Comparison of automated brain segmentation using a brain phantom and patients with early Alzheimer's dementia or mild cognitive impairment

Volumetry of Selected Brain Regions-Can We Compare MRI Examinations of Different Manufacturers and Field Strengths?

PURPOSE

Magnetic Resonance Imaging based brain segmentation and volumetry has become an important tool in clinical routine and research. However the impact of the used hardware is only barely investigated. This study aims to assess the influence of scanner manufacturer, field strength and head-coil on volumetry results.

METHODS

10 healthy subjects (27.4 ± 1.71 years) were prospectively examined in a Philips Achieva 1.5T, Philips Ingenia CX 3T, Siemens MAGNETOM Aera 1.5T and Siemens MAGNETOM Vida 3T, the latter equipped with three different head coils, within one day. Brain volumetry of the whole brain, total white and grey matter, the cortical grey matter of the supratentorial lobes as well as regions important for the differentiation of neurodegenerative diseases of the dementia and movement disorder spectrum and the ventricular system was performed using the CE-certified software mdbrain by mediaire (Berlin, Germany). Both raw volumetry results and percentile allocation provided by the software were analysed.

RESULTS

This study reveals significantly different volumetry results for all examined brain regions beside the ventricular system between the different MRI devices but comparable results between the different head coils. When examining the percentile allocation provided by used software, the Intraclass-Correlation-Coefficient (ICC) values were even lower than the raw volume ICC values ranging from poor to excellent correlation.

CONCLUSION

The present study reveals highly relevant results that need to be considered both in clinical routine when analysing follow-up examinations from different scanner types and clinical research, especially when planning longitudinal and/or multicentre studies.

Changes in cortical grey matter volume with Cognitive Orientation to daily Occupational Performance intervention in children with developmental coordination disorder

Introduction

Cognitive Orientation to daily Occupational Performance (CO-OP) is a cognitive-based, task-specific intervention recommended for children with developmental coordination disorder (DCD). We recently showed structural and functional brain changes after CO-OP, including increased cerebellar grey matter. This study aimed to determine whether CO-OP intervention induced changes in cortical grey matter volume in children with DCD, and if these changes were associated with improvements in motor performance and movement quality.

Methods

This study is part of a randomized waitlist-control trial (ClinicalTrials.gov ID: NCT02597751). Children with DCD (N = 78) were randomized to either a treatment or waitlist group and underwent three MRIs over 6 months. The treatment group received intervention (once weekly for 10 weeks) between the first and second scan; the waitlist group received intervention between the second and third scan. Cortical grey matter volume was measured using voxel-based morphometry (VBM). Behavioral outcome measures included the Performance Quality Rating Scale (PQRS) and Bruininks-Oseretsky Test of Motor Proficiency-2 (BOT-2). Of the 78 children, 58 were excluded (mostly due to insufficient data quality), leaving a final N = 20 for analyses. Due to the small sample size, we combined both groups to examine treatment effects. Cortical grey matter volume differences were assessed using a repeated measures ANOVA, controlling for total intracranial volume. Regression analyses examined the relationship of grey matter volume changes to BOT-2 (motor performance) and PQRS (movement quality).

Results

After CO-OP, children had significantly decreased grey matter in the right superior frontal gyrus and middle/posterior cingulate gyri. We found no significant associations of grey matter volume changes with PQRS or BOT-2 scores.

Conclusion

Decreased cortical grey matter volume generally reflects greater brain maturity. Decreases in grey matter volume after CO-OP intervention were in regions associated with self-regulation and motor control, consistent with our other studies. Decreased grey matter volume may be due to focal increases in synaptic pruning, perhaps as a result of strengthening networks in the brain via the repeated learning and actions in therapy. Findings from this study add to the growing body of literature demonstrating positive neuroplastic changes in the brain after CO-OP intervention.

Cortical grey matter volume differences in children with developmental coordination disorder compared to typically developing children

Introduction

The cause of Developmental Coordination Disorder (DCD) is unknown, but neuroimaging evidence suggests that DCD may be related to altered brain development. Children with DCD show less structural and functional connectivity compared to typically developing (TD) children, but few studies have examined cortical volume in children with DCD. The purpose of this study was to investigate cortical grey matter volume using voxel-based morphometry (VBM) in children with DCD compared to TD children.

Methods

This cross-sectional study was part of a larger randomized-controlled trial (ClinicalTrials.gov ID: NCT02597751) that involved various MRI scans of children with/without DCD. This paper focuses on the anatomical scans, performing VBM of cortical grey matter volume in 30 children with DCD and 12 TD children. Preprocessing and VBM data analysis were conducted using the Computational Anatomy Tool Box-12 and a study-specific brain template. Differences between DCD and TD groups were assessed using a one-way ANOVA, controlling for total intracranial volume. Regression analyses examined if motor and/or attentional difficulties predicted grey matter volume. We used threshold-free cluster enhancement (5,000 permutations) and set an alpha level of 0.05. Due to the small sample size, we did not correct for multiple comparisons.

Results

Compared to the TD group, children with DCD had significantly greater grey matter in the left superior frontal gyrus. Lower motor scores (meaning greater impairment) were related to greater grey matter volume in left superior frontal gyrus, frontal pole, and right middle frontal gyrus. Greater grey matter volume was also significantly correlated with higher scores on the Conners 3 ADHD Index in the left superior frontal gyrus, superior parietal lobe, and precuneus. These results indicate that greater grey matter volume in these regions is associated with poorer motor and attentional skills.

Discussion

Greater grey matter volume in the left superior frontal gyrus in children with DCD may be a result of delayed or absent healthy cortical thinning, potentially due to altered synaptic pruning as seen in other neurodevelopmental disorders. These findings provide further support for the hypothesis that DCD is related to altered brain development.

Predictive value of spinal CSF volume in the preoperative assessment of patients with idiopathic normal-pressure hydrocephalus

Introduction

The pathophysiology, diagnosis, and management of idiopathic normal pressure hydrocephalus (iNPH) remain unclear. Although some prognostic tests recommended in iNPH guidelines should have high sensitivity and high predictive value, there is often no positive clinical response to surgical treatment.

Materials and methods

In our study, 19 patients with clinical and neuroradiological signs of iNPH were selected for preoperative evaluation and possible further surgical treatment according to the guidelines. MR volumetry of the intracranial and spinal space was performed. Patients were exposed to prolonged external lumbar drainage in excess of 10 ml per hour during 3 days. Clinical response to lumbar drainage was assessed by a walk test and a mini-mental test.

