Manual validation of FreeSurfer's automated hippocampal segmentation in normal aging, mild cognitive impairment, and Alzheimer Disease subjects

Comparative assessment of established and deep learning-based segmentation methods for hippocampal volume estimation in brain magnetic resonance imaging analysis

In this study, our objective was to assess the performance of two deep learning-based hippocampal segmentation methods, SynthSeg and TigerBx, which are readily available to the public. We contrasted their performance with that of two established techniques, FreeSurfer-Aseg and FSL-FIRST, using three-dimensional T1-weighted MRI scans (n = 1447) procured from public databases. Our evaluation focused on the accuracy and reproducibility of these tools in estimating hippocampal volume. The findings suggest that both SynthSeg and TigerBx are on a par with Aseg and FIRST in terms of segmentation accuracy and reproducibility, but offer a significant advantage in processing speed, generating results in less than 1 min compared with several minutes to hours for the latter tools. In terms of Alzheimer's disease classification based on the hippocampal atrophy rate, SynthSeg and TigerBx exhibited superior performance. In conclusion, we evaluated the capabilities of two deep learning-based segmentation techniques. The results underscore their potential value in clinical and research environments, particularly when investigating neurological conditions associated with hippocampal structures.

Fully Automated Hippocampus Segmentation using T2-informed Deep Convolutional Neural Networks

Hippocampal atrophy (tissue loss) has become a fundamental outcome parameter in clinical trials on Alzheimer's disease. To accurately estimate hippocampus volume and track its volume loss, a robust and reliable segmentation is essential. Manual hippocampus segmentation is considered the gold standard but is extensive, time-consuming, and prone to rater bias. Therefore, it is often replaced by automated programs like FreeSurfer, one of the most commonly used tools in clinical research. Recently, deep learning-based methods have also been successfully applied to hippocampus segmentation. The basis of all approaches are clinically used T1-weighted whole-brain MR images with approximately 1 mm isotropic resolution. However, such T1 images show low contrast-to-noise ratios (CNRs), particularly for many hippocampal substructures, limiting delineation reliability. To overcome these limitations, high-resolution T2-weighted scans are suggested for better visualization and delineation, as they show higher CNRs and usually allow for higher resolutions. Unfortunately, such time-consuming T2-weighted sequences are not feasible in a clinical routine. We propose an automated hippocampus segmentation pipeline leveraging deep learning with T2-weighted MR images for enhanced hippocampus segmentation of clinical T1-weighted images based on a series of 3D convolutional neural networks and a specifically acquired multi-contrast dataset. This dataset consists of corresponding pairs of T1- and high-resolution T2-weighted images, with the T2 images only used to create more accurate manual ground truth annotations and to train the segmentation network. The T2-based ground truth labels were also used to evaluate all experiments by comparing the masks visually and by various quantitative measures. We compared our approach with four established state-of-the-art hippocampus segmentation algorithms (FreeSurfer, ASHS, HippoDeep, HippMapp3r) and demonstrated a superior segmentation performance. Moreover, we found that the automated segmentation of T1-weighted images benefits from the T2-based ground truth data. In conclusion, this work showed the beneficial use of high-resolution, T2-based ground truth data for training an automated, deep learning-based hippocampus segmentation and provides the basis for a reliable estimation of hippocampal atrophy in clinical studies.

Is Hippocampal Volume a Relevant Early Marker of Dementia?

OBJECTIVE

Hippocampal volume (HV) is a key imaging marker to improve Alzheimer's disease risk prediction. However, longitudinal studies are rare, and hippocampus may also be implicated in the subtle aging-related cognitive decline observed in dementia-free individuals. Our aim was to determine whether HV, measured by manual or automatic segmentation, is associated with dementia risk and cognitive decline in participants with and without incident dementia.

METHODS

At baseline, 510 dementia-free participants from the French longitudinal ESPRIT cohort underwent magnetic resonance imaging. HV was measured by manual and by automatic segmentation (FreeSurfer 6.0). The presence of dementia and cognitive functions were investigated at each follow-up (2, 4, 7, 10, 12, and 15 years). Cox proportional hazards models and linear mixed models were used to assess the association of HV with dementia risk and with cognitive decline, respectively.

RESULTS

During the 15-years follow-up, 42 participants developed dementia. Reduced HV (regardless of the measurement method) was significantly associated with higher dementia risk and cognitive decline in the whole sample. However, only the automatically measured HV was associated with cognitive decline in dementia-free participants.

CONCLUSION

These results suggest that HV can be used to predict the long-term risk of dementia but also cognitive decline in a dementia-free population. This raises the question of the relevance of HV measurement as an early marker of dementia in the general population.

Validating ASHS-T1 automated entorhinal and transentorhinal cortical segmentation in Alzheimer's disease

The current study aimed to validate entorhinal and transentorhinal cortical volumes measured by the automated segmentation tool Automatic Segmentation of Hippocampal Subfields (ASHS-T1). The study sample comprised 34 healthy controls (HCs), 37 individuals with amnestic mild cognitive impairment (aMCI), and 29 individuals with Alzheimer's disease (AD) dementia from the Alzheimer's Disease Neuroimaging Initiative (ADNI) database. Entorhinal and transentorhinal cortical volumes were assessed using ASHS-T1, manual segmentation, as well as a widely used automated segmentation tool, FreeSurfer v6.0.1. Mean differences, intraclass correlation coefficients, and Bland-Altman plots were computed. ASHS-T1 tended to underestimate entorhinal and transentorhinal cortical volumes relative to manual segmentation and FreeSurfer. There was variable consistency and low agreement between ASHS-T1 and manual segmentation volumes. There was low-to-moderate consistency and low agreement between ASHS-T1 and FreeSurfer volumes. There was a trend toward higher consistency and agreement for the entorhinal cortex in the aMCI and AD groups compared to the HC group. Despite the differences in volume measurements, ASHS-T1 was sensitive to entorhinal and transentorhinal cortical atrophy in both early and late disease stages. Based on the current study, ASHS-T1 appears to be a promising tool for automated entorhinal and transentorhinal cortical volume measurement in individuals with likely underlying AD.

Deriving a robust deep-learning model for subcortical brain segmentation by using a large-scale database: Preprocessing, reproducibility, and accuracy of volume estimation

Increasing the accuracy and reproducibility of subcortical brain segmentation is advantageous in various related clinical applications. In this study, we derived a segmentation method based on a convolutional neural network (i.e., U-Net) and a large-scale database consisting of 7039 brain T1-weighted MRI data samples. We evaluated the method by using experiments focused on three distinct topics, namely, the necessity of preprocessing steps, cross-institutional and longitudinal reproducibility, and volumetric accuracy. The optimized model, MX_RW-where "MX" is a mix of RW and nonuniform intensity normalization data and "RW" is raw data with basic preprocessing-did not require time-consuming preprocessing steps, such as nonuniform intensity normalization or image registration, for brain MRI before segmentation. Cross-institutional testing revealed that MX_RW (Dice similarity coefficient: 0.809, coefficient of variation: 4.6%, and Pearson's correlation coefficient: 0.979) exhibited a performance comparable with that of FreeSurfer (Dice similarity coefficient: 0.798, coefficient of variation: 5.6%, and Pearson's correlation coefficient: 0.973). The computation time per dataset of MX_RW was generally less than 5 s (even when tested without graphics processing units), which was notably faster than FreeSurfer. Thus, for time-restricted applications, MX_RW represents a competitive alternative to FreeSurfer.

Longitudinal volumetric evaluation of hippocampus and amygdala subregions in recent trauma survivors

The hippocampus and the amygdala play a central role in post-traumatic stress disorder (PTSD) pathogenesis. While alternations in volumes of both regions have been consistently observed in individuals with PTSD, it remains unknown whether these reflect pre-trauma vulnerability traits or acquired post-trauma consequences of the disorder. Here, we conducted a longitudinal panel study of adult civilian trauma survivors admitted to a general hospital emergency department (ED). One hundred eligible participants (mean age = 32.97 ± 10.97, n = 56 females) completed both clinical interviews and structural MRI scans at 1-, 6-, and 14-months after ED admission (alias T1, T2, and T3). While all participants met PTSD diagnosis at T1, only n = 29 still met PTSD diagnosis at T3 (a "non-Remission" Group), while n = 71 did not (a "Remission" Group). Bayesian multilevel modeling analysis showed robust evidence for smaller right hippocampus volume (P+ of ~0.014) and moderate evidence for larger left amygdala volume (P+ of ~0.870) at T1 in the "non-Remission" group, compared to the "Remission" group. Subregion analysis further demonstrated robust evidence for smaller volume in the subiculum and right CA1 hippocampal subregions (P+ of ~0.021-0.046) in the "non-Remission" group. No time-dependent volumetric changes (T1 to T2 to T3) were observed across all participants or between groups. Results support the "vulnerability trait" hypothesis, suggesting that lower initial volumes of specific hippocampus subregions are associated with non-remitting PTSD. The stable volume of all hippocampal and amygdala subregions does not support the idea of consequential, progressive, stress-related atrophy during the first critical year following trauma exposure.

