covariance estimation (SICE) is increasingly utilized to estimate inter-subject covariance of FDG uptake (FDGcov) as proxy of metabolic brain connectivity. However, this statistical method suffers from the lack of robustness in the connectivity estimation. Patterns of FDGcov were observed to be spatially similar with patterns of structural connectivity as obtained from DTI imaging. Based on this similarity, we propose to regularize the sparse estimation of FDGcov using the structural connectivity.

METHODS

We retrospectively analyzed the FDG-PET and DTI data of 26 healthy controls, 41 patients with Alzheimer's disease (AD), and 30 patients with frontotemporal lobar degeneration (FTLD). Structural connectivity matrix derived from DTI data was introduced as a regularization parameter to assign individual penalties to each potential metabolic connectivity. Leave-one-out cross validation experiments were performed to assess the differential diagnosis ability of structure weighted SICE approach. A few approaches of structure weighted were compared with the standard SICE.

RESULTS

Compared to the standard SICE, structural weighting has shown more stable performance in the supervised classification, especially in the differentiation AD vs. FTLD (accuracy of 89-90%, while unweighted SICE only 85%). There was a significant positive relationship between the minimum number of metabolic connection and the robustness of the classification accuracy (r = 0.57, P < 0.001). Shuffling experiments showed significant differences between classification score derived with true structural weighting and those obtained by randomized structure (P < 0.05).

CONCLUSION

The structure-weighted sparse estimation can enhance the robustness of metabolic connectivity, which may consequently improve the differentiation of pathological phenotypes.

Fully Automated, Semantic Segmentation of Whole-Body 18F-FDG PET/CT Images Based on Data-Centric Artificial Intelligence

We introduce multiple-organ objective segmentation (MOOSE) software that generates subject-specific, multiorgan segmentation using data-centric artificial intelligence principles to facilitate high-throughput systemic investigations of the human body via whole-body PET imaging. Methods: Image data from 2 PET/CT systems were used in training MOOSE. For noncerebral structures, 50 whole-body CT images were used, 30 of which were acquired from healthy controls (14 men and 16 women), and 20 datasets were acquired from oncology patients (14 men and 6 women). Noncerebral tissues consisted of 13 abdominal organs, 20 bone segments, subcutaneous fat, visceral fat, psoas muscle, and skeletal muscle. An expert panel manually segmented all noncerebral structures except for subcutaneous fat, visceral fat, and skeletal muscle, which were semiautomatically segmented using thresholding. A majority-voting algorithm was used to generate a reference-standard segmentation. From the 50 CT datasets, 40 were used for training and 10 for testing. For cerebral structures, 34 ¹⁸F-FDG PET/MRI brain image volumes were used from 10 healthy controls (5 men and 5 women imaged twice) and 14 nonlesional epilepsy patients (7 men and 7 women). Only ¹⁸F-FDG PET images were considered for training: 24 and 10 of 34 volumes were used for training and testing, respectively. The Dice score coefficient (DSC) was used as the primary metric, and the average symmetric surface distance as a secondary metric, to evaluate the automated segmentation performance. Results: An excellent overlap between the reference labels and MOOSE-derived organ segmentations was observed: 92% of noncerebral tissues showed DSCs of more than 0.90, whereas a few organs exhibited lower DSCs (e.g., adrenal glands [0.72], pancreas [0.85], and bladder [0.86]). The median DSCs of brain subregions derived from PET images were lower. Only 29% of the brain segments had a median DSC of more than 0.90, whereas segmentation of 60% of regions yielded a median DSC of 0.80-0.89. The results of the average symmetric surface distance analysis demonstrated that the average distance between the reference standard and the automatically segmented tissue surfaces (organs, bones, and brain regions) lies within the size of image voxels (2 mm). Conclusion: The proposed segmentation pipeline allows automatic segmentation of 120 unique tissues from whole-body ¹⁸F-FDG PET/CT images with high accuracy.

First-in-Human Evaluation of 18F-PF-06445974, a PET Radioligand That Preferentially Labels Phosphodiesterase-4B

Phosphodiesterase-4 (PDE4), which metabolizes the second messenger cyclic adenosine monophosphate (cAMP), has 4 isozymes: PDE4A, PDE4B, PDE4C, and PDE4D. PDE4B and PDE4D have the highest expression in the brain and may play a role in the pathophysiology and treatment of depression and dementia. This study evaluated the properties of the newly developed PDE4B-selective radioligand 18F-PF-06445974 in the brains of rodents, monkeys, and humans. Methods: Three monkeys and 5 healthy human volunteers underwent PET scans after intravenous injection of 18F-PF-06445974. Brain uptake was quantified as total distribution volume (V T) using the standard 2-tissue-compartment model and serial concentrations of parent radioligand in arterial plasma. Results: 18F-PF-06445974 readily distributed throughout monkey and human brain and had the highest binding in the thalamus. The value of V T was well identified by a 2-tissue-compartment model but increased by 10% during the terminal portions (40 and 60 min) of the monkey and human scans, respectively, consistent with radiometabolite accumulation in the brain. The average human V T values for the whole brain were 9.5 ± 2.4 mL ⋅ cm-3 Radiochromatographic analyses in knockout mice showed that 2 efflux transporters-permeability glycoprotein (P-gp) and breast cancer resistance protein (BCRP)-completely cleared the problematic radiometabolite but also partially cleared the parent radioligand from the brain. In vitro studies with the human transporters suggest that the parent radioligand was a partial substrate for BCRP and, to a lesser extent, for P-gp. Conclusion: 18F-PF-06445974 quantified PDE4B in the human brain with reasonable, but not complete, success. The gold standard compartmental method of analyzing brain and plasma data successfully identified the regional densities of PDE4B, which were widespread and highest in the thalamus, as expected. Because the radiometabolite-induced error was only about 10%, the radioligand is, in the opinion of the authors, suitable to extend to clinical studies.