Results

Twelve of 19 patients showed a positive clinical response and underwent a shunting procedure. Volumetric values of intracranial space content in responders and non-responders showed no statistically significant difference. Total CSF volume (sum of cranial and spinal CSF volumes) was higher than previously published. No correlation was found between spinal canal length, CSF pressure, and CSF spinal volume. The results show that there is a significantly higher CSF volume in the spinal space in the responder group (n = 12) (120.5 ± 14.9 ml) compared with the non-responder group (103.1 ± 27.4 ml; n = 7).

Discussion

This study demonstrates for the first time that CSF volume in the spinal space may have predictive value in the preoperative assessment of iNPH patients. The results suggest that patients with increased spinal CSF volume have decreased compliance. Additional prospective randomized clinical trials are needed to confirm our results.

Reproducibility and Reliability of Computing Models in Segmentation and Volumetric Measurement of Brain

Background

Segmentation and morphometric measurement of brain tissue and regions from non-invasive magnetic resonance images have clinical and research applications. Several software tools and models have been developed by different research groups which are increasingly used for segmentation and morphometric measurements. Variability in results has been observed in the imaging data processed with different neuroimaging pipelines which have increased the focus on standardization.

Purpose

The availability of several tools and models for brain morphometry poses challenges as an analysis done on the same set of data using different sets of tools and pipelines may result in different results and interpretations and there is a need for understanding the reliability and accuracy of such models.

Methods

T1-weighted (T1-w) brain volumes from the publicly available OASIS3 dataset have been analysed using recent versions of FreeSurfer, FSL-FAST, CAT12, and ANTs pipelines. grey matter (GM), white matter (WM), and estimated total intracranial volume (eTIV) have been extracted and compared for inter-method variability and accuracy.

Results

All four methods are consistent and strongly reproducible in their measurement across subjects however there is a significant degree of variability between these methods.

Conclusion

CAT12 and FreeSurfer methods have the highest degree of agreement in tissue class segmentation and are most reproducible compared to others.

Impact of brain segmentation methods on regional metabolism quantification in 18F-FDG PET/MR analysis

BACKGROUND

Accurate analysis of quantitative PET data plays a crucial role in studying small, specific brain structures. The integration of PET and MRI through an integrated PET/MR system presents an opportunity to leverage the benefits of precisely aligned structural MRI and molecular PET images in both spatial and temporal dimensions. However, in many clinical workflows, PET studies are often performed without the aid of individually matched structural MRI scans, primarily for the sake of convenience in the data collection and brain segmentation possesses. Currently, two commonly employed segmentation strategies for brain PET analysis are distinguished: methods with or without MRI registration and methods employing either atlas-based or individual-based algorithms. Moreover, the development of artificial intelligence (AI)-assisted methods for predicting brain segmentation holds promise but requires further validation of their efficiency and accuracy for clinical applications. This study aims to compare and evaluate the correlations, consistencies, and differences among the above-mentioned brain segmentation strategies in quantification of brain metabolism in 18F-FDG PET/MR analysis.

RESULTS

Strong correlations were observed among all methods (r = 0.932 to 0.999, P < 0.001). The variances attributable to subject and brain region were higher than those caused by segmentation methods (P < 0.001). However, intraclass correlation coefficient (ICC)s between methods with or without MRI registration ranged from 0.924 to 0.975, while ICCs between methods with atlas- or individual-based algorithms ranged from 0.741 to 0.879. Brain regions exhibiting significant standardized uptake values (SUV) differences due to segmentation methods were the basal ganglia nuclei (maximum to 11.50 ± 4.67%), and various cerebral cortexes in temporal and occipital regions (maximum to 18.03 ± 5.52%). The AI-based method demonstrated high correlation (r = 0.998 and 0.999, P < 0.001) and ICC (0.998 and 0.997) with FreeSurfer, substantially reducing the time from 8.13 h to 57 s on per subject.

CONCLUSIONS

Different segmentation methods may have impact on the calculation of brain metabolism in basal ganglia nuclei and specific cerebral cortexes. The AI-based approach offers improved efficiency and is recommended for its enhanced performance.

Machine learning-based automatic estimation of cortical atrophy using brain computed tomography images

Cortical atrophy is measured clinically according to established visual rating scales based on magnetic resonance imaging (MRI). Although brain MRI is the primary imaging marker for neurodegeneration, computed tomography (CT) is also widely used for the early detection and diagnosis of dementia. However, they are seldom investigated. Therefore, we developed a machine learning algorithm for the automatic estimation of cortical atrophy on brain CT. Brain CT images (259 Alzheimer's dementia and 55 cognitively normal subjects) were visually rated by three neurologists and used for training. We constructed an algorithm by combining the convolutional neural network and regularized logistic regression (RLR). Model performance was then compared with that of neurologists, and feature importance was measured. RLR provided fast and reliable automatic estimations of frontal atrophy (75.2% accuracy, 93.6% sensitivity, 67.2% specificity, and 0.87 area under the curve [AUC]), posterior atrophy (79.6% accuracy, 87.2% sensitivity, 75.9% specificity, and 0.88 AUC), right medial temporal atrophy (81.2% accuracy, 84.7% sensitivity, 79.6% specificity, and 0.88 AUC), and left medial temporal atrophy (77.7% accuracy, 91.1% sensitivity, 72.3% specificity, and 0.90 AUC). We concluded that RLR-based automatic estimation of brain CT provided a comprehensive rating of atrophy that can potentially support physicians in real clinical settings.

The instant impact of a single hemodialysis session on brain morphological measurements in patients with end-stage renal disease

Objective

To investigate the instant impact of hemodialysis (HD) on the cerebral morphological measurements of patients with end-stage renal disease (ESRD).

Materials and methods

Twenty-five patients undergoing maintenance HD and twenty-eight age-, sex-, and education-matched healthy control (HC) were included. The HD group and HC group had 3D high-resolution structural magnetic resonance imaging (MRI) scans twice and once, respectively. Both groups underwent neuropsychologic tests. The morphological measurements of structural MRI were measured using CAT12 and these measures were compared among three groups. The relationship between morphological measures and clinical parameters and neuropsychological tests were investigated through multiple regression analysis.

Results

Compared to the HC group, the cortical thickness before HD significantly decreased in the bilateral temporal lobe and significantly decreased in the left superior temporal gyrus after HD. The cortical thickness significantly increased in the bilateral temporal lobe, frontal lobe and occipital lobe after HD compared to before HD. The sulcus depth in the bilateral insula, frontal lobe, and parietal lobe after HD significantly increased compared to before HD. No significant differences in sulcus depth between HD and HC were detected. After HD, the cortical thickness of the right parsopercularis was positively correlated with the number connection test-A. Cortical thickness in multiple regions were positively correlated with blood flow velocity and cortical thickness in the left parahippocampal gyrus was negatively correlated with ultrafiltration volume. Patients showed better performance in the digit symbol test and line tracing test after HD compared to before HD, but there were no significant differences in the comparison of neuropsychologic tests between patients and HC.