Changes in Measures of Vestibular and Balance Function and Hippocampus Volume in Alzheimer's Disease and Mild Cognitive Impairment

OBJECTIVE

To test the hypotheses that people with Alzheimer's disease and mild cognitive impairment have increased frequency of vestibular impairments and decreased hippocampal volume compared with healthy age-matched controls.

STUDY DESIGN

Retrospective, with some historical controls.

SETTING

Out-patient, tertiary care center.

SUBJECTS

People with mild to moderate dementia diagnosed with Alzheimer's disease and with mild cognitive impairment. Main Outcome Measures: A standard clinical battery of objective tests of the vestibular system, and screening for balance; available clinical diagnostic magnetic resonance imaging (MRIs) were reviewed and postprocessed to quantify the left and right hippocampal volumes utilizing both manual segmentation and computer automated segmentation.

RESULTS

Study subjects (N = 26) had significantly more vestibular impairments, especially on Dix-Hallpike maneuvers and cervical vestibular evoked myogenic potentials (cVEMP), than historical controls. No differences were found between mild and moderate dementia subjects. Independence on instrumental activities of daily living in subjects with age-normal balance approached statistical differences from subjects with age-abnormal balance. MRI data were available for 11 subjects. Subjects with abnormal cVEMP had significantly reduced left hippocampal MRIs using manual segmentation compared with subjects with normal cVEMP.

CONCLUSION

The data from this small sample support and extend previous evidence for vestibular impairments in this population. The small MRI sample set should be considered preliminary evidence, and suggests the need for further research, with a more robust sample and high-resolution MRIs performed for the purpose of hippocampal analysis.

Validation of an automatic tool for the rapid measurement of brain atrophy and white matter hyperintensity: QyScore®

OBJECTIVES

QyScore® is an imaging analysis tool certified in Europe (CE marked) and the US (FDA cleared) for the automatic volumetry of grey and white matter (GM and WM respectively), hippocampus (HP), amygdala (AM), and white matter hyperintensity (WMH). Here we compare QyScore® performances with the consensus of expert neuroradiologists.

METHODS

Dice similarity coefficient (DSC) and the relative volume difference (RVD) for GM, WM volumes were calculated on 50 3DT1 images. DSC and the F1 metrics were calculated for WMH on 130 3DT1 and FLAIR images. For each index, we identified thresholds of reliability based on current literature review results. We hypothesized that DSC/F1 scores obtained using QyScore® markers would be higher than the threshold. In contrast, RVD scores would be lower. Regression analysis and Bland-Altman plots were obtained to evaluate QyScore® performance in comparison to the consensus of three expert neuroradiologists.

RESULTS

The lower bound of the DSC/F1 confidence intervals was higher than the threshold for the GM, WM, HP, AM, and WMH, and the higher bounds of the RVD confidence interval were below the threshold for the WM, GM, HP, and AM. QyScore®, compared with the consensus of three expert neuroradiologists, provides reliable performance for the automatic segmentation of the GM and WM volumes, and HP and AM volumes, as well as WMH volumes.

CONCLUSIONS

QyScore® represents a reliable medical device in comparison with the consensus of expert neuroradiologists. Therefore, QyScore® could be implemented in clinical trials and clinical routine to support the diagnosis and longitudinal monitoring of neurological diseases.

KEY POINTS

• QyScore® provides reliable automatic segmentation of brain structures in comparison with the consensus of three expert neuroradiologists. • QyScore® automatic segmentation could be performed on MRI images using different vendors and protocols of acquisition. In addition, the fast segmentation process saves time over manual and semi-automatic methods. • QyScore® could be implemented in clinical trials and clinical routine to support the diagnosis and longitudinal monitoring of neurological diseases.

Agreement Between Automated and Manual MRI Volumetry in Alzheimer's Disease: A Systematic Review and Meta-Analysis

BACKGROUND

Automated magnetic resonance imaging (MRI) volumetry is a promising tool to evaluate regional brain volumes in dementia and especially Alzheimer's disease (AD).

PURPOSE

To compare automated methods and the gold standard manual segmentation in measuring regional brain volumes on MRI across healthy controls, patients with mild cognitive impairment, and patients with dementia due to AD.

STUDY TYPE

Systematic review and meta-analysis.

DATA SOURCES

MEDLINE, Embase, and PsycINFO were searched through October 2021.

FIELD STRENGTH

1.0 T, 1.5 T, or 3.0 T.

ASSESSMENT

Two review authors independently identified studies for inclusion and extracted data. Methodological quality was assessed using the Quality Assessment of Diagnostic Accuracy Studies 2 (QUADAS-2).

STATISTICAL TESTS

Standardized mean differences (SMD; Hedges' g) were pooled using random-effects meta-analysis with robust variance estimation. Subgroup analyses were undertaken to explore potential sources of heterogeneity. Sensitivity analyses were conducted to examine the impact of the within-study correlation between effect estimates on the meta-analysis results.

RESULTS

Seventeen studies provided sufficient data to evaluate the hippocampus, lateral ventricles, and parahippocampal gyrus. The pooled SMD for the hippocampus, lateral ventricles, and parahippocampal gyrus were 0.22 (95% CI -0.50 to 0.93), 0.12 (95% CI -0.13 to 0.37), and -0.48 (95% CI -1.37 to 0.41), respectively. For the hippocampal data, subgroup analyses suggested that the pooled SMD was invariant across clinical diagnosis and field strength. Subgroup analyses could not be conducted on the lateral ventricles data and the parahippocampal gyrus data due to insufficient data. The results were robust to the selected within-study correlation value.

DATA CONCLUSION

While automated methods are generally comparable to manual segmentation for measuring hippocampal, lateral ventricle, and parahippocampal gyrus volumes, wide 95% CIs and large heterogeneity suggest that there is substantial uncontrolled variance. Thus, automated methods may be used to measure these regions in patients with AD but should be used with caution.

EVIDENCE LEVEL

3 TECHNICAL EFFICACY: Stage 3.

Charting the human amygdala development across childhood and adolescence: Manual and automatic segmentation

The developmental pattern of the amygdala throughout childhood and adolescence has been inconsistently reported in previous neuroimaging studies. Given the relatively small size of the amygdala on full brain MRI scans, discrepancies may be partly due to methodological differences in amygdalar segmentation. To investigate the impact of volume extraction methods on amygdala volume, we compared FreeSurfer, FSL and volBrain segmentation measurements with those obtained by manual tracing. The manual tracing method, which we used as the 'gold standard', exhibited almost perfect intra- and inter-rater reliability. We observed systematic differences in amygdala volumes between automatic (FreeSurfer and volBrain) and manual methods. Specifically, compared with the manual tracing, FreeSurfer estimated larger amygdalae, and volBrain produced smaller amygdalae while FSL demonstrated a mixed pattern. The tracing bias was not uniform, but higher for smaller amygdalae. We further modeled amygdalar growth curves using accelerated longitudinal cohort data from the Chinese Color Nest Project (http://deepneuro.bnu.edu.cn/?p=163). Trajectory modeling and statistical assessments of the manually traced amygdalae revealed linearly increasing and parallel developmental patterns for both girls and boys, although the amygdalae of boys were larger than those of girls. Compared to these trajectories, the shapes of developmental curves were similar when using the volBrain derived volumes. FreeSurfer derived trajectories had more nonlinearities and appeared flatter. FSL derived trajectories demonstrated an inverted U shape and were significantly different from those derived from manual tracing method. The use of amygdala volumes adjusted for total gray-matter volumes, but not intracranial volumes, resolved the shape discrepancies and led to reproducible growth curves between manual tracing and the automatic methods (except FSL). Our findings revealed steady growth of the human amygdala, mirroring its functional development across the school age. Methodological improvements are warranted for current automatic tools to achieve more accurate amygdala structure at school age, calling for next generation tools.

Seizure control does not predict hippocampal subfield volume change in children with focal drug-resistant epilepsy

Background and purpose

Recurrent seizures have been reported to induce neuronal loss in the hippocampus. It is unclear whether seizure control influences hippocampal volume. The aims of this study were to determine if there was a change in total or subfield hippocampal volume over time in children with focal drug-resistant epilepsy, and whether seizure control influenced total or subfield hippocampal volumes.

Methods

Using FreeSurfer’s automated segmentation of brain magnetic resonance imaging scans, we calculated the total and subfield (including CA1, CA3, CA4, subiculum, presubiculum, parasubiculum, molecular layer and dentate gyrus) hippocampal volumes of children with non-lesional focal epilepsy. Seizure frequency and hippocampal volumes were assessed at baseline and follow-up. Patients were classified into those who were seizure free or have improvement in seizures (group 1) and those with no improvement in seizures (group 2) at follow-up.