A zero-dose synthetic baseline for the personalized analysis of 2-Deoxy-2-[18F]fluoroglucose: Application in Alzheimer’s disease

Purpose

Brain 2-Deoxy-2-[18F]fluoroglucose ([18F]FDG-PET) is widely used in the diagnostic workup of Alzheimer’s disease (AD). Current tools for uptake analysis rely on non-personalized templates, which poses a challenge as decreased glucose uptake could reflect neuronal dysfunction, or heterogeneous brain morphology associated with normal aging. Overcoming this, we propose a deep learning method for synthesizing a personalized [18F]FDG-PET baseline from the patient’s own MRI, and showcase its applicability in detecting AD pathology.

Methods

We included [18F]FDG-PET/MRI data from 123 patients of a local cohort and 600 patients from ADNI. A supervised, adversarial model with two connected Generative Adversarial Networks (GANs) was trained on cognitive normal (CN) patients with transfer-learning to generate full synthetic baseline volumes (sbPET) (192 × 192 × 192) which reflect healthy uptake conditioned on brain anatomy. Synthetic accuracy was measured by absolute relative %-difference (Abs%), relative %-difference (RD%), and peak signal-to-noise ratio (PSNR). Lastly, we deployed the sbPET images in a fully personalized method for localizing metabolic abnormalities.

Results

The model achieved a spatially uniform Abs% of 9.4%, RD% of 0.5%, and a PSNR of 26.3 for CN subjects. The sbPET images conformed to the anatomical information dictated by the MRI and proved robust in presence of atrophy. The personalized abnormality method correctly mapped the pathology of AD subjects while showing little to no anomalies for CN subjects.

Conclusion

This work demonstrated the feasibility of synthesizing fully personalized, healthy-appearing [18F]FDG-PET images. Using these, we showcased a promising application in diagnosing AD, and theorized the potential value of sbPET images in other neuroimaging routines.

3T sodium MR imaging in Alzheimer's disease shows stage-dependent sodium increase influenced by age and local brain volume

INTRODUCTION

Application of MRI in clinical routine mainly addresses structural alterations. However, pathological changes at a cellular level are expected to precede the occurrence of brain atrophy clusters and of clinical symptoms. In this context, ²³Na-MRI examines sodium changes in the brain as a potential metabolic parameter. Recently, we have shown that ²³Na-MRI at ultra-high-field (7 T) was able to detect increased tissue sodium concentration (TSC) in Alzheimer's disease (AD). In this work, we aimed at assessing AD-pathology with ²³Na-MRI in a larger cohort and on a clinical 3T MR scanner.

METHODS

We used a multimodal MRI protocol on 52 prodromal to mild AD patients and 34 cognitively healthy control subjects on a clinical 3T MR scanner. We examined the TSC, brain volume, and cortical thickness in association with clinical parameters. We further compared TSC with intra-individual normalized TSC for the reduction of inter-individual TSC variability resulting from physiological as well as experimental conditions. Normalized TSC maps were created by normalizing each voxel to the mean TSC inside the brain stem.

RESULTS

We found increased normalized TSC in the AD cohort compared to elderly control subjects both on global as well as on a region-of-interest-based level. We further confirmed a significant association of local brain volume as well as age with TSC. TSC increase in the left temporal lobe was further associated with the cognitive state, evaluated via the Montreal cognitive assessment (MoCA) screening test. An increase of normalized TSC depending on disease stage reflected by the Clinical Dementia Rating (CDR) was found in our AD patients in temporal lobe regions. In comparison to classical brain volume and cortical thickness assessments, normalized TSC had a higher discriminative power between controls and prodromal AD patients in several regions of the temporal lobe.

DISCUSSION

We confirm the feasibility of ²³Na-MRI at 3T and report an increase of TSC in AD in several regions of the brain, particularly in brain regions of the temporal lobe. Furthermore, to reduce inter-subject variability caused by physiological factors such as circadian rhythms and experimental conditions, we introduced normalized TSC maps. This showed a higher discriminative potential between different clinical groups in comparison to the classical TSC analysis. In conclusion, ²³Na-MRI represents a potential translational imaging marker applicable e.g.for diagnostics and the assessment of intervention outcomes in AD even under clinically available field strengths such as 3T. Implication of ²³Na-MRI in association with other metabolic imaging marker needs to be further elucidated.