Conclusion

The instant morphological changes were captured during a single hemodialysis in HD patients. There was an association between these instant changes in the brain and clinical parameters and neuropsychologic tests. This work implied the instant impact of a single hemodialysis impact on the brain in HD patients.

Investigation of biases in convolutional neural networks for semantic segmentation using performance sensitivity analysis

The application of deep neural networks for segmentation in medical imaging has gained substantial interest in recent years. In many cases, this variant of machine learning has been shown to outperform other conventional segmentation approaches. However, little is known about its general applicability. Especially the robustness against image modifications (e.g., intensity variations, contrast variations, spatial alignment) has hardly been investigated. Data augmentation is often used to compensate for sensitivity to such changes, although its effectiveness has not yet been studied. Therefore, the goal of this study was to systematically investigate the sensitivity to variations in input data with respect to segmentation of medical images using deep learning. This approach was tested with two publicly available segmentation frameworks (DeepMedic and TractSeg). In the case of DeepMedic, the performance was tested using ground truth data, while in the case of TractSeg, the STAPLE technique was employed. In both cases, sensitivity analysis revealed significant dependence of the segmentation performance on input variations. The effects of different data augmentation strategies were also shown, making this type of analysis a useful tool for selecting the right parameters for augmentation. The proposed analysis should be applied to any deep learning image segmentation approach, unless the assessment of sensitivity to input variations can be directly derived from the network.

Advantages of Using Both Voxel- and Surface-based Morphometry in Cortical Morphology Analysis: A Review of Various Applications

Surface-based morphometry (SBM) is extremely useful for estimating the indices of cortical morphology, such as volume, thickness, area, and gyrification, whereas voxel-based morphometry (VBM) is a typical method of gray matter (GM) volumetry that includes cortex measurement. In cases where SBM is used to estimate cortical morphology, it remains controversial as to whether VBM should be used in addition to estimate GM volume. Therefore, this review has two main goals. First, we summarize the differences between the two methods regarding preprocessing, statistical analysis, and reliability. Second, we review studies that estimate cortical morphological changes using VBM and/or SBM and discuss whether using VBM in conjunction with SBM produces additional values. We found cases in which detection of morphological change in either VBM or SBM was superior, and others that showed equivalent performance between the two methods. Therefore, we concluded that using VBM and SBM together can help researchers and clinicians obtain a better understanding of normal neurobiological processes of the brain. Moreover, the use of both methods may improve the accuracy of the detection of morphological changes when comparing the data of patients and controls.

In addition, we introduce two other recent methods as future directions for estimating cortical morphological changes: a multi-modal parcellation method using structural and functional images, and a synthetic segmentation method using multi-contrast images (such as T1- and proton density-weighted images).

A comparison of automated atrophy measures across the frontotemporal dementia spectrum: Implications for trials

Background

Frontotemporal dementia (FTD) is a common cause of young onset dementia, and whilst there are currently no treatments, there are several promising candidates in development and early phase trials. Comprehensive investigations of neuroimaging markers of disease progression across the full spectrum of FTD disorders are lacking and urgently needed to facilitate these trials.

Objective

To investigate the comparative performance of multiple automated segmentation and registration pipelines used to quantify longitudinal whole-brain atrophy across the clinical, genetic and pathological subgroups of FTD, in order to inform upcoming trials about suitable neuroimaging-based endpoints.

Methods

Seventeen fully automated techniques for extracting whole-brain atrophy measures were applied and directly compared in a cohort of 226 participants who had undergone longitudinal structural 3D T1-weighted imaging. Clinical diagnoses were behavioural variant FTD (n = 56) and primary progressive aphasia (PPA, n = 104), comprising semantic variant PPA (n = 38), non-fluent variant PPA (n = 42), logopenic variant PPA (n = 18), and PPA-not otherwise specified (n = 6). 49 of these patients had either a known pathogenic mutation or postmortem confirmation of their underlying pathology. 66 healthy controls were included for comparison. Sample size estimates to detect a 30% reduction in atrophy (80% power; 0.05 significance) were computed to explore the relative feasibility of these brain measures as surrogate markers of disease progression and their ability to detect putative disease-modifying treatment effects.

Results

Multiple automated techniques showed great promise, detecting significantly increased rates of whole-brain atrophy (p<0.001) and requiring sample sizes of substantially less than 100 patients per treatment arm. Across the different FTD subgroups, direct measures of volume change consistently outperformed their indirect counterparts, irrespective of the initial segmentation quality. Significant differences in performance were found between both techniques and patient subgroups, highlighting the importance of informed biomarker choice based on the patient population of interest.

Conclusion

This work expands current knowledge and builds on the limited longitudinal investigations currently available in FTD, as well as providing valuable information about the potential of fully automated neuroimaging biomarkers for sporadic and genetic FTD trials.

RT-NET: real-time reconstruction of neural activity using high-density electroencephalography

High-density electroencephalography (hdEEG) has been successfully used for large-scale investigations of neural activity in the healthy and diseased human brain. Because of their high computational demand, analyses of source-projected hdEEG data are typically performed offline. Here, we present a real-time noninvasive electrophysiology toolbox, RT-NET, which has been specifically developed for online reconstruction of neural activity using hdEEG. RT-NET relies on the Lab Streaming Layer for acquiring raw data from a large number of EEG amplifiers and for streaming the processed data to external applications. RT-NET estimates a spatial filter for artifact removal and source activity reconstruction using a calibration dataset. This spatial filter is then applied to the hdEEG data as they are acquired, thereby ensuring low latencies and computation times. Overall, our analyses show that RT-NET can estimate real-time neural activity with performance comparable to offline analysis methods. It may therefore enable the development of novel brain–computer interface applications such as source-based neurofeedback.

