Edited by: Fernando Cendes, Universidade Estadual de Campinas, Brazil
Reviewed by: Ian Brian Malone, University College London, United Kingdom; Christopher D. Whelan, University of Southern California, United States
†These authors have contributed equally to this work.
Specialty section: This article was submitted to Epilepsy, a section of the journal Frontiers in Neurology
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
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 (
Identifying brain morphological changes is a challenging task in neuroimaging studies. Voxel-based morphometry (VBM), introduced by Ashburner and Friston (
The brain volumetric changes in Alzheimer’s disease ( Tissue segmentation: the aim of this step is to classify the MRI scans into white matter (WM), gray matter (GM), and cerebrospinal fluid (CSF) images. Spatial normalization: this step contributes to the alignment of the images by registering the MRI images to a standard Montreal Neurological Institute (MNI) space A VBM analysis generally uses two registration methods: affine registration and non-linear registration. The affine registration is a linear mapping method that is used to achieve a global geometric transformation of the brain images, and this method is applied identically to each part of the image. With the non-linear registration, a finer-resolution match between images is achieved by allowing local transformations that adjust the different parts of each image in different manners. Modulating: this step contributes to the correction of changes in the volume of the segmented images by applying a linear deformation or a non-linear deformation. Smoothing: in the smoothing step, the segmented images are convolved with the use of an isotropic Gaussian kernel. This step helps to increase the signal-to-noise ratio, reducing the impact of misregistration between images and benefits on the normality of the statistics. Gaussian kernel sizes between 8 and 14 mm are usually used. Matrix design: here, a structural measures model and a general linear model (GLM) are used to test hypotheses regarding the brain structures. Modeling brain imaging data using a GLM is described in greater detail in Ref. ( Statistical inference: a statistical inference analysis is conducted to identify any significant differences between subject groups.
The details of a standard VBM procedure have been described (
The VBM8
The processing framework in a standard voxel-based morphometry analysis using Statistical Parametric Mapping software. GM, gray matter; WM, white matter.
In this study, we compared the results obtained using the CAT12 toolbox and those using the older software program VBM8 in whole-brain VBM analyses conducted to identify significant brain morphological abnormalities in five groups of adult subjects: (1) healthy controls (
All data used in this study were obtained from the National Center of Neurology and Psychiatry Hospital (Tokyo) for patients examined during the period from November 2013 through January 2017. The MRI scans were acquired from 3 T scanners manufactured by Philips (Best, The Netherlands) with the Digital Imaging and Communications in Medicine (DICOM) format with following protocol: repetition time/echo time: 7.12 ms/3.4 ms; flip angle: 10°; number of excitations: 1; 0.81 mm × 0.81 mm in plane resolution, 0.6-mm effective slice thickness with no gap, 300 slices, matrix of 260 × cm 320 cm; 26 cm × 24 cm field of view; acquisition time 4:01 min.
Table
Characteristics of the healthy controls and TLE patients.
HC ( |
RTLE-HS ( |
RTLE-no ( |
LTLE-HS ( |
RTLE-no ( |
|
---|---|---|---|---|---|
Age, years (mean ± SD) | 40.67 ± 10.97 | 42.07 ± 11.53 | 43.76 ± 13.78 | 38.00 ± 13.11 | 39.14 ± 13.12 |
Female/male | 12/16 | 14/12 | 14/16 | 17/8 | 14/13 |
The patients with an HS or non-HS diagnosis were assessed by visual inspections of MRI findings, and thus the patients with an HS diagnosis were recognized based on different criteria: ipsilateral reduced hippocampal volume; increased T2 signal on the hippocampus; and abnormal morphology (i.e., a loss of internal architecture of the stratum radiatum, a thin layer of WM that separates the dentate nucleus and Ammon’s horn). All participants gave written informed consent for their data to be used in this study and to be published. The study was approved by the Institutional Review Board at the National Center of Neurology and Psychiatry Hospital.
As the first step, we reviewed and converted the raw DICOM scans into the Neuroimaging Informatics Technology Initiative format, using MRICRON software.
In the present study, as part of the modulation step we performed a non-linear deformation on the normalized segmented images with both the VBM8 and CAT12 toolboxes. This modulation provides a comparison of the absolute amounts of tissue corrected for individual differences in brain size (
To identify the GM and WM morphological abnormalities in the present study’s TLE patients with and without HS, we used the GM and WM images. All segmented, modulated, and normalized GM and WM images were smoothed using 8-mm full-width-half-maximum Gaussian smoothing and then fed into a flexible factorial analysis in SPM8 and SPM12, separately.
In both the VBM8 and CAT12 toolboxes, the total GM volume, WM volume, and CSF volume were obtained, separately, on the basis of segmented images. The total intracranial volume (TIV) was calculated as the sum of the GM, WM, and CSF volumes for each toolbox, separately. As some authors have described using the age, gender, and head size of subjects in MRI studies (
We conducted an analysis of variance followed by Tukey’s multiple comparison test for the statistical analysis of the demographics among the five groups. We accepted probability values (
Figure
The significant alterations of regional gray matter (GM) volume revealed by the voxel-based morphometry (VBM) analyses using VBM8 versus CAT12. Family-wise error corrected at
Clusters of GM alterations shown by the VBM analysis using VBM8 versus CAT12.