Larger gray matter volumes in neuropsychiatric long-COVID syndrome

Neuropsychiatric symptoms are the most common sequelae of long-COVID. As accumulating evidence suggests an impact of survived SARS-CoV-2-infection on brain physiology, it is necessary to further investigate brain structural changes in relation to course and neuropsychiatric symptom burden in long-COVID. To this end, the present study investigated 3T-MRI scans from long-COVID patients suffering from neuropsychiatric symptoms (n = 30), and healthy controls (n = 20). Whole-brain comparison of gray matter volume (GMV) was conducted by voxel-based morphometry. To determine whether changes in GMV are predicted by neuropsychiatric symptom burden and/or initial severity of symptoms of COVID-19 and time since onset of COVID-19 stepwise linear regression analysis was performed. Significantly enlarged GMV in long-COVID patients was present in several clusters (spanning fronto-temporal areas, insula, hippocampus, amygdala, basal ganglia, and thalamus in both hemispheres) when compared to controls. Time since onset of COVID-19 was a significant regressor in four of these clusters with an inverse relationship. No associations with clinical symptom burden were found. GMV alterations in limbic and secondary olfactory areas are present in long-COVID patients and might be dynamic over time. Larger samples and longitudinal data in long-COVID patients are required to further clarify the mediating mechanisms between COVID-19, GMV and neuropsychiatric symptoms.

High-resolution atlasing and segmentation of the subcortex: Review and perspective on challenges and opportunities created by machine learning

This paper reviews almost three decades of work on atlasing and segmentation methods for subcortical structures in human brain MRI. In writing this survey, we have three distinct aims. First, to document the evolution of digital subcortical atlases of the human brain, from the early MRI templates published in the nineties, to the complex multi-modal atlases at the subregion level that are available today. Second, to provide a detailed record of related efforts in the automated segmentation front, from earlier atlas-based methods to modern machine learning approaches. And third, to present a perspective on the future of high-resolution atlasing and segmentation of subcortical structures in in vivo human brain MRI, including open challenges and opportunities created by recent developments in machine learning.

Serotonin and dopamine transporter availability in social anxiety disorder after combined treatment with escitalopram and cognitive-behavioral therapy

Selective serotonin reuptake inhibitors (SSRIs) and internet-based cognitive behavioral therapy (ICBT) are recommended treatments of social anxiety disorder (SAD), and often combined, but their effects on monoaminergic signaling are not well understood. In this multi-tracer positron emission tomography (PET) study, 24 patients with SAD were randomized to treatment with escitalopram+ICBT or placebo+ICBT under double-blind conditions. Before and after 9 weeks of treatment, patients were examined with positron emission tomography and the radioligands [¹¹C]DASB and [¹¹C]PE2I, probing the serotonin (SERT) and dopamine (DAT) transporter proteins respectively. Both treatment combinations resulted in significant improvement as measured by the Liebowitz Social Anxiety Scale (LSAS). At baseline, SERT-DAT co-expression was high and, in the putamen and thalamus, co-expression showed positive associations with symptom severity. SERT-DAT co-expression was also predictive of treatment success, but predictor-outcome associations differed in direction between the treatments. After treatment, average SERT occupancy in the SSRI + ICBT group was >80%, with positive associations between symptom improvement and occupancy in the nucleus accumbens, putamen and anterior cingulate cortex. Following placebo+ICBT, SERT binding increased in the raphe nuclei. DAT binding increased in both groups in limbic and striatal areas, but relations with symptom improvement differed, being negative for SSRI + ICBT and positive for placebo + ICBT. Thus, serotonin-dopamine transporter co-expression exerts influence on symptom severity and remission rate in the treatment of social anxiety disorder. However, the monoamine transporters are modulated in dissimilar ways when cognitive-behavioral treatment is given concomitantly with either SSRI-medication or pill placebo.

Estimating medical image registration error and confidence: A taxonomy and scoping review

Given that image registration is a fundamental and ubiquitous task in both clinical and research domains of the medical field, errors in registration can have serious consequences. Since such errors can mislead clinicians during image-guided therapies or bias the results of a downstream analysis, methods to estimate registration error are becoming more popular. To give structure to this new heterogenous field we developed a taxonomy and performed a scoping review of methods that quantitatively and automatically provide a dense estimation of registration error. The taxonomy breaks down error estimation methods into Approach (Image- or Transformation-based), Framework (Machine Learning or Direct) and Measurement (error or confidence) components. Following the PRISMA guidelines for scoping reviews, the 570 records found were reduced to twenty studies that met inclusion criteria, which were then reviewed according to the proposed taxonomy. Trends in the field, advantages and disadvantages of the methods, and potential sources of bias are also discussed. We provide suggestions for best practices and identify areas of future research.

Meta-analysis of molecular imaging of translocator protein in major depression

Molecular neuroimaging studies provide mounting evidence that neuroinflammation plays a contributory role in the pathogenesis of major depressive disorder (MDD). This has been the focus of a number of positron emission tomography (PET) studies of the 17-kDa translocator protein (TSPO), which is expressed by microglia and serves as a marker of neuroinflammation. In this meta-analysis, we compiled and analyzed all available molecular imaging studies comparing cerebral TSPO binding in MDD patients with healthy controls. Our systematic literature search yielded eight PET studies encompassing 238 MDD patients and 164 healthy subjects. The meta-analysis revealed relatively increased TSPO binding in several cortical regions (anterior cingulate cortex: Hedges' g = 0.6, 95% CI: 0.36, 0.84; hippocampus: g = 0.54, 95% CI: 0.26, 0.81; insula: g = 0.43, 95% CI: 0.17, 0.69; prefrontal cortex: g = 0.36, 95% CI: 0.14, 0.59; temporal cortex: g = 0.39, 95% CI: -0.04, 0.81). While the high range of effect size in the temporal cortex might reflect group-differences in body mass index (BMI), exploratory analyses failed to reveal any relationship between elevated TSPO availability in the other four brain regions and depression severity, age, BMI, radioligand, or the binding endpoint used, or with treatment status at the time of scanning. Taken together, this meta-analysis indicates a widespread ∼18% increase of TSPO availability in the brain of MDD patients, with effect sizes comparable to those in earlier molecular imaging studies of serotonin transporter availability and monoamine oxidase A binding.