23 Na MRI in ischemic stroke: Acquisition time reduction using postprocessing with convolutional neural networks

Quantitative ²³ Na magnetic resonance imaging (MRI) provides tissue sodium concentration (TSC), which is connected to cell viability and vitality. Long acquisition times are one of the most challenging aspects for its clinical establishment. K-space undersampling is an approach for acquisition time reduction, but generates noise and artifacts. The use of convolutional neural networks (CNNs) is increasing in medical imaging and they are a useful tool for MRI postprocessing. The aim of this study is ²³ Na MRI acquisition time reduction by k-space undersampling. CNNs were applied to reduce the resulting noise and artifacts. A retrospective analysis from a prospective study was conducted including image datasets from 46 patients (aged 72 ± 13 years; 25 women, 21 men) with ischemic stroke; the ²³ Na MRI acquisition time was 10 min. The reconstructions were performed with full dataset (FI) and with a simulated dataset an image that was acquired in 2.5 min (RI). Eight different CNNs with either U-Net-based or ResNet-based architectures were implemented with RI as input and FI as label, using batch normalization and the number of filters as varying parameters. Training was performed with 9500 samples and testing included 400 samples. CNN outputs were evaluated based on signal-to-noise ratio (SNR) and structural similarity (SSIM). After quantification, TSC error was calculated. The image quality was subjectively rated by three neuroradiologists. Statistical significance was evaluated by Student's t-test. The average SNR was 21.72 ± 2.75 (FI) and 10.16 ± 0.96 (RI). U-Nets increased the SNR of RI to 43.99 and therefore performed better than ResNet. SSIM of RI to FI was improved by three CNNs to 0.91 ± 0.03. CNNs reduced TSC error by up to 15%. The subjective rating of CNN-generated images showed significantly better results than the subjective image rating of RI. The acquisition time of ²³ Na MRI can be reduced by 75% due to postprocessing with a CNN on highly undersampled data.

Tissue Sodium Concentration within White Matter Correlates with the Extent of Small Vessel Disease

INTRODUCTION

Sodium MRI (23Na MRI) derived biomarkers such as tissue sodium concentration (TSC) provide valuable information on cell function and brain tissue viability and has become a reliable tool for the assessment of brain tumors and ischemic stroke beyond pathoanatomical morphology. Patients with major stroke often suffer from different degrees of underlying white matter lesions (WMLs) attributed to chronic small vessel disease. This study aimed to evaluate the WM TSC in patients with an acute ischemic stroke and to correlate the TSC with the extent of small vessel disease. Furthermore, the reliability of relative TSC (rTSC) compared to absolute TSC in these patients was analyzed.

METHODOLOGY

We prospectively examined 62 patients with acute ischemic stroke (73 ± 13 years) between November 2016 and August 2019 from which 18 patients were excluded and thus 44 patients were evaluated. A 3D 23Na MRI was acquired in addition to a T2-TIRM and a diffusion-weighted image. Coregistration and segmentation were performed with SPM 12 based on the T2-TIRM image. The extension of WM T2 hyperintense lesions in each patient was classified using the Fazekas scale of WMLs. The absolute TSC in the WM region was correlated to the Fazekas grades. The stroke region was manually segmented on the coregistered absolute diffusion coefficient image and absolute, and rTSC was calculated in the stroke region and compared to nonischemic WM region. Statistical significance was evaluated using the Student t-test.

RESULTS

For patients with Fazekas grade I (n = 25, age: 68.5 ± 15.1 years), mean TSC in WM was 55.57 ± 7.43 mM, and it was not statistically significant different from patients with Fazekas grade II (n = 7, age: 77.9 ± 6.4 years) with a mean TSC in WM of 53.9 ± 6.4 mM, p = 0.58. For patients with Fazekas grade III (n = 9, age: 81.4 ± 7.9 years), mean TSC in WM was 68.7 ± 10.5 mM, which is statistically significantly higher than the TSC in patients with Fazekas grade I and II (p < 0.001 and p = 0.05, respectively). There was a positive correlation between the TSC in WM and the Fazekas grade with r = 0.48 p < 0.001. The rTSC in the stroke region was statistically significant difference between low (0 and I) and high (2 and 3) Fazekas grades (p = 0.0353) whereas there was no statistically significant difference in absolute TSC in the stroke region between low (0 and I) and high (2 and 3) Fazekas grades.

CONCLUSION

The significant difference in absolute TSC in WM in patients with severe small vessel disease; Fazekas grade 3 can lead to inaccuracies using rTSC quantification for evaluation of acute ischemic stroke using 23 Na MRI. The study, therefore, emphasizes the importance of absolute tissue sodium quantification.

Clinically Applicable Quantitative Magnetic Resonance Morphologic Measurements of Grey Matter Changes in the Human Brain

(1) Purpose: Quantitative magnetic resonance imaging (qMRI) measurements can be used to sensitively estimate brain morphological alterations and may support clinical diagnosis of neurodegenerative diseases (ND). We aimed to establish a normative reference database for a clinical applicable quantitative MR morphologic measurement on neurodegenerative changes in patients; (2) Methods: Healthy subjects (HCs, n = 120) with an evenly distribution between 21 to 70 years and amyotrophic lateral sclerosis (ALS) patients (n = 11, mean age = 52.45 ± 6.80 years), as an example of ND patients, underwent magnetic resonance imaging (MRI) examinations under routine diagnostic conditions. Regional cortical thickness (rCTh) in 68 regions of interest (ROIs) and subcortical grey matter volume (SGMV) in 14 ROIs were determined from all subjects by using Computational Anatomy Toolbox. Those derived from HCs were analyzed to determine age-related differences and subsequently used as reference to estimate ALS-related alterations; (3) Results: In HCs, the rCTh (in 49/68 regions) and the SGMV (in 9/14 regions) in elderly subjects were less than those in younger subjects and exhibited negative linear correlations to age (p < 0.0007 for rCTh and p < 0.004 for SGMV). In comparison to age- and sex-matched HCs, the ALS patients revealed significant decreases of rCTh in eight ROIs, majorly located in frontal and temporal lobes; (4) Conclusion: The present study proves an overall grey matter decline with normal ageing as reported previously. The provided reference may be used for detection of grey matter alterations in neurodegenerative diseases that are not apparent in standard MR scans, indicating the potential of using qMRI as an add-on diagnostic tool in a clinical setting.

A Review of Publicly Available Automatic Brain Segmentation Methodologies, Machine Learning Models, Recent Advancements, and Their Comparison

BACKGROUND

The noninvasive study of the structure and functions of the brain using neuroimaging techniques is increasingly being used for its clinical and research perspective. The morphological and volumetric changes in several regions and structures of brains are associated with the prognosis of neurological disorders such as Alzheimer's disease, epilepsy, schizophrenia, etc. and the early identification of such changes can have huge clinical significance. The accurate segmentation of three-dimensional brain magnetic resonance images into tissue types (i.e., grey matter, white matter, cerebrospinal fluid) and brain structures, thus, has huge importance as they can act as early biomarkers. The manual segmentation though considered the "gold standard" is time-consuming, subjective, and not suitable for bigger neuroimaging studies. Several automatic segmentation tools and algorithms have been developed over the years; the machine learning models particularly those using deep convolutional neural network (CNN) architecture are increasingly being applied to improve the accuracy of automatic methods.