Results

Thirty-seven patients were included, with mean age 10.31 ± 3.68 years at baseline. The interval between the two magnetic resonance imaging scans was 2.59 ± 1.25 years. There was no significant difference in the total and subfield hippocampal volumes for the whole cohort at follow-up compared to baseline (all P > 0.002). Seizure control of the two groups did not predict total or subfield hippocampal volume, after controlling for baseline volume, age, severity of seizure frequency at baseline and time interval between the magnetic resonance imaging scans (all P > 0.002).

Conclusion

We have found that total and subfield hippocampal volumes did not change, and seizure control did not predict hippocampal volumes at follow-up in children with drug-resistant epilepsy.

Diagnostic Performance of Automated MRI Volumetry by icobrain dm for Alzheimer's Disease in a Clinical Setting: A REMEMBER Study

Background:

Magnetic resonance imaging (MRI) has become important in the diagnostic work-up of neurodegenerative diseases. icobrain dm, a CE-labeled and FDA-cleared automated brain volumetry software, has shown potential in differentiating cognitively healthy controls (HC) from Alzheimer’s disease (AD) dementia (ADD) patients in selected research cohorts.

Objective:

This study examines the diagnostic value of icobrain dm for AD in routine clinical practice, including a comparison to the widely used FreeSurfer software, and investigates if combined brain volumes contribute to establish an AD diagnosis.

Methods:

The study population included HC (n = 90), subjective cognitive decline (SCD, n = 93), mild cognitive impairment (MCI, n = 357), and ADD (n = 280) patients. Through automated volumetric analyses of global, cortical, and subcortical brain structures on clinical brain MRI T1w (n = 820) images from a retrospective, multi-center study (REMEMBER), icobrain dm’s (v.4.4.0) ability to differentiate disease stages via ROC analysis was compared to FreeSurfer (v.6.0). Stepwise backward regression models were constructed to investigate if combined brain volumes can differentiate between AD stages.

Results:

icobrain dm outperformed FreeSurfer in processing time (15–30 min versus 9–32 h), robustness (0 versus 67 failures), and diagnostic performance for whole brain, hippocampal volumes, and lateral ventricles between HC and ADD patients. Stepwise backward regression showed improved diagnostic accuracy for pairwise group differentiations, with highest performance obtained for distinguishing HC from ADD (AUC = 0.914; Specificity 83.0%; Sensitivity 86.3%).

Conclusion:

Automated volumetry has a diagnostic value for ADD diagnosis in routine clinical practice. Our findings indicate that combined brain volumes improve diagnostic accuracy, using real-world imaging data from a clinical setting.

APOEɛ4 Gene Dose and Sex Effects on Alzheimer's Disease MRI Biomarkers in Older Adults with Mild Cognitive Impairment

Background:

APOEɛ4 and sex have been linked to increased risk for conversion to Alzheimer’s disease (AD). However, the relationship between APOEɛ4 gene dose, sex, and AD biomarkers remains understudied.

Objective:

To investigate the effect of APOEɛ4 dose on AD biomarkers in a sample of older adults with mild cognitive impairment (MCI), and to examine whether APOEɛ4 dose modifies AD risk differently in MCI women and men.

Methods:

We examined cross-sectional AD biomarkers for participants with MCI (n = 930, 55–96 years old) from three large aging cohorts. Region of interest MRI volumes, global cognition, and episodic memory were analyzed by number of APOEɛ4 alleles and stratified by sex.

Results:

Across all participants, number of APOEɛ4 alleles was associated with smaller hippocampal and amygdala volumes and poorer cognition. When stratified by sex, women showed an APOEɛ4 dose effect for bilateral hippocampal and left amygdala volumes and cognition. In contrast, men showed an APOEɛ4 dose effect for hippocampal volumes with a trend in amygdala, but cognition did not differ between men with 1 and 2 APOEɛ4 alleles. Women with 2 APOEɛ4 alleles had poorer memory between 65–69 and poorer global cognition between 70–74 compared to men with 2 APOEɛ4 alleles.

Conclusion:

APOEɛ4 confers a dose effect on AD biomarkers in patients with MCI, and the number of APOEɛ4 alleles has a greater detrimental impact in women than men, which may be specific to a critical time window.

Validity of automated FreeSurfer segmentation compared to manual tracing in detecting prenatal alcohol exposure-related subcortical and corpus callosal alterations in 9- to 11-year-old children

In recent years a number of semi-automated and automated segmentation tools and brain atlases have been developed to facilitate morphometric analyses of large MRI datasets. These tools are much faster than manual tracing and demonstrate excellent test-retest reliabilities. Reliabilities of automated segmentations relative to "gold standard" manual tracings have, however, been shown to vary by brain region and in different cohorts. It remains uncertain to what extent smaller brain volumes and potential changes in grey/white matter contrasts in paediatric brains impact on the performance of automated methods, and how pathology may influence performance. This study examined whether using data from automated FreeSurfer segmentation would alter our ability, compared to manual segmentation, to detect prenatal alcohol exposure (PAE)-related volume changes in subcortical regions and the corpus callosum (CC) in pre-adolescent children. High-resolution T1-weighted images were acquired, using a sequence optimized for morphometric neuroanatomical analysis, on a Siemens 3T Allegra MRI scanner in 71 right-handed, 9- to 11-year-old children (27 fetal alcohol syndrome (FAS) and partial FAS (PFAS), 25 non-syndromal heavily exposed (HE) and 19 non-exposed controls) from a high-risk community in Cape Town, South Africa. Data from timeline follow-back interviews administered to the mothers prospectively during pregnancy were used to quantify the amount of alcohol (in ounces absolute alcohol per day, AA/day) that the children had been exposed to prenatally. Volumes of corpus callosum (CC) and bilateral caudate nuclei, hippocampi and nucleus accumbens (NA) were obtained by manual tracing and automated segmentation using both FreeSurfer versions 5.1 and 6.0. Reliability across methods was assessed using intraclass correlation (ICC) estimates for consistency and absolute agreement, and Cronbach's α. Ability to detect regions showing PAE effects was assessed separately for each segmentation method using ANOVA and linear regression of regional volumes with AA/day. Our results support findings from other studies showing excellent reliability across methods for easy-to-segment structures, such as the CC and caudate nucleus. Volumes from FreeSurfer 6.0 were smaller than those from version 5.1 in all regions except the right caudate, for which they were similar, and right hippocampus and CC, for which they were larger. Despite poor absolute agreement between methods in the NA and hippocampus, all three segmentation methods detected dose-dependent volume reductions in regions for which reliabilities on ICC consistency across methods reached at least 0.70, namely the CC, and bilateral caudate nuclei and hippocampi. PAE-related changes in the NA for which ICC consistency did not reach this minimum were inconsistent across methods and should be interpreted with caution. This is the first study to demonstrate in a pre-adolescent cohort the ability of automated segmentation with FreeSurfer to detect regional volume changes associated with pathology similar to those found using manual tracing.

The impact of hippocampal segmentation methods on correlations with clinical data

BACKGROUND

In vivo measurement of hippocampal volume with magnetic resonance imaging (MRI) has become an important element in neuroimaging research. However, hippocampal volumetric findings and their relationship with cardiovascular risk factors and memory performance are still controversial and inconsistent for non-demented adults.

PURPOSE

To compare total and regional hippocampal volumes from manual tracing and automated Freesurfer segmentation methods and their relationship with mid-life clinical data and late-life verbal episodic memory performance in older women.

MATERIAL AND METHODS

This study used structural MRI datasets from 161 women who were scanned in 2012 and underwent neuropsychological assessments. Of these participants, 135 women had completed baseline measures of cardiovascular risk factors in 1992.

RESULTS

Our results showed a significant correlation between manual tracing and automated Freesurfer output segmentations of total (r = 0.71), anterior (r = 0.65), and posterior (r = 0.38) hippocampal volumes. Mid-life Framingham Cardiovascular Risk Profile score is not associated with late-life hippocampal volumes, adjusted for intracranial volume, age, education, and apolipoprotein E gene ε4 status. Anterior hippocampal volume segmented either with manual tracing or automated Freesurfer software is sensitive to changes in mid-life high-density lipoprotein (HDL) cholesterol level, while posterior hippocampal volume is linked with verbal episodic memory performance in elderly women.

CONCLUSION

These findings support the use of Freesurfer automated segmentation measures for large datasets as being highly correlated with the manual tracing method. In addition, our results suggest intervention strategies that target mid-life HDL cholesterol level in women.