White matter hyperintensity distribution differences in aging and neurodegenerative disease cohorts

INTRODUCTION

White matter hyperintensities (WMHs) are common magnetic resonance imaging (MRI) findings in the aging population in general, as well as in patients with neurodegenerative diseases. They are known to exacerbate the cognitive deficits and worsen the clinical outcomes in the patients. However, it is not well-understood whether there are disease-specific differences in prevalence and distribution of WMHs in different neurodegenerative disorders.

METHODS

Data included 976 participants with cross-sectional T1-weighted and fluid attenuated inversion recovery (FLAIR) MRIs from the Comprehensive Assessment of Neurodegeneration and Dementia (COMPASS-ND) cohort of the Canadian Consortium on Neurodegeneration in Aging (CCNA) with eleven distinct diagnostic groups: cognitively intact elderly (CIE), subjective cognitive impairment (SCI), mild cognitive impairment (MCI), vascular MCI (V-MCI), Alzheimer's dementia (AD), vascular AD (V-AD), frontotemporal dementia (FTD), Lewy body dementia (LBD), cognitively intact elderly with Parkinson's disease (PD-CIE), cognitively impaired Parkinson's disease (PD-CI), and mixed dementias. WMHs were segmented using a previously validated automated technique. WMH volumes in each lobe and hemisphere were compared against matched CIE individuals, as well as each other, and between men and women.

RESULTS

All cognitively impaired diagnostic groups had significantly greater overall WMH volumes than the CIE group. Vascular groups (i.e. V-MCI, V-AD, and mixed dementia) had significantly greater WMH volumes than all other groups, except for FTD, which also had significantly greater WMH volumes than all non-vascular groups. Women tended to have lower WMH burden than men in most groups and regions, controlling for age. The left frontal lobe tended to have a lower WMH burden than the right in all groups. In contrast, the right occipital lobe tended to have greater WMH volumes than the left.

CONCLUSIONS

There were distinct differences in WMH prevalence and distribution across diagnostic groups, sexes, and in terms of asymmetry. WMH burden was significantly greater in all neurodegenerative dementia groups, likely encompassing areas exclusively impacted by neurodegeneration as well as areas related to cerebrovascular disease pathology.

In vivo molecular imaging of the neuroinflammatory response to peripheral acute bacterial infection in older patients with cognitive dysfunction: A cross-sectional controlled study

Introduction

Chronic neuroinflammatory events have been implicated in the pathophysiology of neurodegenerative conditions but no studies have directly examined the neuroinflammatory response to acute systemic infection in older people with dementia. The objective of this study was to determine the magnitude of the neuroinflammatory response triggered by acute systemic infection in older subjects with dementia and/or delirium compared to cognitively healthy controls.

Methods

We recruited 19 participants (4 with delirium, 4 with dementia, 4 with delirium superimposed on dementia, 7 cognitively healthy) hospitalized with acute systemic bacterial infection not involving the Central Nervous System. Participants underwent [11C]-PK11195 PET and a neuropsychological assessment during hospital stay. The distribution volume ratio was estimated in the regions-of-interest using the Hammers’ brain atlas.

Results

In the subcortical analysis, we found that the cognitively healthy group presented regions with significantly higher DVR intensity than the other groups in the choroid plexus. Mean choroid plexus DVR positively correlated with MoCA (r = 0.66, p = 0.036).

Conclusion

This study suggests that dementia and/or delirium is associated with a reduced neuroinflammatory response to acute systemic bacterial infection which can be the result of an immunosuppressive brain environment.

Neuropsychological and Structural Neuroimaging Outcomes in LGI1-Limbic Encephalitis: A Case Study

OBJECTIVE

Anti-leucine-rich glioma-inactivated 1 limbic encephalitis (LGI1-LE) is a rare autoimmune condition that affects the structural integrity and functioning of the brain's limbic system. Little is known about its impact on long-term neuropsychological functioning and the structural integrity of the medial temporal lobe. Here we examined the long-term neuropsychological and neuroanatomical outcomes of a 68-year-old male who acquired LGI1-LE.

METHODS

Our case patient underwent standardized neuropsychological testing at two time points. Volumetric analyses of T1-weighted images were undertaken at four separate time points and qualitatively compared with a group of age-matched healthy controls.

RESULTS

At the time of initial assessment, our case study exhibited focal impairments in verbal and visual episodic memory and these impairments continued to persist after undergoing a course of immunotherapy. Furthermore, in reference to an age-matched healthy control group, over the course of 11 months, volumetric brain imaging analyses revealed that areas of the medial temporal lobe including specific hippocampal subfields (e.g., CA1 and dentate gyrus) underwent a subacute period of volumetric enlargement followed by a chronic period of volumetric reduction in the same regions.

CONCLUSIONS

In patients with persisting neurocognitive deficits, LGI1-LE may produce chronic volume loss in specific areas of the medial temporal lobe; however, this appears to follow a subacute period of volume enlargement possibly driven by neuro-inflammatory processes.