Analysis | Location of peak voxels | Hemisphere | Cluster size (no of voxels) | Talairach coordinates ( |
MNI coordinates ( |
||
---|---|---|---|---|---|---|---|
VBM8 | (a) HC > LTLE-HS | – | – | – | – | – | – |
(b) HC > RTLE-HS | Hippocampus | R | 110 | 31 −21 −8 | 33 −19 −14 | 6.30 | |
(c) RTLE-no > HC | Amygdala | R | 935 | 25 −11 −10 | 26 −9 −17 | 7.21 | |
CAT12 | (a) HC > LTLE-HS | Hippocampus | L | 1,281 | −26 −18 −11 | −27 −16 −18 | 7.93 |
(b) HC > RTLE-HS | Hippocampus | R | 2,201 | 27 −28 −2 | 28 −27 −8 | 10.36 | |
(c) RTLE-no > HC | Amygdala | R | 1,256 | 29 −5 −19 | 30 −2 −27 | 4.8 |
The VBM analyses using the VBM8 and CAT12 procedures each revealed a significant increase in the GM in the right amygdala in the RTLE-no patients compared to the healthy controls. Both the VBM8 and CAT12 procedures showed no significant GM volume alterations in the LTLE-no patients compared to the healthy controls or significant differences in the reverse contrast between these groups.
The significant WM volume alterations in the five subject groups revealed by the VBM analyses using the VBM8 and CAT12 toolboxes are shown in Figure
The significant alterations of regional white matter (WM) volume shown by voxel-based morphometry (VBM) analyses using VBM8 and CAT12. family-wise error (FWE) corrected at
Clusters of WM alterations shown by the VBM analysis using VBM8 versus CAT12.
Analysis | Location of peak voxels | Hemisphere | Cluster size (no of voxels) | Talairach coordinates ( |
MNI coordinates ( |
||
---|---|---|---|---|---|---|---|
VBM8 | (a) HC > LTLE-HS | – | – | – | – | – | – |
(b) HC > RTLE-HS | – | – | – | – | – | – | |
CAT12 | (a) HC > LTLE-HS | Para hippocampal | L | 1,376 | −26 −28 −17 | −28 −27 −24 | 8.10 |
(b) HC > RTLE-HS | Para hippocampal | R | 1,291 | 27 −23 −14 | 28 −22 −21 | 7.98 |
The reliability of different automatic brain segmentation programs such as SPM, FreeSurfer, and FSL was recently evaluated in patients with Alzheimer’s disease or mild cognitive impairment (
In the VBM analysis using the older toolbox (i.e., VBM8), we observed a slight reduction in GM compared to the healthy controls only in the right hippocampus region of the RTLE-HS patients, whereas the VBM analysis using the newer program CAT12 revealed significant GM reductions at the left and right hippocampus regions in the LTLE-HS and RTLE-HS patients, respectively. Our VBM analysis with CAT12 results are in line with those of studies that reported ipsilateral mesial temporal volume reductions in the GM of TLE-HS patients compared to healthy individuals (
In our direct comparisons between the patients with a non-HS diagnosis versus the healthy controls, we observed a significant amygdala GM swelling in the RTLE-no patients in the VBM analysis using the VBM8 toolbox and in the same analysis using the CAT12 toolbox. This finding is in agreement with those of earlier studies that demonstrated TLE with amygdala enlargement (
Our comparison of WM alterations in our TLE-HS patients versus the healthy controls showed that ipsilateral mesial temporal WM reductions were identified by the VBM analysis using CAT12, whereas the VBM analysis using VBM8 did not detect any WM reduction in the LTLE-HS and RTLE-HS patients. The reason for this may be due to the improved and/or new segmentation algorithms incorporated into SPM12 compared to SPM8. Our VBM analysis with CAT12 findings are broadly consistent with studies describing ipsilateral WM abnormalities in TLE-HS patients compared to healthy controls (
One limitation of our study might be that the subject groups were gender imbalanced; the LTLE-HS group in particular was predominantly female, and the healthy controls were mostly male. In addition, given that statistical significance can sometimes be affected by various factors, we should pay careful attentions to interpreting the significance of the results.
The authors in Ref. (
To identify the brain morphological changes in TLE patients with and without HS, we performed two whole-brain VBM analyses—one using the toolbox VBM8 and the other using the CAT12 toolbox. These analyses provided disparate results. The results of the two analyses demonstrated that compared to the use of VBM8, a VBM analysis using the CAT12 toolbox identifies brain morphological abnormalities in patients with TLE that are more consistent with the literature- and pathology-based knowledge of TLE. The reason for this may be various improvements of the normalization and segmentation methods provided by SPM12 compared to the older program SPM8. It should be noted that the DARTEL process of normalizing to an averaged group template is not updated in SPM12 (
Our findings also demonstrate that brain morphological abnormalities in TLE patients identified using CAT12 are consistent with other studies that investigated the gray- and white-matter abnormalities in TLE using different methods such as optimized VBM (
In this work, FF, IB, DS and HM contributed equally.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
We thank the reviewers for their constructive comments. This research was partially supported by an Intramural Research Grant (27-8) for Neurological and Psychiatric Disorders of the National Center of Neurology and Psychiatry.
1
2
3
4
5
6
7