Comparative Utility of Manual versus Automated Segmentation of Hippocampus and Entorhinal Cortex Volumes in a Memory Clinic Sample

Structural neuroimaging is a useful non-invasive biomarker commonly employed to evaluate the integrity of mesial temporal lobe structures that are typically compromised in Alzheimer's disease. Advances in quantitative neuroimaging have permitted the development of automated segmentation protocols (e.g., FreeSurfer) with significantly increased efficiency compared to earlier manual techniques. While these protocols have been found to be suitable for large-scale, multi-site research studies, we were interested in assessing the practical utility and reliability of automated FreeSurfer protocols compared to manual volumetry on routinely acquired clinical scans. Independent validation studies with newer automated segmentation protocols are scarce. Two FreeSurfer protocols for each of two regions of interest-the hippocampus and entorhinal cortex-were compared against manual volumetry. High reliability and agreement was found between FreeSurfer and manual hippocampal protocols, however, there was lower reliability and agreement between FreeSurfer and manual entorhinal protocols. Although based on a the relatively small sample of subjects drawn from a memory clinic (n = 27), our study findings suggest further refinements to improve measurement error and most accurately depict true regional brain volumes using automated segmentation protocols are required, especially for non-hippocampal mesial temporal structures, to achieve maximal utility for routine clinical evaluations.

Neuroanatomical Risk Factors for Posttraumatic Stress Disorder in Recent Trauma Survivors

BACKGROUND

Low hippocampal volume could serve as an early risk factor for posttraumatic stress disorder (PTSD) in interaction with other brain anomalies of developmental origin. One such anomaly may well be the presence of a large cavum septum pellucidum (CSP), which has been loosely associated with PTSD. We performed a longitudinal prospective study of recent trauma survivors. We hypothesized that at 1 month after trauma exposure the relation between hippocampal volume and PTSD symptom severity will be moderated by CSP volume, and that this early interaction will account for persistent PTSD symptoms at subsequent time points.

METHODS

One hundred seventy-one adults (87 women, average age 34.22 years [range, 18-65 years of age]) who were admitted to a general hospital's emergency department after a traumatic event underwent clinical assessment and structural magnetic resonance imaging within 1 month after trauma. Follow-up clinical evaluations were conducted at 6 (n = 97) and 14 (n = 78) months after trauma. Hippocampal and CSP volumes were measured automatically by FreeSurfer software and verified manually by a neuroradiologist (D.N.).

RESULTS

At 1 month after trauma, CSP volume significantly moderated the relation between hippocampal volume and PTSD severity (p = .026), and this interaction further predicted symptom severity at 14 months posttrauma (p = .018). Specifically, individuals with a smaller hippocampus and larger CSP at 1 month posttrauma showed more severe symptoms at 1 and 14 months after trauma exposure.

CONCLUSIONS

Our study provides evidence for an early neuroanatomical risk factors for PTSD, which could also predict the progression of the disorder in the year after trauma exposure. Such a simple-to-acquire neuroanatomical signature for PTSD could guide early management as well as long-term monitoring.

Hippocampal subfields and visuospatial associative memory across stages of schizophrenia-spectrum disorder

BACKGROUND

While previous studies have identified relationships between hippocampal volumes and memory performance in schizophrenia, these relationships are not apparent in healthy individuals. Further, few studies have examined the role of hippocampal subfields in illness-related memory deficits, and no study has examined potential differences across varying illness stages. The current study aimed to investigate whether individuals with early and established psychosis exhibited differential relationships between visuospatial associative memory and hippocampal subfield volumes.

METHODS

Measurements of visuospatial associative memory performance and grey matter volume were obtained from 52 individuals with a chronic schizophrenia-spectrum disorder, 28 youth with recent-onset psychosis, 52 older healthy controls, and 28 younger healthy controls.

RESULTS

Both chronic and recent-onset patients had impaired visuospatial associative memory performance, however, only chronic patients showed hippocampal subfield volume loss. Both chronic and recent-onset patients demonstrated relationships between visuospatial associative memory performance and hippocampal subfield volumes in the CA4/dentate gyrus and the stratum that were not observed in older healthy controls. There were no group by volume interactions when chronic and recent-onset patients were compared.

CONCLUSIONS

The current study extends the findings of previous studies by identifying particular hippocampal subfields, including the hippocampal stratum layers and the dentate gyrus, that appear to be related to visuospatial associative memory ability in individuals with both chronic and first-episode psychosis.

Testing a deep convolutional neural network for automated hippocampus segmentation in a longitudinal sample of healthy participants

Subtle changes in hippocampal volumes may occur during both physiological and pathophysiological processes in the human brain. Assessing hippocampal volumes manually is a time-consuming procedure, however, creating a need for automated segmentation methods that are both fast and reliable over time. Segmentation algorithms that employ deep convolutional neural networks (CNN) have emerged as a promising solution for large longitudinal neuroimaging studies. However, for these novel algorithms to be useful in clinical studies, the accuracy and reproducibility should be established on independent datasets. Here, we evaluate the performance of a CNN-based hippocampal segmentation algorithm that was developed by Thyreau and colleagues - Hippodeep. We compared its segmentation outputs to manual segmentation and FreeSurfer 6.0 in a sample of 200 healthy participants scanned repeatedly at seven sites across Canada, as part of the Canadian Biomarker Integration Network in Depression consortium. The algorithm demonstrated high levels of stability and reproducibility of volumetric measures across all time points compared to the other two techniques. Although more rigorous testing in clinical populations is necessary, this approach holds promise as a viable option for tracking volumetric changes in longitudinal neuroimaging studies.

FAst Segmentation Through SURface Fairing (FASTSURF): A novel semi-automatic hippocampus segmentation method

Objective

The objective is to present a proof-of-concept of a semi-automatic method to reduce hippocampus segmentation time on magnetic resonance images (MRI).

Materials and methods

FAst Segmentation Through SURface Fairing (FASTSURF) is based on a surface fairing technique which reconstructs the hippocampus from sparse delineations. To validate FASTSURF, simulations were performed in which sparse delineations extracted from full manual segmentations served as input. On three different datasets with different diagnostic groups, FASTSURF hippocampi were compared to the original segmentations using Jaccard overlap indices and percentage volume differences (PVD). In one data set for which back-to-back scans were available, unbiased estimates of overlap and PVD were obtained. Using longitudinal scans, we compared hippocampal atrophy rates measured by manual, FASTSURF and two automatic segmentations (FreeSurfer and FSL-FIRST).

Results

With only seven input contours, FASTSURF yielded mean Jaccard indices ranging from 72(±4.3)% to 83(±2.6)% and PVDs ranging from 0.02(±2.40)% to 3.2(±3.40)% across the three datasets. Slightly poorer results were obtained for the unbiased analysis, but the performance was still considerably better than both tested automatic methods with only five contours.

Conclusions

FASTSURF segmentations have high accuracy and require only a fraction of the delineation effort of fully manual segmentation. Atrophy rate quantification based on completely manual segmentation is well reproduced by FASTSURF. Therefore, FASTSURF is a promising tool to be implemented in clinical workflow, provided a future prospective validation confirms our findings.

Volume of the Human Hippocampus and Clinical Response Following Electroconvulsive Therapy

BACKGROUND

Hippocampal enlargements are commonly reported after electroconvulsive therapy (ECT). To clarify mechanisms, we examined if ECT-induced hippocampal volume change relates to dose (number of ECT sessions and electrode placement) and acts as a biomarker of clinical outcome.

METHODS

Longitudinal neuroimaging and clinical data from 10 independent sites participating in the Global ECT-Magnetic Resonance Imaging Research Collaboration (GEMRIC) were obtained for mega-analysis. Hippocampal volumes were extracted from structural magnetic resonance images, acquired before and after patients (n = 281) experiencing a major depressive episode completed an ECT treatment series using right unilateral and bilateral stimulation. Untreated nondepressed control subjects (n = 95) were scanned twice.

RESULTS

The linear component of hippocampal volume change was 0.28% (SE 0.08) per ECT session (p < .001). Volume change varied by electrode placement in the left hippocampus (bilateral, 3.3 ± 2.2%, d = 1.5; right unilateral, 1.6 ± 2.1%, d = 0.8; p < .0001) but not the right hippocampus (bilateral, 3.0 ± 1.7%, d = 1.8; right unilateral, 2.7 ± 2.0%, d = 1.4; p = .36). Volume change for electrode placement per ECT session varied similarly by hemisphere. Individuals with greater treatment-related volume increases had poorer outcomes (Montgomery-Åsberg Depression Rating Scale change -1.0 [SE 0.35], per 1% volume increase, p = .005), although the effects were not significant after controlling for ECT number (slope -0.69 [SE 0.38], p = .069).

CONCLUSIONS

The number of ECT sessions and electrode placement impacts the extent and laterality of hippocampal enlargement, but volume change is not positively associated with clinical outcome. The results suggest that the high efficacy of ECT is not explained by hippocampal enlargement, which alone might not serve as a viable biomarker for treatment outcome.