PURPOSE

The purpose of the study is to understand the current and emerging state of automatic segmentation tools, their comparison, machine learning models, their reliability, and shortcomings with an intent to focus on the development of improved methods and algorithms.

METHODS

The study focuses on the review of publicly available neuroimaging tools, their comparison, and emerging machine learning models particularly those based on CNN architecture developed and published during the last five years.

CONCLUSION

Several software tools developed by various research groups and made publicly available for automatic segmentation of the brain show variability in their results in several comparison studies and have not attained the level of reliability required for clinical studies. The machine learning models particularly three dimensional fully convolutional network models can provide a robust and efficient alternative with relation to publicly available tools but perform poorly on unseen datasets. The challenges related to training, computation cost, reproducibility, and validation across distinct scanning modalities for machine learning models need to be addressed.

Expert Opinions and Recommendations for the Clinical Use of Quantitative Analysis Software for MRI-Based Brain Volumetry

치매를 비롯한 퇴행성 신경 질환의 초기 진단에 자기공명영상을 이용한 뇌 위축 평가와 정량적 용적 분석이 중요하다. 뇌 위축의 시각적 평가는 주관적으로 평가자에 따라 다른 결과를 보여주기 때문에, 객관적인 결과를 제공하면서 임상 적용도 가능한 소프트웨어의 수요와 개발이 늘어나고 있다. 이러한 임상용 소프트웨어의 실제 임상 적용은 영상 검사의 표준화가 선행되어야 하고, 개발된 소프트웨어의 검증이 반드시 필요하다. 따라서 대한신경두경부영상의학회는 뇌용적 분석 임상용 소프트웨어의 임상적 활용에 대한 의견을 제시하기 위해 전문위원회를 구성하고 현재까지 발표된 연구를 정리하였다. 그리고, 정량화 분석을 위한 영상 검사의 표준화 및 소프트웨어의 임상 적용에 대한 전문가 의견을 제시하기 위하여 공동 작업을 수행하였다. 본 종설에서는 뇌 자기공명영상의 정량화 분석의 필요성 및 배경, 정량화 분석을 위한 임상용 소프트웨어의 소개 및 기존의 표준품(reference standard)과의 진단능 비교, 영상 획득의 표준화, 분석 및 평가의 표준화, 소프트웨어의 임상 적용에 대한 전문가 의견, 제한점 및 대처 방법 등 대한신경두경부영상의학회의 전문가 권고안을 소개하는 것이 목적이다.

Brain Morphological Alterations Are Detected in Early-Stage Parkinson's Disease with MRI Morphometry

BACKGROUND AND PURPOSE

To detect brain morphological alterations in patients with early Parkinson's disease (PD) by using magnetic resonance imaging (MRI) morphometry under radiological diagnostic conditions.

METHODS

T1-weighted brain images of 18 early PD patients and 18 age-sex-matched healthy controls (HCs) were analyzed with free software Computational Anatomy Toolbox (CAT12). Regional cortical thickness (rCTh) in 68 atlas-defined regions-of-interest (ROIs) and subcortical gray matter volume (SGMV) in 14 atlas-defined ROIs were determined and compared between patients and HCs by paired comparison using both ROI-wise and voxel-wise analyses. False-discovery rate (FDR) was used multiple comparison correction. Possible correlations between brain morphological changes in patients and clinical observations were also analyzed.

RESULTS

Comparing to the HCs, the ROI-wise analysis revealed rCTh thinning significantly in left medial orbitofrontal (P = .001), by trend (P < .05 but not significant after FDR correction) in four other ROIs located in frontal and temporal lobes, and a volume decreasing trend in left pallidum of the PD patients, while the voxel-wise analysis revealed one cluster with rCTh thinning trend located between left insula and superior temporal region of the patients. In addition, the patients showed more distinct rCTh thinning in ipsilateral hemisphere and SGMV deceasing trends in contralateral hemisphere in respect of the symptom-onset body side.

CONCLUSION

Brain morphological alterations in early PD patients are evident despite of their inconspicuous findings in standard MRI. Quantitative morphological measurements with CAT12 may be an applicable add-on tool for clinical diagnosis of early PD. These results have to be verified in future studies with larger patient samples.

Creation of an anthropomorphic CT head phantom for verification of image segmentation

Purpose

Many methods are available to segment structural magnetic resonance (MR) images of the brain into different tissue types. These have generally been developed for research purposes but there is some clinical use in the diagnosis of neurodegenerative diseases such as dementia. The potential exists for computed tomography (CT) segmentation to be used in place of MRI segmentation, but this will require a method to verify the accuracy of CT processing, particularly if algorithms developed for MR are used, as MR has notably greater tissue contrast.

Methods

To investigate these issues we have created a three‐dimensional (3D) printed brain with realistic Hounsfield unit (HU) values based on tissue maps segmented directly from an individual T1 MRI scan of a normal subject. Several T1 MRI scans of normal subjects from the ADNI database were segmented using SPM12 and used to create stereolithography files of different tissues for 3D printing. The attenuation properties of several material blends were investigated, and three suitable formulations were used to print an object expected to have realistic geometry and attenuation properties. A skull was simulated by coating the object with plaster of Paris impregnated bandages. Using two CT scanners, the realism of the phantom was assessed by the measurement of HU values, SPM12 segmentation and comparison with the source data used to create the phantom.

Results

Realistic relative HU values were measured although a subtraction of 60 was required to obtain equivalence with the expected values (gray matter 32.9–35.8 phantom, 29.9–34.2 literature). Segmentation of images acquired at different kVps/mAs showed excellent agreement with the source data (Dice Similarity Coefficient 0.79 for gray matter). The performance of two scanners with two segmentation methods was compared, with the scanners found to have similar performance and with one segmentation method clearly superior to the other.

Conclusion

The ability to use 3D printing to create a realistic (in terms of geometry and attenuation properties) head phantom has been demonstrated and used in an initial assessment of CT segmentation accuracy using freely available software developed for MRI.