Longitudinal change in ATN biomarkers in cognitively normal individuals

BACKGROUND

Biomarkers for amyloid, tau, and neurodegeneration (ATN) have predictive value for clinical progression, but it is not clear how individuals move through these stages. We examined changes in ATN profiles over time, and investigated determinants of change in A status, in a sample of cognitively normal individuals presenting with subjective cognitive decline (SCD).

METHODS

We included 92 individuals with SCD from the SCIENCe project with [¹⁸F]florbetapir PET (A) available at two time points (65 ± 8y, 42% female, MMSE 29 ± 1, follow-up 2.5 ± 0.7y). We additionally used [¹⁸F]flortaucipir PET for T and medial temporal atrophy score on MRI for N. Thirty-nine individuals had complete biomarker data at baseline and follow-up, enabling the construction of ATN profiles at two time points. All underwent extensive neuropsychological assessments (follow-up time 4.9 ± 2.8y, median number of visits n = 4). We investigated changes in biomarker status and ATN profiles over time. We assessed which factors predisposed for a change from A- to A+ using logistic regression. We additionally used linear mixed models to assess change from A- to A+, compared to the group that remained A- at follow-up, as predictor for cognitive decline.

RESULTS

At baseline, 62% had normal AD biomarkers (A-T-N- n = 24), 5% had non-AD pathologic change (A-T-N+ n = 2,) and 33% fell within the Alzheimer's continuum (A+T-N- n = 9, A+T+N- n = 3, A+T+N+ n = 1). Seventeen subjects (44%) changed to another ATN profile over time. Only 6/17 followed the Alzheimer's disease sequence of A → T → N, while 11/17 followed a different order (e.g., reverted back to negative biomarker status). APOE ε4 carriership inferred an increased risk of changing from A- to A+ (OR 5.2 (95% CI 1.2-22.8)). Individuals who changed from A- to A+, showed subtly steeper decline on Stroop I (β - 0.03 (SE 0.01)) and Stroop III (- 0.03 (0.01)), compared to individuals who remained A-.

CONCLUSION

We observed considerable variability in the order of ATN biomarkers becoming abnormal. Individuals who became A+ at follow-up showed subtle decline on tests for attention and executive functioning, confirming clinical relevance of amyloid positivity.

Distinct stress-related changes in intrinsic amygdala connectivity predict subsequent positive and negative memory performance

Experiencing stress before an event can influence how that event is later remembered. In the current study, we examine how individual differences in one's physiological response to a stressor are related to changes to underlying brain states and memory performance. Specifically, we examined how changes in intrinsic amygdala connectivity relate to positive and negative memory performance as a function of stress response, defined as a change in cortisol. Twenty-five participants underwent a social stressor before an incidental emotional memory encoding task. Cortisol samples were obtained before and after the stressor to measure individual differences in stress response. Three resting state scans (pre-stressor, post-stressor/pre-encoding and post-encoding) were conducted to evaluate pre- to post-stressor and pre- to post-encoding changes to intrinsic amygdala connectivity. Analyses examined relations between greater cortisol changes and connectivity changes. Greater cortisol increases were associated with a greater decrease in prefrontal-amygdala connectivity following the stressor and a reversal in the relation between prefrontal-amygdala connectivity and negative vs. positive memory performance. Greater cortisol increases were also associated with a greater increase in amygdala connectivity with a number of posterior sensory regions following encoding. Consistent with prior findings in non-stressed individuals, pre- to post-encoding increases in amygdala-posterior connectivity were associated with greater negative relative to positive memory performance, although this was specific to lateral rather than medial posterior regions and to participants with the greatest cortisol changes. These findings suggest that stress response is associated with changes in intrinsic connectivity that have downstream effects on the valence of remembered emotional content.

Musical Activity During Life Is Associated With Multi-Domain Cognitive and Brain Benefits in Older Adults

Regular musical activity as a complex multimodal lifestyle activity is proposed to be protective against age-related cognitive decline and Alzheimer’s disease. This cross-sectional study investigated the association and interplay between musical instrument playing during life, multi-domain cognitive abilities and brain morphology in older adults (OA) from the DZNE-Longitudinal Cognitive Impairment and Dementia Study (DELCODE) study. Participants reporting having played a musical instrument across three life periods (n = 70) were compared to controls without a history of musical instrument playing (n = 70), well-matched for reserve proxies of education, intelligence, socioeconomic status and physical activity. Participants with musical activity outperformed controls in global cognition, working memory, executive functions, language, and visuospatial abilities, with no effects seen for learning and memory. The musically active group had greater gray matter volume in the somatosensory area, but did not differ from controls in higher-order frontal, temporal, or hippocampal volumes. However, the association between gray matter volume in distributed frontal-to-temporal regions and cognitive abilities was enhanced in participants with musical activity compared to controls. We show that playing a musical instrument during life relates to better late-life cognitive abilities and greater brain capacities in OA. Musical activity may serve as a multimodal enrichment strategy that could help preserve cognitive and brain health in late life. Longitudinal and interventional studies are needed to support this notion.