Amygdala subnuclei and healthy cognitive aging

Amygdala is a group of nuclei involved in the neural circuits of fear, reward learning, and stress. The main goal of this magnetic resonance imaging (MRI) study was to investigate the relationship between age and the amygdala subnuclei volumes in a large cohort of healthy individuals. Our second goal was to determine effects of the apolipoprotein E (APOE) and brain-derived neurotrophic factor (BDNF) polymorphisms on the amygdala structure. One hundred and twenty-six healthy participants (18-85 years old) were recruited for this study. MRI datasets were acquired on a 4.7 T system. Amygdala was manually segmented into five major subdivisions (lateral, basal, accessory basal nuclei, and cortical, and centromedial groups). The BDNF (methionine and homozygous valine) and APOE genotypes (ε2, homozygous ε3, and ε4) were obtained using single nucleotide polymorphisms. We found significant nonlinear negative associations between age and the total amygdala and its lateral, basal, and accessory basal nuclei volumes, while the cortical amygdala showed a trend. These age-related associations were found only in males but not in females. Centromedial amygdala did not show any relationship with age. We did not observe any statistically significant effects of APOE and BDNF polymorphisms on the amygdala subnuclei volumes. In contrast to APOE ε2 allele carriers, both older APOE ε4 and ε3 allele carriers had smaller lateral, basal, accessory basal nuclei volumes compared to their younger counterparts. This study indicates that amygdala subnuclei might be nonuniformly affected by aging and that age-related association might be gender specific.

Longitudinal Assessment of Hippocampal Atrophy in Midlife and Early Old Age: Contrasting Manual Tracing and Semi-automated Segmentation (FreeSurfer)

It is important to have accurate estimates of normal age-related brain structure changes and to understand how the choice of measurement technique may bias those estimates. We compared longitudinal change in hippocampal volume, laterality and atrophy measured by manual tracing and FreeSurfer (version 5.3) in middle age (n = 244, 47.2[1.4] years) and older age (n = 199, 67.0[1.4] years) individuals over 8 years. The proportion of overlap (Dice coefficient) between the segmented hippocampi was calculated and we hypothesised that the proportion of overlap would be higher for older individuals as a consequence of higher atrophy. Hippocampal volumes produced by FreeSurfer were larger than manually traced volumes. Both methods produced a left less than right volume laterality difference. Over time this laterality difference increased for manual tracing and decreased for FreeSurfer leading to laterality differences in left and right estimated atrophy rates. The overlap proportion between methods was not significantly different for older individuals, but was greater for the right hippocampus. Estimated middle age annualised atrophy rates were - 0.39(1.0) left, 0.07(1.01) right, - 0.17(0.88) total for manual tracing and - 0.15(0.69) left, - 0.20(0.63) right, - 0.18(0.57) total for FreeSurfer. Older age atrophy rates were - 0.43(1.32) left, - 0.15(1.41) right, - 0.30 (1.23) total for manual tracing and - 0.34(0.79) left, - 0.68(0.78) right, - 0.51(0.65) total for FreeSurfer. FreeSurfer reliably segments the hippocampus producing atrophy rates that are comparable to manual tracing with some biases that need to be considered in study design. FreeSurfer is suited for use in large longitudinal studies where it is not cost effective to use manual tracing.

A comparison of manual tracing and FreeSurfer for estimating hippocampal volume over the adult lifespan

MRI has become an indispensable tool for brain volumetric studies, with the hippocampus an important region of interest. Automation of the MRI segmentation process has helped advance the field by facilitating the volumetric analysis of larger cohorts and more studies. FreeSurfer has emerged as the de facto standard tool for these analyses, but studies validating its output are all based on older versions. To characterize FreeSurfer's validity, we compare several versions of FreeSurfer software with traditional hand-tracing. Using MRI images of 262 males and 402 females aged 38 to 84, we directly compare estimates of hippocampal volume from multiple versions of FreeSurfer, its hippocampal subfield routines, and our manual tracing protocol. We then use those estimates to assess asymmetry and atrophy, comparing performance of different estimators with each other and with brain atrophy measures. FreeSurfer consistently reports larger volumes than manual tracing. This difference is smaller in larger hippocampi or older people, with these biases weaker in version 6.0.0 than prior versions. All methods tested agree qualitatively on rightward asymmetry and increasing atrophy in older people. FreeSurfer saves time and money, and approximates the same atrophy measures as manual tracing, but it introduces biases that could require statistical adjustments in some studies.

Coupled changes in hippocampal structure and cognitive ability in later life

Introduction

The hippocampus plays an important role in cognitive abilities which often decline with advancing age.

Methods

In a longitudinal study of community-dwelling adults, we investigated whether there were coupled changes in hippocampal structure and verbal memory, working memory, and processing speed between the ages of 73 (N = 655) and 76 years (N = 469). Hippocampal structure was indexed by hippocampal volume, hippocampal volume as a percentage of intracranial volume (H_ICV), fractional anisotropy (FA), mean diffusivity (MD), and longitudinal relaxation time (T1).

Results

Mean levels of hippocampal volume, H_ICV, FA, T1, and all three cognitive abilities domains decreased, whereas MD increased, from age 73 to 76. At baseline, higher hippocampal volume was associated with better working memory and verbal memory, but none of these correlations survived correction for multiple comparisons. Higher FA, lower MD, and lower T1 at baseline were associated with better cognitive abilities in all three domains; only the correlation between baseline hippocampal MD and T1, and change in the three cognitive domains, survived correction for multiple comparisons. Individuals with higher hippocampal MD at age 73 experienced a greater decline in all three cognitive abilities between ages 73 and 76. However, no significant associations with changes in cognitive abilities were found with hippocampal volume, FA, and T1 measures at baseline. Similarly, no significant associations were found between cognitive abilities at age 73 and changes in the hippocampal MRI biomarkers between ages 73 and 76.

Conclusion

Our results provide evidence to better understand how the hippocampus ages in healthy adults in relation to the cognitive domains in which it is involved, suggesting that better hippocampal MD at age 73 predicts less relative decline in three important cognitive domains across the next 3 years. It can potentially assist in diagnosing early stages of aging-related neuropathologies, because in some cases, accelerated decline could predict pathologies.

Brain Atrophy Estimated from Structural Magnetic Resonance Imaging as a Marker of Large-Scale Network-Based Neurodegeneration in Aging and Stroke

Brain atrophy is a normal part of healthy aging, and stroke appears to have neurodegenerative effects, accelerating this atrophy to pathological levels. The distributed pattern of atrophy in healthy aging suggests that large-scale brain networks may be involved. At the same time, the network wide effects of stroke are beginning to be appreciated. There is now widespread use of network methods to understand the brain in terms of coordinated brain activity or white matter connectivity. Examining brain morphology on a network level presents a powerful method of understanding brain structure and has been successfully applied to charting the course of brain development. This review will introduce recent advances in structural magnetic resonance imaging (MRI) acquisition and analyses that have allowed for reliable and reproducible estimates of atrophy in large-scale brain networks in aging and after stroke. These methods are currently underutilized despite their ease of acquisition and potential to clarify the progression of brain atrophy as a normal part of healthy aging and in the context of stroke. Understanding brain atrophy at the network level may be key to clarifying healthy aging processes and the pathway to neurodegeneration after stroke.

Prenatal methamphetamine exposure is associated with reduced subcortical volumes in neonates

OBJECTIVES

Prenatal exposure to methamphetamine is associated with a range of neuropsychological, behavioural and cognitive deficits. A small number of imaging studies suggests that these may be mediated by neurostructural changes, including reduced volumes of specific brain regions. This study investigated potential volumetric changes in the brains of neonates with prenatal methamphetamine exposure. To our knowledge no previous studies have examined methamphetamine effects on regional brain volumes at this age.

STUDY DESIGN

Mothers were recruited antenatally and interviewed regarding methamphetamine use during pregnancy. Mothers in the exposure group reported using methamphetamine≥twice/month during pregnancy; control infants had no exposure to methamphetamine or other drugs and minimal exposure to alcohol. MRI scans were performed in the first postnatal month, following which anatomical images were processed using FreeSurfer. Subcortical and cerebellar regions were manually segmented and their volumes determined using FreeView. Pearson correlations were used to analyse potential associations between methamphetamine exposure and regional volumes. The associations between methamphetamine exposure and regional volumes were then examined adjusting for potential confounding variables.

RESULTS

Methamphetamine exposure was associated with reduced left and right caudate and thalamus volumes. The association in the right caudate remained significant following adjustment for potential confounding variables.