Test-retest variability of brain morphometry analysis: an investigation of sequence and coil effects

Background

Precise and reliable brain morphometry analysis is critical for clinical and research purposes. The magnetization-prepared rapid gradient echo (MPRAGE), multi-echo MPRAGE (MEMPRAGE) and magnetization-prepared 2 rapid acquisition gradient echo (MP2RAGE) sequences have all been used to acquire brain structural images, but it is unclear which of these sequences is the most suitable for brain morphometry and whether the number of coil channels (20 or 32) affects scan precision. This study aimed to assess the impact of T1-weighted image acquisition variables (sequence and head coil) on the repeatability of resultant automated volumetric measurements.

Methods

Twenty-four healthy volunteers underwent back-to-back scanning protocols with three sequences and two different coils (i.e., six scanning conditions in total) presented in a randomized order in a single session. MorphoBox prototype and FreeSurfer were used for brain segmentation. Brain structures were divided into cortical and subcortical regions for more precise analysis. The acquired volume and thickness values were used to calculate test-retest variability (TRV) values. TRV values from the six different combinations were compared for total brain structures, total cortical structures, total subcortical structures, and every single structure.

Results

The median TRV value for all brain regions was 1.23% with MorphoBox and 3.14% with FreeSurfer. When using FreeSurfer results to compare the six combinations, for total brain structures volume and total cortical structures volume and thickness, the MEMPRAGE-32 channel combination showed significantly lower TRV values than the others (P<0.01). Similar results were observed with MorphoBox. For total subcortical structures, the MP2RAGE-32 channel combination showed the lowest TRV values with both MorphoBox (lower about 0.01% to 0.17%) and FreeSurfer analyses (lower about 0.02% to 0.37%).

Conclusions

TRV values were generally low, indicating generally high reliability for every region. The MEMPRAGE sequence was the most reliable of the three sequences for total brain structures and cortical structures. However, MP2RAGE was the most reliable for subcortical structures. The 32-channel coil showed better repeatability results than the 20-channel coil.

Cortical surface morphology in long-term cannabis users: A multi-site MRI study

Cannabis exerts its psychoactive effect through cannabinoid receptors that are widely distributed across the cortical surface of the human brain. It is suggested that cannabis use may contribute to structural alterations across the cortical surface. In a large, multisite dataset of 120 controls and 141 cannabis users, we examined whether differences in key characteristics of the cortical surface - including cortical thickness, surface area, and gyrification index were related to cannabis use characteristics, including (i) cannabis use vs. non-use, (ii) cannabis dependence vs. non-dependence vs. non-use, and (iii) early adolescent vs. late adolescent onset of cannabis use vs. non-use. Our results revealed that cortical morphology was not associated with cannabis use, dependence, or onset age. The lack of effect of regular cannabis use, including problematic use, on cortical structure in our study is contrary to previous evidence of cortical morphological alterations (particularly in relation to cannabis dependence and cannabis onset age) in cannabis users. Careful reevaluation of the evidence on cannabis-related harm will be necessary to address concerns surrounding the long-term effects of cannabis use and inform policies in a changing cannabis regulation climate.

White matter volume alterations in hair-pulling disorder (trichotillomania)

Trichotillomania (TTM) is a disorder characterized by repetitive hair-pulling resulting in hair loss. Key processes affected in TTM comprise affective, cognitive, and motor functions. Emerging evidence suggests that brain matter aberrations in fronto-striatal and fronto-limbic brain networks and the cerebellum may characterize the pathophysiology of TTM. The aim of the present voxel-based morphometry (VBM) study was to evaluate whole brain grey and white matter volume alteration in TTM and its correlation with hair-pulling severity. High-resolution magnetic resonance imaging (3 T) data were acquired from 29 TTM patients and 28 age-matched healthy controls (CTRLs). All TTM participants completed the Massachusetts General Hospital Hair-Pulling Scale (MGH-HPS) to assess illness/pulling severity. Using whole-brain VBM, between-group differences in regional brain volumes were measured. Additionally, within the TTM group, the relationship between MGH-HPS scores, illness duration and brain volumes were examined. All data were corrected for multiple comparisons using family-wise error (FWE) correction at p < 0.05. Patients with TTM showed larger white matter volumes in the parahippocampal gyrus and cerebellum compared to CTRLs. Estimated white matter volumes showed no significant association with illness duration or MGH-HPS total scores. No significant between-group differences were found for grey matter volumes. Our observations suggest regional alterations in cortico-limbic and cerebellar white matter in patients with TTM, which may underlie deficits in cognitive and affective processing. Such volumetric white matter changes may precipitate impaired cortico-cerebellar communication leading to a reduced ability to control hair pulling behavior.

Alteration to hippocampal volume and shape confined to cannabis dependence: a multi-site study

Cannabis use is highly prevalent and often considered to be relatively harmless. Nonetheless, a subset of regular cannabis users may develop dependence, experiencing poorer quality of life and greater mental health problems relative to non-dependent users. The neuroanatomy characterizing cannabis use versus dependence is poorly understood. We aimed to delineate the contributing role of cannabis use and dependence on morphology of the hippocampus, one of the most consistently altered brain regions in cannabis users, in a large multi-site dataset aggregated across four research sites. We compared hippocampal volume and vertex-level hippocampal shape differences (1) between 121 non-using controls and 140 cannabis users; (2) between 106 controls, 50 non-dependent users and 70 dependent users; and (3) between a subset of 41 controls, 41 non-dependent users and 41 dependent users, matched on sample characteristics and cannabis use pattern (onset age and dosage). Cannabis users did not differ from controls in hippocampal volume or shape. However, cannabis-dependent users had significantly smaller right and left hippocampi relative to controls and non-dependent users, irrespective of cannabis dosage. Shape analysis indicated localized deflations in the superior-medial body of the hippocampus. Our findings support neuroscientific theories postulating dependence-specific neuroadaptations in cannabis users. Future efforts should uncover the neurobiological risk and liabilities separating dependent and non-dependent use of cannabis.

Comparative analysis of methods of volume adjustment in hippocampal volumetry for the diagnosis of Alzheimer disease

INTRODUCTION

Hippocampal volumetry can discriminate normal subjects from patients with amnestic mild cognitive impairment (MCI) or Alzheimer disease (AD). We have analyzed the effects of different methods of hippocampal volume (HV) adjustment on the diagnostic accuracy of this technique.

METHODS

Cross-sectional analysis of 148 subjects of the ADNI database (48 normal, 66 MCI, 34 AD). Brain volumes were calculated from 3T MRI scans with gm extractor, a fully automated script based on FSL. A series of logistic regression models was obtained using 9 volumes of reference and 3 methods of adjustment (normalization, covariance, bilinear regression). Diagnostic accuracy was evaluated with the receiver operating characteristic curve method. External validity was assessed with 10-fold cross-validation.