Discriminating Aging Cognitive Decline Spectrum Using PET and Magnetic Resonance Image Features

Background: The population aging increased the prevalence of brain diseases, like Alzheimer’s disease (AD), and early identification of individuals with higher odds of cognitive decline is essential to maintain quality of life. Imaging evaluation of individuals at risk of cognitive decline includes biomarkers extracted from brain positron emission tomography (PET) and structural magnetic resonance imaging (MRI). Objective: We propose investigating ensemble models to classify groups in the aging cognitive decline spectrum by combining features extracted from single imaging modalities and combinations of imaging modalities (FDG+AMY+MRI, and a PET ensemble). Methods: We group imaging data of 131 individuals into four classes related to the individuals’ cognitive assessment in baseline and follow-up: stable cognitive non-impaired; individuals converting to mild cognitive impairment (MCI) syndrome; stable MCI; and Alzheimer’s clinical syndrome. We assess the performance of four algorithms using leave-one-out cross-validation: decision tree classifier, random forest (RF), light gradient boosting machine (LGBM), and categorical boosting (CAT). The performance analysis of models is evaluated using balanced accuracy before and after using Shapley Additive exPlanations with recursive feature elimination (SHAP-RFECV) method. Results: Our results show that feature selection with CAT or RF algorithms have the best overall performance in discriminating early cognitive decline spectrum mainly using MRI imaging features. Conclusion: Use of CAT or RF algorithms with SHAP-RFECV shows good discrimination of early stages of aging cognitive decline, mainly using MRI image features. Further work is required to analyze the impact of selected brain regions and their correlation with cognitive decline spectrum.

Integrated 3d flow-based multi-atlas brain structure segmentation

MRI brain structure segmentation plays an important role in neuroimaging studies. Existing methods either spend much CPU time, require considerable annotated data, or fail in segmenting volumes with large deformation. In this paper, we develop a novel multi-atlas-based algorithm for 3D MRI brain structure segmentation. It consists of three modules: registration, atlas selection and label fusion. Both registration and label fusion leverage an integrated flow based on grayscale and SIFT features. We introduce an effective and efficient strategy for atlas selection by employing the accompanying energy generated in the registration step. A 3D sequential belief propagation method and a 3D coarse-to-fine flow matching approach are developed in both registration and label fusion modules. The proposed method is evaluated on five public datasets. The results show that it has the best performance in almost all the settings compared to competitive methods such as ANTs, Elastix, Learning to Rank and Joint Label Fusion. Moreover, our registration method is more than 7 times as efficient as that of ANTs SyN, while our label transfer method is 18 times faster than Joint Label Fusion in CPU time. The results on the ADNI dataset demonstrate that our method is applicable to image pairs that require a significant transformation in registration. The performance on a composite dataset suggests that our method succeeds in a cross-modality manner. The results of this study show that the integrated 3D flow-based method is effective and efficient for brain structure segmentation. It also demonstrates the power of SIFT features, multi-atlas segmentation and classical machine learning algorithms for a medical image analysis task. The experimental results on public datasets show the proposed method's potential for general applicability in various brain structures and settings.

Alzheimer’s Disease Prediction Using Attention Mechanism with Dual-Phase 18F-Florbetaben Images

Introduction

Amyloid-beta (Aβ) imaging test plays an important role in the early diagnosis and research of biomarkers of Alzheimer’s disease (AD) but a single test may produce Aβ-negative AD or Aβ-positive cognitively normal (CN). In this study, we aimed to distinguish AD from CN with dual-phase 18F-Florbetaben (FBB) via a deep learning–based attention method and evaluate the AD positivity scores compared to late-phase FBB which is currently adopted for AD diagnosis.

Materials and Methods

A total of 264 patients (74 CN and 190 AD), who underwent FBB imaging test and neuropsychological tests, were retrospectively analyzed. Early- and delay-phase FBB images were spatially normalized with an in-house FBB template. The regional standard uptake value ratios were calculated with the cerebellar region as a reference region and used as independent variables that predict the diagnostic label assigned to the raw image.

Results

AD positivity scores estimated from dual-phase FBB showed better accuracy (ACC) and area under the receiver operating characteristic curve (AUROC) for AD detection (ACC: 0.858, AUROC: 0.831) than those from delay phase FBB imaging (ACC: 0.821, AUROC: 0.794). AD positivity score estimated by dual-phase FBB (R: −0.5412) shows a higher correlation with psychological test compared to only dFBB (R: −0.2975). In the relevance analysis, we observed that LSTM uses different time and regions of early-phase FBB for each disease group for AD detection.

Conclusions

These results show that the aggregated model with dual-phase FBB with long short-term memory and attention mechanism can be used to provide a more accurate AD positivity score, which shows a closer association with AD, than the prediction with only a single phase FBB.

Quantitative assessment of myelin density using [11C]MeDAS PET in patients with multiple sclerosis: a first-in-human study

PURPOSE

Multiple sclerosis (MS) is a disease characterized by inflammatory demyelinated lesions. New treatment strategies are being developed to stimulate myelin repair. Quantitative myelin imaging could facilitate these developments. This first-in-man study aimed to evaluate [11C]MeDAS as a PET tracer for myelin imaging in humans.

METHODS

Six healthy controls and 11 MS patients underwent MRI and dynamic [11C]MeDAS PET scanning with arterial sampling. Lesion detection and classification were performed on MRI. [11C]MeDAS time-activity curves of brain regions and MS lesions were fitted with various compartment models for the identification of the best model to describe [11C]MeDAS kinetics. Several simplified methods were compared to the optimal compartment model.