CONCLUSIONS

Our findings showing reduced caudate and thalamus volumes in neonates with prenatal methamphetamine exposure are consistent with previous findings in older exposed children, and demonstrate that these changes are already detectable in neonates. Continuing research is warranted to examine whether reduced subcortical volumes are predictive of cognitive, behavioural and affective impairment in older children.

The anteroposterior and primary-to-posterior limbic ratios as MRI-derived volumetric markers of Alzheimer's disease

BACKGROUND/AIMS

Alzheimer's disease (AD) shows a characteristic pattern of brain atrophy, with predominant involvement of posterior limbic structures, and relative preservation of rostral limbic and primary cortical regions. We aimed to investigate the diagnostic utility of two gray matter volume ratios based on this pattern, and to develop a fully automated method to calculate them from unprocessed MRI files.

PATIENTS AND METHODS

Cross-sectional study of 118 subjects from the ADNI database, including normal controls and patients with mild cognitive impairment (MCI) and AD. Clinical variables and 3T T1-weighted MRI files were analyzed. Regional gray matter and total intracranial volumes were calculated with a shell script (gm_extractor) based on FSL. Anteroposterior and primary-to-posterior limbic ratios (APL and PPL) were calculated from these values. Diagnostic utility of variables was tested in logistic regression models using Bayesian model averaging for variable selection. External validity was evaluated with bootstrap sampling and a test set of 60 subjects.

RESULTS

gm_extractor showed high test-retest reliability and high concurrent validity with FSL's FIRST. Volumetric measurements agreed with the expected anatomical pattern associated with AD. APL and PPL ratios were significantly different between groups, and were selected instead of hippocampal and entorhinal volumes to differentiate normal from MCI or cognitively impaired (MCI plus AD) subjects.

CONCLUSION

APL and PPL ratios may be useful components of models aimed to differentiate normal subjects from patients with MCI or AD. These values, and other gray matter volumes, may be reliably calculated with gm_extractor.

Reference standard space hippocampus labels according to the European Alzheimer's Disease Consortium-Alzheimer's Disease Neuroimaging Initiative harmonized protocol: Utility in automated volumetry

INTRODUCTION

A harmonized protocol (HarP) for manual hippocampal segmentation on magnetic resonance imaging (MRI) has recently been developed by an international European Alzheimer's Disease Consortium-Alzheimer's Disease Neuroimaging Initiative project. We aimed at providing consensual certified HarP hippocampal labels in Montreal Neurological Institute (MNI) standard space to serve as reference in automated image analyses.

METHODS

Manual HarP tracings on the high-resolution MNI152 standard space template of four expert certified HarP tracers were combined to obtain consensual bilateral hippocampus labels. Utility and validity of these reference labels is demonstrated in a simple atlas-based morphometry approach for automated calculation of HarP-compliant hippocampal volumes within SPM software.

RESULTS

Individual tracings showed very high agreement among the four expert tracers (pairwise Jaccard indices 0.82-0.87). Automatically calculated hippocampal volumes were highly correlated (r_L/R = 0.89/0.91) with gold standard volumes in the HarP benchmark data set (N = 135 MRIs), with a mean volume difference of 9% (standard deviation 7%).

CONCLUSION

The consensual HarP hippocampus labels in the MNI152 template can serve as a reference standard for automated image analyses involving MNI standard space normalization.

Regional analysis of volumes and reproducibilities of automatic and manual hippocampal segmentations

PURPOSE

Precise and reproducible hippocampus outlining is important to quantify hippocampal atrophy caused by neurodegenerative diseases and to spare the hippocampus in whole brain radiation therapy when performing prophylactic cranial irradiation or treating brain metastases. This study aimed to quantify systematic differences between methods by comparing regional volume and outline reproducibility of manual, FSL-FIRST and FreeSurfer hippocampus segmentations.

MATERIALS AND METHODS

This study used a dataset from ADNI (Alzheimer's Disease Neuroimaging Initiative), including 20 healthy controls, 40 patients with mild cognitive impairment (MCI), and 20 patients with Alzheimer's disease (AD). For each subject back-to-back (BTB) T1-weighted 3D MPRAGE images were acquired at time-point baseline (BL) and 12 months later (M12). Hippocampi segmentations of all methods were converted into triangulated meshes, regional volumes were extracted and regional Jaccard indices were computed between the hippocampi meshes of paired BTB scans to evaluate reproducibility. Regional volumes and Jaccard indices were modelled as a function of group (G), method (M), hemisphere (H), time-point (T), region (R) and interactions.

RESULTS

For the volume data the model selection procedure yielded the following significant main effects G, M, H, T and R and interaction effects G-R and M-R. The same model was found for the BTB scans. For all methods volumes reduces with the severity of disease. Significant fixed effects for the regional Jaccard index data were M, R and the interaction M-R. For all methods the middle region was most reproducible, independent of diagnostic group. FSL-FIRST was most and FreeSurfer least reproducible.

DISCUSSION/CONCLUSION

A novel method to perform detailed analysis of subtle differences in hippocampus segmentation is proposed. The method showed that hippocampal segmentation reproducibility was best for FSL-FIRST and worst for Freesurfer. We also found systematic regional differences in hippocampal segmentation between different methods reinforcing the need of adopting harmonized protocols.

Whole-Brain High-Resolution Structural Connectome: Inter-Subject Validation and Application to the Anatomical Segmentation of the Striatum

The present study describes extraction of high-resolution structural connectome (HRSC) in 99 healthy subjects, acquired and made available by the Human Connectome Project. Single subject connectomes were then registered to the common surface space to allow assessment of inter-individual reproducibility of this novel technique using a leave-one-out approach. The anatomic relevance of the surface-based connectome was examined via a clustering algorithm, which identified anatomic subdivisions within the striatum. The connectivity of these striatal subdivisions were then mapped on the cortical and other subcortical surfaces. Findings demonstrate that HRSC analysis is robust across individuals and accurately models the actual underlying brain networks related to the striatum. This suggests that this method has the potential to model and characterize the healthy whole-brain structural network at high anatomic resolution.

MRI-Based Classification Models in Prediction of Mild Cognitive Impairment and Dementia in Late-Life Depression

Objective: Late-life depression (LLD) is associated with development of different types of dementia. Identification of LLD patients, who will develop cognitive decline, i.e., the early stage of dementia would help to implement interventions earlier. The purpose of this study was to assess whether structural brain magnetic resonance imaging (MRI) in LLD patients can predict mild cognitive impairment (MCI) or dementia 1 year prior to the diagnosis.

Methods: LLD patients underwent brain MRI at baseline and repeated clinical assessment after 1-year. Structural brain measurements were obtained using Freesurfer software (v. 5.1) from the T1W brain MRI images. MRI-based Random Forest classifier was used to discriminate between LLD who developed MCI or dementia after 1-year follow-up and cognitively stable LLD. Additionally, a previously established Random Forest model trained on 185 patients with Alzheimer’s disease (AD) vs. 225 cognitively normal elderly from the Alzheimer’s disease Neuroimaging Initiative was tested on the LLD data set (ADNI model).

Results: MCI and dementia diagnoses were predicted in LLD patients with 76%/68%/84% accuracy/sensitivity/specificity. Adding the baseline Mini-Mental State Examination (MMSE) scores to the models improved accuracy/sensitivity/specificity to 81%/75%/86%. The best model predicted MCI status alone using MRI and baseline MMSE scores with accuracy/sensitivity/specificity of 89%/85%/90%. The most important region for all the models was right ventral diencephalon, including hypothalamus. Its volume correlated negatively with the number of depressive episodes. ADNI model trained on AD vs. Controls using SV could predict MCI-DEM patients with 67% accuracy.

Conclusion: LDD patients developing MCI and dementia can be discriminated from LLD patients remaining cognitively stable with good accuracy based on baseline structural MRI alone. Baseline MMSE score improves prediction accuracy. Ventral diencephalon, including the hypothalamus might play an important role in preservation of cognitive functions in LLD.

A New MRI-Based Pediatric Subcortical Segmentation Technique (PSST)

Volumetric and morphometric neuroimaging studies of the basal ganglia and thalamus in pediatric populations have utilized existing automated segmentation tools including FIRST (Functional Magnetic Resonance Imaging of the Brain's Integrated Registration and Segmentation Tool) and FreeSurfer. These segmentation packages, however, are mostly based on adult training data. Given that there are marked differences between the pediatric and adult brain, it is likely an age-specific segmentation technique will produce more accurate segmentation results. In this study, we describe a new automated segmentation technique for analysis of 7-year-old basal ganglia and thalamus, called Pediatric Subcortical Segmentation Technique (PSST). PSST consists of a probabilistic 7-year-old subcortical gray matter atlas (accumbens, caudate, pallidum, putamen and thalamus) combined with a customized segmentation pipeline using existing tools: ANTs (Advanced Normalization Tools) and SPM (Statistical Parametric Mapping). The segmentation accuracy of PSST in 7-year-old data was compared against FIRST and FreeSurfer, relative to manual segmentation as the ground truth, utilizing spatial overlap (Dice's coefficient), volume correlation (intraclass correlation coefficient, ICC) and limits of agreement (Bland-Altman plots). PSST achieved spatial overlap scores ≥90% and ICC scores ≥0.77 when compared with manual segmentation, for all structures except the accumbens. Compared with FIRST and FreeSurfer, PSST showed higher spatial overlap (p FDR  < 0.05) and ICC scores, with less volumetric bias according to Bland-Altman plots. PSST is a customized segmentation pipeline with an age-specific atlas that accurately segments typical and atypical basal ganglia and thalami at age 7 years, and has the potential to be applied to other pediatric datasets.