RESULTS

The models with the highest area under the curve (AUC) were those including the HV normalized by total intracranial volume (TIV). The differences with bilinear regression and the covariance method adjusted by TIV were minor and not statistically significant. The lowest AUCs corresponded to the models based on raw (unadjusted) HVs. The results were qualitatively similar in two clinical settings (normal versus MCI, and normal versus AD), but the differences were higher in the normal versus MCI context.

CONCLUSION

The accuracy of hippocampal volumetry for the differential diagnosis between normal subjects and patients with MCI or AD was maximized by normalizing the HV by the TIV. Our results do not exclude the potential superiority of non-linear models.

T1-MPRAGE and T2-FLAIR segmentation of cortical and subcortical brain regions-an MRI evaluation study

PURPOSE

Development of a warp-based automated brain segmentation approach of 3D fluid-attenuated inversion recovery (FLAIR) images and comparison to 3D T1-based segmentation.

METHODS

3D FLAIR and 3D T1-weighted sequences of 30 healthy subjects (mean age 29.9 ± 8.3 years, 8 female) were acquired on the same 3T MR scanner. Warp-based segmentation was applied for volumetry of total gray matter (GM), white matter (WM), and 116 atlas regions. Segmentation results of both sequences were compared using Pearson correlation (r).

RESULTS

Correlation of GM segmentation results based on FLAIR and T1 was overall good for cortical structures (mean r across all cortical structures = 0.76). Comparatively weaker results were found in the occipital lobe (r = 0.77), central region (mean r = 0.58), basal ganglia (mean r = 0.59), thalamus (r = 0.30), and cerebellum (r = 0.73). FLAIR segmentation underestimated volume of the central region compared to T1, but showed a better anatomic concordance with the occipital lobe on visual review and subcortical structures, when also compared to manual segmentation. Visual analysis of FLAIR-based WM segmentation revealed frequent misclassification of regions of high signal intensity as GM.

CONCLUSION

Warp-based FLAIR segmentation yields comparable results to T1 segmentation for most cortical GM structures and may provide anatomically more congruent segmentation of subcortical GM structures. Selected cortical regions, especially the central region and total WM, seem to be underestimated on FLAIR segmentation.

Future Brain and Spinal Cord Volumetric Imaging in the Clinic for Monitoring Treatment Response in MS

PURPOSE OF REVIEW

Volumetric analysis of brain imaging has emerged as a standard approach used in clinical research, e.g., in the field of multiple sclerosis (MS), but its application in individual disease course monitoring is still hampered by biological and technical limitations. This review summarizes novel developments in volumetric imaging on the road towards clinical application to eventually monitor treatment response in patients with MS.

RECENT FINDINGS

In addition to the assessment of whole-brain volume changes, recent work was focused on the volumetry of specific compartments and substructures of the central nervous system (CNS) in MS. This included volumetric imaging of the deep brain structures and of the spinal cord white and gray matter. Volume changes of the latter indeed independently correlate with clinical outcome measures especially in progressive MS. Ultrahigh field MRI and quantitative MRI added to this trend by providing a better visualization of small compartments on highly resolving MR images as well as microstructural information. New developments in volumetric imaging have the potential to improve sensitivity as well as specificity in detecting and hence monitoring disease-related CNS volume changes in MS.

Recommendations for the Use of Automated Gray Matter Segmentation Tools: Evidence from Huntington's Disease

The selection of an appropriate segmentation tool is a challenge facing any researcher aiming to measure gray matter (GM) volume. Many tools have been compared, yet there is currently no method that can be recommended above all others; in particular, there is a lack of validation in disease cohorts. This work utilizes a clinical dataset to conduct an extensive comparison of segmentation tools. Our results confirm that all tools have advantages and disadvantages, and we present a series of considerations that may be of use when selecting a GM segmentation method, rather than a ranking of these tools. Seven segmentation tools were compared using 3 T MRI data from 20 controls, 40 premanifest Huntington's disease (HD), and 40 early HD participants. Segmented volumes underwent detailed visual quality control. Reliability and repeatability of total, cortical, and lobular GM were investigated in repeated baseline scans. The relationship between each tool was also examined. Longitudinal within-group change over 3 years was assessed via generalized least squares regression to determine sensitivity of each tool to disease effects. Visual quality control and raw volumes highlighted large variability between tools, especially in occipital and temporal regions. Most tools showed reliable performance and the volumes were generally correlated. Results for longitudinal within-group change varied between tools, especially within lobular regions. These differences highlight the need for careful selection of segmentation methods in clinical neuroimaging studies. This guide acts as a primer aimed at the novice or non-technical imaging scientist providing recommendations for the selection of cohort-appropriate GM segmentation software.

Comparing CAT12 and VBM8 for Detecting Brain Morphological Abnormalities in Temporal Lobe Epilepsy

The identification of the brain morphological alterations that play important roles in neurodegenerative/neurological diseases will contribute to our understanding of the causes of these diseases. Various automated software programs are designed to provide an automatic framework to detect brain morphological changes in structural magnetic resonance imaging (MRI) data. A voxel-based morphometry (VBM) analysis can also be used for the detection of brain volumetric abnormalities. Here, we compared gray matter (GM) and white matter (WM) abnormality results obtained by a VBM analysis using the Computational Anatomy Toolbox (CAT12) via the current version of Statistical Parametric Mapping software (SPM12) with the results obtained by a VBM analysis using the VBM8 toolbox implemented in the older software SPM8, in adult temporal lobe epilepsy (TLE) patients with (n = 51) and without (n = 57) hippocampus sclerosis (HS), compared to healthy adult controls (n = 28). The VBM analysis using CAT12 showed that compared to the healthy controls, significant GM and WM reductions were located in ipsilateral mesial temporal lobes in the TLE-HS patients, and slight GM amygdala swelling was present in the right TLE-no patients (n = 27). In contrast, the VBM analysis via the VBM8 toolbox showed significant GM and WM reductions only in the left TLE-HS patients (n = 25) compared to the healthy controls. Our findings thus demonstrate that compared to VBM8, a VBM analysis using CAT12 provides a more accurate volumetric analysis of the brain regions in TLE. Our results further indicate that a VBM analysis using CAT12 is more robust and accurate against volumetric alterations than the VBM8 toolbox.

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.

Orbitofrontal and caudate volumes in cannabis users: a multi-site mega-analysis comparing dependent versus non-dependent users

RATIONALE

Cannabis (CB) use and dependence are associated with regionally specific alterations to brain circuitry and substantial psychosocial impairment.