RESULTS

Visual analysis of the fits of [11C]MeDAS time-activity curves showed no preference for irreversible (2T3k) or reversible (2T4k) two-tissue compartment model. Both volume of distribution and binding potential estimates showed a high degree of variability. As this was not the case for 2T3k-derived net influx rate (Ki), the 2T3k model was selected as the model of choice. Simplified methods, such as SUV and MLAIR2 correlated well with 2T3k-derived Ki, but SUV showed subject-dependent bias when compared to 2T3k. Both the 2T3k model and the simplified methods were able to differentiate not only between gray and white matter, but also between lesions with different myelin densities.

CONCLUSION

[11C]MeDAS PET can be used for quantification of myelin density in MS patients and is able to distinguish differences in myelin density within MS lesions. The 2T3k model is the optimal compartment model and MLAIR2 is the best simplified method for quantification.

TRIAL REGISTRATION

NL7262. Registered 18 September 2018.

[Correction of Partial Volume Effect Using CT Images in Brain 18F-FDG PET]

PURPOSE

We performed partial volume effect correction of PET images using ¹⁸F-FDG-PET and CT images taken consecutively, compared it with correction using MRI images, and investigated the usefulness of correction using CT images.

METHODS

A total of 9 clinically normal subjects were included in the study, and the CT and MRI images of each subject were segmented and normalized. PET images were coregistered to each morphological image and then normalized. The normalized morphological images of each subject were used to mask the brain atlas and to correct for the partial volume effect. For each brain region, comparison of counts, two-group test between CT- and MRI-corrected groups, and correlation analysis were performed.

RESULTS

As a result of correction, some error was observed between the two groups. Correlation analysis showed strong positive correlations in many areas, but weak correlations were found in some areas. In the region where significant differences were found, the two groups showed strong positive correlation, and in the region where weak correlation was found, the error tended to be small.

CONCLUSION

It is suggested that the correction by CT can be performed with the same accuracy, although some errors are generated compared with MRI.

Does Loss of Integrity of the Cingulum Bundle Link Amyloid-β Accumulation and Neurodegeneration in Alzheimer’s Disease?

Background: Alzheimer’s disease is characterized by the accumulation of amyloid-β (Aβ) into plaques, aggregation of tau into neurofibrillary tangles, and neurodegenerative processes including atrophy. However, there is a poorly understood spatial discordance between initial Aβ deposition and local neurodegeneration. Objective: Here, we test the hypothesis that the cingulum bundle links Aβ deposition in the cingulate cortex to medial temporal lobe (MTL) atrophy. Methods: 21 participants with mild cognitive impairment (MCI) from the UMC Utrecht memory clinic (UMCU, discovery sample) and 37 participants with MCI from Alzheimer’s Disease Neuroimaging Initiative (ADNI, replication sample) with available Aβ-PET scan, T1-weighted and diffusion-weighted MRI were included. Aβ load of the cingulate cortex was measured by the standardized uptake value ratio (SUVR), white matter integrity of the cingulum bundle was assessed by mean diffusivity and atrophy of the MTL by normalized MTL volume. Relationships were tested with linear mixed models, to accommodate multiple measures for each participant. Results: We found at most a weak association between cingulate Aβ and MTL volume (added R2 <0.06), primarily for the posterior hippocampus. In neither sample, white matter integrity of the cingulum bundle was associated with cingulate Aβ or MTL volume (added R2 <0.01). Various sensitivity analyses (Aβ-positive individuals only, posterior cingulate SUVR, MTL sub region volume) provided similar results. Conclusion: These findings, consistent in two independent cohorts, do not support our hypothesis that loss of white matter integrity of the cingulum is a connecting factor between cingulate gyrus Aβ deposition and MTL atrophy.

The Relationship Between Frontostriatal Connectivity and Striatal Dopamine Function in Schizophrenia: An 18F-DOPA PET and Diffusion Tensor Imaging Study in Treatment Responsive and Resistant Patients

OBJECTIVE

Striatal dopamine dysfunction caused by cortical abnormalities is a leading hypothesis of schizophrenia. Although prefrontal cortical pathology is negatively correlated with striatal dopamine synthesis, the relationship between structural frontostriatal connectivity and striatal dopamine synthesis has not been proved in patients with schizophrenia with different treatment response. We therefore investigated the relationship between frontostriatal connectivity and striatal dopamine synthesis in treatment-responsive schizophrenia (non-TRS) and compared them to treatment-resistant schizophrenia (TRS) and healthy controls (HC).

METHODS

Twenty-four patients with schizophrenia and twelve HC underwent [18F] DOPA PET scans to measure dopamine synthesis capacity (the influx rate constant Kicer) and diffusion 3T MRI to measure structural connectivity (fractional anisotropy, FA). Connectivity was assessed in 2 major frontostriatal tracts. Associations between Kicer and FA in each group were evaluated using Spearman's rho correlation coefficients.

RESULTS

Non-TRS showed a negative correlation (r=-0.629, p=0.028) between connectivity of dorsolateral prefrontal cortex-associative striatum (DLPFC-AST) and dopamine synthesis capacity of associative striatum but this was not evident in TRS (r=-0.07, p=0.829) and HC (r=-0.277, p=0.384).

CONCLUSION

Our findings are consistent with the hypothesis of dysregulation of the striatal dopaminergic system being related to prefrontal cortex pathology localized to connectivity of DLPFC-AST in non-TRS, and also extend the hypothesis to suggest that different mechanisms underlie the pathophysiology of non-TRS and TRS.