Motion-related artifacts in structural brain images revealed with independent estimates of in-scanner head motion

Motion‐contaminated T1‐weighted (T1w) magnetic resonance imaging (MRI) results in misestimates of brain structure. Because conventional T1w scans are not collected with direct measures of head motion, a practical alternative is needed to identify potential motion‐induced bias in measures of brain anatomy. Head movements during functional MRI (fMRI) scanning of 266 healthy adults (20–89 years) were analyzed to reveal stable features of in‐scanner head motion. The magnitude of head motion increased with age and exhibited within‐participant stability across different fMRI scans. fMRI head motion was then related to measurements of both quality control (QC) and brain anatomy derived from a T1w structural image from the same scan session. A procedure was adopted to “flag” individuals exhibiting excessive head movement during fMRI or poor T1w quality rating. The flagging procedure reliably reduced the influence of head motion on estimates of gray matter thickness across the cortical surface. Moreover, T1w images from flagged participants exhibited reduced estimates of gray matter thickness and volume in comparison to age‐ and gender‐matched samples, resulting in inflated effect sizes in the relationships between regional anatomical measures and age. Gray matter thickness differences were noted in numerous regions previously reported to undergo prominent atrophy with age. Recommendations are provided for mitigating this potential confound, and highlight how the procedure may lead to more accurate measurement and comparison of anatomical features. Hum Brain Mapp 38:472–492, 2017. © 2016 Wiley Periodicals, Inc.

Associations between Family Adversity and Brain Volume in Adolescence: Manual vs. Automated Brain Segmentation Yields Different Results

Associations between brain structure and early adversity have been inconsistent in the literature. These inconsistencies may be partially due to methodological differences. Different methods of brain segmentation may produce different results, obscuring the relationship between early adversity and brain volume. Moreover, adolescence is a time of significant brain growth and certain brain areas have distinct rates of development, which may compromise the accuracy of automated segmentation approaches. In the current study, 23 adolescents participated in two waves of a longitudinal study. Family aggression was measured when the youths were 12 years old, and structural scans were acquired an average of 4 years later. Bilateral amygdalae and hippocampi were segmented using three different methods (manual tracing, FSL, and NeuroQuant). The segmentation estimates were compared, and linear regressions were run to assess the relationship between early family aggression exposure and all three volume segmentation estimates. Manual tracing results showed a positive relationship between family aggression and right amygdala volume, whereas FSL segmentation showed negative relationships between family aggression and both the left and right hippocampi. However, results indicate poor overlap between methods, and different associations were found between early family aggression exposure and brain volume depending on the segmentation method used.

Reproducibility of hippocampal atrophy rates measured with manual, FreeSurfer, AdaBoost, FSL/FIRST and the MAPS-HBSI methods in Alzheimer's disease

The purpose of this study is to assess the reproducibility of hippocampal atrophy rate measurements of commonly used fully-automated algorithms in Alzheimer disease (AD). The reproducibility of hippocampal atrophy rate for FSL/FIRST, AdaBoost, FreeSurfer, MAPS independently and MAPS combined with the boundary shift integral (MAPS-HBSI) were calculated. Back-to-back (BTB) 3D T1-weighted MPRAGE MRI from the Alzheimer's Disease Neuroimaging Initiative (ADNI1) study at baseline and year one were used. Analysis on 3 groups of subjects was performed - 562 subjects at 1.5T, a 75 subject group that also had manual segmentation and 111 subjects at 3T. A simple and novel statistical test based on the binomial distribution was used that handled outlying data points robustly. Median hippocampal atrophy rates were -1.1%/year for healthy controls, -3.0%/year for mildly cognitively impaired and -5.1%/year for AD subjects. The best reproducibility was observed for MAPS-HBSI (1.3%), while the other methods tested had reproducibilities at least 50% higher at 1.5T and 3T which was statistically significant. For a clinical trial, MAPS-HBSI should require less than half the subjects of the other methods tested. All methods had good accuracy versus manual segmentation. The MAPS-HBSI method has substantially better reproducibility than the other methods considered.

Robust Machine Learning-Based Correction on Automatic Segmentation of the Cerebellum and Brainstem

Automated segmentation is a useful method for studying large brain structures such as the cerebellum and brainstem. However, automated segmentation may lead to inaccuracy and/or undesirable boundary. The goal of the present study was to investigate whether SegAdapter, a machine learning-based method, is useful for automatically correcting large segmentation errors and disagreement in anatomical definition. We further assessed the robustness of the method in handling size of training set, differences in head coil usage, and amount of brain atrophy. High resolution T1-weighted images were acquired from 30 healthy controls scanned with either an 8-channel or 32-channel head coil. Ten patients, who suffered from brain atrophy because of fragile X-associated tremor/ataxia syndrome, were scanned using the 32-channel head coil. The initial segmentations of the cerebellum and brainstem were generated automatically using Freesurfer. Subsequently, Freesurfer's segmentations were both manually corrected to serve as the gold standard and automatically corrected by SegAdapter. Using only 5 scans in the training set, spatial overlap with manual segmentation in Dice coefficient improved significantly from 0.956 (for Freesurfer segmentation) to 0.978 (for SegAdapter-corrected segmentation) for the cerebellum and from 0.821 to 0.954 for the brainstem. Reducing the training set size to 2 scans only decreased the Dice coefficient ≤0.002 for the cerebellum and ≤ 0.005 for the brainstem compared to the use of training set size of 5 scans in corrective learning. The method was also robust in handling differences between the training set and the test set in head coil usage and the amount of brain atrophy, which reduced spatial overlap only by <0.01. These results suggest that the combination of automated segmentation and corrective learning provides a valuable method for accurate and efficient segmentation of the cerebellum and brainstem, particularly in large-scale neuroimaging studies, and potentially for segmenting other neural regions as well.

Lack of an association of BDNF Val66Met polymorphism and plasma BDNF with hippocampal volume and memory

Brain-derived neurotrophic factor (BDNF) has been shown to be important for neuronal survival and synaptic plasticity in the hippocampus in nonhuman animals. The Val66Met polymorphism in the BDNF gene, involving a valine (Val) to methionine (Met) substitution at codon 66, has been associated with lower BDNF secretion in vitro. However, there have been mixed results regarding associations between either circulating BDNF or the BDNF Val66Met polymorphism with hippocampal volume and memory in humans. The current study examined the association of BDNF genotype and plasma BDNF with hippocampal volume and memory in two large independent cohorts of middle-aged and older adults (both cognitively normal and early-stage dementia). Sample sizes ranged from 123 to 649. Measures of the BDNF genotype, plasma BDNF, MRI-based hippocampal volume, and memory performance were obtained from the Knight Alzheimer Disease Research Center (ADRC) and the Alzheimer's Disease Neuroimaging Initiative (ADNI). There were no significant differences between BDNF Met+ and Met- groups on either hippocampal volume or memory in either cohort. In addition, plasma BDNF was not significantly associated with either hippocampal volume or memory in either cohort. Neither age, cognitive status, nor gender moderated any of the relationships. Overall, current findings suggest that BDNF genotype and plasma BDNF may not be robust predictors for variance in hippocampal volume and memory in middle age and older adult cohorts.

Fornix and medial temporal lobe lesions lead to comparable deficits in complex visual perception

Recent research dealing with the structures of the medial temporal lobe (MTL) has shifted away from exclusively investigating memory-related processes and has repeatedly incorporated the investigation of complex visual perception. Several studies have demonstrated that higher level visual tasks can recruit structures like the hippocampus and perirhinal cortex in order to successfully perform complex visual discriminations, leading to a perceptual-mnemonic or representational view of the medial temporal lobe. The current study employed a complex visual discrimination paradigm in two patients suffering from brain lesions with differing locations and origin. Both patients, one with extensive medial temporal lobe lesions (VG) and one with a small lesion of the anterior fornix (HJK), were impaired in complex discriminations while showing otherwise mostly intact cognitive functions. The current data confirmed previous results while also extending the perceptual-mnemonic theory of the MTL to the main output structure of the hippocampus, the fornix.