OBJECTIVES

The objective of this study was to investigate the association between CB use and dependence, and the volumes of brain regions critically involved in goal-directed learning and behaviour-the orbitofrontal cortex (OFC) and caudate.

METHODS

In the largest multi-site structural imaging study of CB users vs healthy controls (HC), 140 CB users and 121 HC were recruited from four research sites. Group differences in OFC and caudate volumes were investigated between HC and CB users and between 70 dependent (CB-dep) and 50 non-dependent (CB-nondep) users. The relationship between quantity of CB use and age of onset of use and caudate and OFC volumes was explored.

RESULTS

CB users (consisting of CB-dep and CB-nondep) did not significantly differ from HC in OFC or caudate volume. CB-dep compared to CB-nondep users exhibited significantly smaller volume in the medial and the lateral OFC. Lateral OFC volume was particularly smaller in CB-dep females, and reduced volume in the CB-dep group was associated with higher monthly cannabis dosage.

CONCLUSIONS

Smaller medial OFC volume may be driven by CB dependence-related mechanisms, while smaller lateral OFC volume may be due to ongoing exposure to cannabinoid compounds. The results highlight a distinction between cannabis use and dependence and warrant examination of gender-specific effects in studies of CB dependence.

Neuroanatomical and Neuropsychological Markers of Amnestic MCI: A Three-Year Longitudinal Study in Individuals Unaware of Cognitive Decline

Structural brain changes underlying mild cognitive impairment (MCI) have been well-researched, but most previous studies required subjective cognitive complaints (SCC) as a diagnostic criterion, diagnosed MCI based on a single screening test or lacked analyses in relation to neuropsychological impairment. This longitudinal voxel-based morphometry study aimed to overcome these limitations: The relationship between regional gray matter (GM) atrophy and behavioral performance was investigated over the course of 3 years in individuals unaware of cognitive decline, identified as amnestic MCI based on an extensive neuropsychological test battery. Region of interest analyses revealed GM atrophy in the left amygdala, hippocampus, and parahippocampus in MCI individuals compared to normally aging participants, which was specifically related to verbal memory impairment and evident already at the first measurement point. These findings demonstrate that GM atrophy is detectable in individuals with amnestic MCI despite unawareness of beginning cognitive decline. Thus, individuals with GM atrophy in regions associated with verbal memory impairment do not necessarily need to experience SCC before meeting neuropsychological criteria for MCI. These results have important implications for future research and diagnostic procedures of MCI.

Predicting brain metastases for non-small cell lung cancer based on magnetic resonance imaging

In this study the relationship between brain structure and brain metastases (BM) occurrence was analyzed. A model for predicting the time of BM onset in patients with non-small cell lung cancer (NSCLC) was proposed. Twenty patients were used to develop the model, whereas the remaining 69 were used for independent validation and verification of the model. Magnetic resonance images were segmented into cerebrospinal fluid, gray matter (GM), and white matter using voxel-based morphometry. Automatic anatomic labeling template was used to extract 116 brain regions from the GM volume. The elapsed time between the MRI acquisitions and BM diagnosed was analyzed using the least absolute shrinkage and selection operator method. The model was validated using the leave-one-out cross validation (LOOCV) and permutation test. The GM volume of the extracted 11 regions of interest increased with the progression of BM from NSCLC. LOOCV test on the model indicated that the measured and predicted BM onset were highly correlated (r = 0.834, P = 0.0000). For the 69 independent validating patients, accuracy, sensitivity, and specificity of the model for predicting BM occurrence were 70, 75, and 66%, respectively, in 6 months and 74, 82, and 60%, respectively, in 1 year. The extracted brain GM volumes and interval times for BM occurrence were correlated. The established model based on MRI data may reliably predict BM in 6 months or 1 year. Further studies with larger sample size are needed to validate the findings in a clinical setting.

Inter-Method Discrepancies in Brain Volume Estimation May Drive Inconsistent Findings in Autism

Previous studies applying automatic preprocessing methods on Structural Magnetic Resonance Imaging (sMRI) report inconsistent neuroanatomical abnormalities in Autism Spectrum Disorder (ASD). In this study we investigate inter-method differences as a possible cause behind these inconsistent findings. In particular, we focus on the estimation of the following brain volumes: gray matter (GM), white matter (WM), cerebrospinal fluid (CSF), and total intra cranial volume (TIV). T1-weighted sMRIs of 417 ASD subjects and 459 typically developing controls (TDC) from the ABIDE dataset were estimated using three popular preprocessing methods: SPM, FSL, and FreeSurfer (FS). Brain volumes estimated by the three methods were correlated but had significant inter-method differences; except TIVSPM vs. TIVFS, all inter-method differences were significant. ASD vs. TDC group differences in all brain volume estimates were dependent on the method used. SPM showed that TIV, GM, and CSF volumes of ASD were larger than TDC with statistical significance, whereas FS and FSL did not show significant differences in any of the volumes; in some cases, the direction of the differences were opposite to SPM. When methods were compared with each other, they showed differential biases for autism, and several biases were larger than ASD vs. TDC differences of the respective methods. After manual inspection, we found inter-method segmentation mismatches in the cerebellum, sub-cortical structures, and inter-sulcal CSF. In addition, to validate automated TIV estimates we performed manual segmentation on a subset of subjects. Results indicate that SPM estimates are closest to manual segmentation, followed by FS while FSL estimates were significantly lower. In summary, we show that ASD vs. TDC brain volume differences are method dependent and that these inter-method discrepancies can contribute to inconsistent neuroimaging findings in general. We suggest cross-validation across methods and emphasize the need to develop better methods to increase the robustness of neuroimaging findings.

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.

The traveling heads: multicenter brain imaging at 7 Tesla

OBJECTIVE

This study evaluates the inter-site and intra-site reproducibility of 7 Tesla brain imaging and compares it to literature values for other field strengths.

MATERIALS AND METHODS

The same two subjects were imaged at eight different 7 T sites. MP2RAGE, TSE, TOF, SWI, EPI as well as B1 and B0 field maps were analyzed quantitatively to assess inter-site reproducibility. Intra-site reproducibility was measured with rescans at three sites.

RESULTS

Quantitative measures of MP2RAGE scans showed high agreement. Inter-site and intra-site reproducibility errors were comparable to 1.5 and 3 T. Other sequences also showed high reproducibility between the sites, but differences were also revealed. The different RF coils used were the main source for systematic differences between the sites.

CONCLUSION

Our results show for the first time that multi-center brain imaging studies of the supratentorial brain can be performed at 7 T with high reproducibility and similar reliability as at 3T. This study develops the basis for future large-scale 7 T multi-site studies.