Multiview Feature Learning With Multiatlas-Based Functional Connectivity Networks for MCI Diagnosis

Functional connectivity (FC) networks built from resting-state functional magnetic resonance imaging (rs-fMRI) has shown promising results for the diagnosis of Alzheimer's disease and its prodromal stage, that is, mild cognitive impairment (MCI). FC is usually estimated as a temporal correlation of regional mean rs-fMRI signals between any pair of brain regions, and these regions are traditionally parcellated with a particular brain atlas. Most existing studies have adopted a predefined brain atlas for all subjects. However, the constructed FC networks inevitably ignore the potentially important subject-specific information, particularly, the subject-specific brain parcellation. Similar to the drawback of the "single view" (versus the "multiview" learning) in medical image-based classification, FC networks constructed based on a single atlas may not be sufficient to reveal the underlying complicated differences between normal controls and disease-affected patients due to the potential bias from that particular atlas. In this study, we propose a multiview feature learning method with multiatlas-based FC networks to improve MCI diagnosis. Specifically, a three-step transformation is implemented to generate multiple individually specified atlases from the standard automated anatomical labeling template, from which a set of atlas exemplars is selected. Multiple FC networks are constructed based on these preselected atlas exemplars, providing multiple views of the FC network-based feature representations for each subject. We then devise a multitask learning algorithm for joint feature selection from the constructed multiple FC networks. The selected features are jointly fed into a support vector machine classifier for multiatlas-based MCI diagnosis. Extensive experimental comparisons are carried out between the proposed method and other competing approaches, including the traditional single-atlas-based method. The results indicate that our method significantly improves the MCI classification, demonstrating its promise in the brain connectome-based individualized diagnosis of brain diseases.

A framework For brain atlases: Lessons from seizure dynamics

Brain maps, or atlases, are essential tools for studying brain function and organization. The abundance of available atlases used across the neuroscience literature, however, creates an implicit challenge that may alter the hypotheses and predictions we make about neurological function and pathophysiology. Here, we demonstrate how parcellation scale, shape, anatomical coverage, and other atlas features may impact our prediction of the brain's function from its underlying structure. We show how network topology, structure-function correlation (SFC), and the power to test specific hypotheses about epilepsy pathophysiology may change as a result of atlas choice and atlas features. Through the lens of our disease system, we propose a general framework and algorithm for atlas selection. This framework aims to maximize the descriptive, explanatory, and predictive validity of an atlas. Broadly, our framework strives to provide empirical guidance to neuroscience research utilizing the various atlases published over the last century.

Alzheimer's disease pattern derived from relative cerebral flow as an alternative for the metabolic pattern using SSM/PCA

BACKGROUND

2-Deoxy-2-[¹⁸F]fluoroglucose (FDG) PET is an important tool for the identification of Alzheimer's disease (AD) patients through the characteristic neurodegeneration pattern that these patients present. Regional cerebral blood flow (rCBF) images derived from dynamic ¹¹C-labelled Pittsburgh Compound B (PIB) have been shown to present a similar pattern as FDG. Moreover, multivariate analysis techniques, such as scaled subprofile modelling using principal component analysis (SSM/PCA), can be used to generate disease-specific patterns (DP) that may aid in the classification of subjects. Therefore, the aim of this study was to compare rCBF AD-DPs with FDG AD-DP and their respective performances. Therefore, 52 subjects were included in this study. Fifteen AD and 16 healthy control subjects were used to generate four AD-DP: one based on relative cerebral trace blood (R₁), two based on time-weighted average of initial frame intervals (ePIB), and one based on FDG images. Furthermore, 21 subjects diagnosed with mild cognitive impairment were tested against these AD-DPs.

RESULTS

In general, the rCBF and FDG AD-DPs were characterized by a reduction in cortical frontal, temporal, and parietal lobes. FDG and rCBF methods presented similar score distribution.

CONCLUSION

rCBF images may provide an alternative for FDG PET scans for the identification of AD patients through SSM/PCA.

Combined Multi-Atlas and Multi-Layer Perception for Alzheimer's Disease Classification

Alzheimer's disease (AD) is a progressive and irreversible neurodegenerative disease. To distinguish the stage of the disease, AD classification technology challenge has been proposed in Pattern Recognition and Computer Vision 2021 (PRCV 2021) which provides the gray volume and average cortical thickness data extracted in multiple atlases from magnetic resonance imaging (MRI). Traditional methods either train with convolutional neural network (CNN) by MRI data to adapt the spatial features of images or train with recurrent neural network (RNN) by temporal features to predict the next stage. However, the morphological features from the challenge have been extracted into discrete values. We present a multi-atlases multi-layer perceptron (MAMLP) approach to deal with the relationship between morphological features and the stage of the disease. The model consists of multiple multi-layer perceptron (MLP) modules, and morphological features extracted from different atlases will be classified by different MLP modules. The final vote of all classification results obtains the predicted disease stage. Firstly, to preserve the diversity of brain features, the most representative atlases are chosen from groups of similar atlases, and one atlas is selected in each group. Secondly, each atlas is fed into one MLP to fetch the score of the classification. Thirdly, to obtain more stable results, scores from different atlases are combined to vote the result of the classification. Based on this approach, we rank 10th among 373 teams in the challenge. The results of the experiment indicate as follows: (1) Group selection of atlas reduces the number of features required without reducing the accuracy of the model; (2) The MLP architecture achieves better performance than CNN and RNN networks in morphological features; and (3) Compared with other networks, the combination of multiple MLP networks has faster convergence of about 40% and makes the classification more stable.