Subcortical volume changes in dementia with Lewy bodies and Alzheimer's disease. A comparison with healthy aging

BACKGROUND

Differentiating Alzheimer's disease (AD) and dementia with Lewy bodies (DLB), two of the commonest forms of dementia in older age, remains a diagnostic challenge. To assist with better understanding of the differences between the conditions during life, we assessed limbic and subcortical brain volumes in AD, DLB, and healthy older individuals using magnetic resonance imaging (MRI), with the hypothesis that when compared with controls, subcortical volumes would be reduced to a greater extent in DLB than in AD.

METHODS

One hundred participants (35 healthy controls, 32 AD, and 33 DLB) underwent 3 Tesla T1 weighted MR scanning. Volumes were automatically segmented for each participant using FreeSurfer, then expressed as a percentage of their total intracranial volumes. Group effects were assessed using multivariate analysis of covariance, controlling for age and gender.

RESULTS

Significant group effects were apparent among subcortical brain volumes (F 28,162 = 4.8, p < 0.001; Wilk's Λ = 0.30, partial η 2 = 0.45), while univariate tests showed differences in all volumetric measures (p AD, p < 0.008).

CONCLUSIONS

For similar levels of dementia severity, DLB appears to have greater involvement of subcortical brain atrophy than AD. Further investigation of the subcortical brain structures in DLB is warranted to fully understand their neurobiological role in this disease.

Hippocampus and amygdala volumes from magnetic resonance images in children: Assessing accuracy of FreeSurfer and FSL against manual segmentation

The volumetric quantification of brain structures is of great interest in pediatric populations because it allows the investigation of different factors influencing neurodevelopment. FreeSurfer and FSL both provide frequently used packages for automatic segmentation of brain structures. In this study, we examined the accuracy and consistency of those two automated protocols relative to manual segmentation, commonly considered as the "gold standard" technique, for estimating hippocampus and amygdala volumes in a sample of preadolescent children aged between 6 to 11 years. The volumes obtained with FreeSurfer and FSL-FIRST were evaluated and compared with manual segmentations with respect to volume difference, spatial agreement and between- and within-method correlations. Results highlighted a tendency for both automated techniques to overestimate hippocampus and amygdala volumes, in comparison to manual segmentation. This was more pronounced when using FreeSurfer than FSL-FIRST and, for both techniques, the overestimation was more marked for the amygdala than the hippocampus. Pearson correlations support moderate associations between manual tracing and FreeSurfer for hippocampus (right r=0.69, p<0.001; left r=0.77, p<0.001) and amygdala (right r=0.61, p<0.001; left r=0.67, p<0.001) volumes. Correlation coefficients between manual segmentation and FSL-FIRST were statistically significant (right hippocampus r=0.59, p<0.001; left hippocampus r=0.51, p<0.001; right amygdala r=0.35, p<0.001; left amygdala r=0.31, p<0.001) but were significantly weaker, for all investigated structures. When computing intraclass correlation coefficients between manual tracing and automatic segmentation, all comparisons, except for left hippocampus volume estimated with FreeSurfer, failed to reach 0.70. When looking at each method separately, correlations between left and right hemispheric volumes showed strong associations between bilateral hippocampus and bilateral amygdala volumes when assessed using manual segmentation or FreeSurfer. These correlations were significantly weaker when volumes were assessed with FSL-FIRST. Finally, Bland-Altman plots suggest that the difference between manual and automatic segmentation might be influenced by the volume of the structure, because smaller volumes were associated with larger volume differences between techniques. These results demonstrate that, at least in a pediatric population, the agreement between amygdala and hippocampus volumes obtained with automated FSL-FIRST and FreeSurfer protocols and those obtained with manual segmentation is not strong. Visual inspection by an informed individual and, if necessary, manual correction of automated segmentation outputs are important to ensure validity of volumetric results and interpretation of related findings.

Quantitative Neuroimaging Software for Clinical Assessment of Hippocampal Volumes on MR Imaging

BACKGROUND

Multiple neurological disorders including Alzheimer's disease (AD), mesial temporal sclerosis, and mild traumatic brain injury manifest with volume loss on brain MRI. Subtle volume loss is particularly seen early in AD. While prior research has demonstrated the value of this additional information from quantitative neuroimaging, very few applications have been approved for clinical use. Here we describe a US FDA cleared software program, NeuroreaderTM, for assessment of clinical hippocampal volume on brain MRI.

OBJECTIVE

To present the validation of hippocampal volumetrics on a clinical software program.

METHOD

Subjects were drawn (n = 99) from the Alzheimer Disease Neuroimaging Initiative study. Volumetric brain MR imaging was acquired in both 1.5 T (n = 59) and 3.0 T (n = 40) scanners in participants with manual hippocampal segmentation. Fully automated hippocampal segmentation and measurement was done using a multiple atlas approach. The Dice Similarity Coefficient (DSC) measured the level of spatial overlap between NeuroreaderTM and gold standard manual segmentation from 0 to 1 with 0 denoting no overlap and 1 representing complete agreement. DSC comparisons between 1.5 T and 3.0 T scanners were done using standard independent samples T-tests.

RESULTS

In the bilateral hippocampus, mean DSC was 0.87 with a range of 0.78-0.91 (right hippocampus) and 0.76-0.91 (left hippocampus). Automated segmentation agreement with manual segmentation was essentially equivalent at 1.5 T (DSC = 0.879) versus 3.0 T (DSC = 0.872).

CONCLUSION

This work provides a description and validation of a software program that can be applied in measuring hippocampal volume, a biomarker that is frequently abnormal in AD and other neurological disorders.

Impact of monoamine-related gene polymorphisms on hippocampal volume in treatment-resistant depression

OBJECTIVE

In major depressive disorder (MDD), single nucleotide polymorphisms (SNPs) in monoaminergic genes may impact disease susceptibility, treatment response, and brain volume. The objective of this study was to examine the effect of such polymorphisms on hippocampal volume in patients with treatment-resistant MDD and healthy controls. Candidate gene risk alleles were hypothesised to be associated with reductions in hippocampal volume.

METHODS

A total of 26 outpatients with treatment-resistant MDD and 27 matched healthy controls underwent magnetic resonance imaging and genotyping for six SNPs in monoaminergic genes [serotonin transporter (SLC6A4), norepinephrine transporter (SLC6A2), serotonin 1A and 2A receptors (HTR1A and HTR2A), catechol-O-methyltransferase (COMT), and brain-derived neurotrophic factor (BDNF)]. Hippocampal volume was estimated using an automated segmentation algorithm (FreeSurfer).

RESULTS

Hippocampal volume did not differ between patients and controls. Within the entire study sample irrespective of diagnosis, C allele-carriers for both the NET-182 T/C [rs2242446] and 5-HT1A-1019C/G [rs6295] polymorphisms had smaller hippocampal volumes relative to other genotypes. For the 5-HTTLPR (rs25531) polymorphism, there was a significant diagnosis by genotype interaction effect on hippocampal volume. Among patients only, homozygosity for the 5-HTTLPR short (S) allele was associated with smaller hippocampal volume. There was no association between the 5-HT2A, COMT, and BDNF SNPs and hippocampal volume.

CONCLUSION

The results indicate that the volume of the hippocampus may be influenced by serotonin- and norepinephrine-related gene polymorphisms. The NET and 5-HT1A polymorphisms appear to have similar effects on hippocampal volume in patients and controls while the 5-HTTLPR polymorphism differentially affects hippocampal volume in the presence of depression.

Semi-automated hippocampal segmentation in people with cognitive impairment using an age appropriate template for registration

BACKGROUND

To evaluate a new semi-automated segmentation method for calculating hippocampal volumes and to compare results with standard software tools in a cohort of people with subjective memory complaints (SMC) and mild cognitive impairment (MCI).

METHODS

Data from 58 participants, 39 with SMC (17 male, 22 female, mean age 72.6) and 19 with MCI (6 male, 13 female, mean age 74.3), were analyzed. For each participant, T1-weighted images were acquired using an MPRAGE sequence on a 3 Tesla MRI system. Hippocampal volumes (left, right, and total) were calculated with a new, age appropriate registration template, based on older people and using the advanced software tool ANTs (Advanced Normalization Tools). The results were compared with manual tracing (seen as the reference standard) and two widely accepted automated software tools (FSL, FreeSurfer).

RESULTS

The hippocampal volumes, calculated by using the age appropriate registration template were significantly (P < 0.05) more accurate (mean volume accuracy more than 90%) than those obtained with FreeSurfer and FSL (both less than 70%). Dice coefficients for the hippocampal segmentations with the new template method (75.3%) were slightly, but significantly (P < 0.05) higher than those from FreeSurfer (72.4%).

CONCLUSION

These results suggest that an age appropriate registration template might be a more accurate alternative to calculate hippocampal volumes when manual segmentation is not feasible.