Voxel-based relaxometry: a new approach for analysis of T2 relaxometry changes in epilepsy
Introduction
Since the early applications of nuclear magnetic resonance techniques to medicine, it was hoped that the tissue relaxation times, T1 and T2, would be a sensitive indicator of diseased tissue. However, the variability of these parameters across the wide range of varied pathologies and MRI acquisition methods precluded the fixing of standards for healthy and compromised tissue. Nevertheless, measurement of the relaxation times (“relaxometry”), especially T2, has been established as a reliable and objective tool for the quantitative measure of tissue abnormalities in conditions such as temporal lobe epilepsy (Jackson et al., 1993), Alzheimer's disease (Laakso et al., 1996) and various mental health disorders. In these cases, T2 relaxometry has been found to be useful for identifying abnormalities that are not obvious on visual assessment of MRI images. The measurement of the T2 time is commonly achieved by the relatively simple acquisition of spin echo images acquired at a range of echo times. The time constant of the exponential signal decay represents that rate of T2 relaxation. This report describes a novel approach to the analysis of this quantitative data for the purpose of the study of a group of patients with temporal lobe epilepsy.
Temporal lobe epilepsy is one of the most common medically intractable seizure disorders. Hippocampal sclerosis (HS) is the most frequent abnormality found in the temporal lobes of these patients. Surgical outcome is most often excellent and reliable identification of HS is therefore crucial. It has become well established that patients with unilateral HS display a unilateral increase in the hippocampal T2 time in comparison with controls Grünewald et al., 1994, Jackson et al., 1993, Kuzniecky et al., 1997. This change likely reflects gliosis of the sclerotic hippocampus (Briellmann et al., 2002). The range of variation of normal hippocampal T2 relaxation times is well defined and precise (Grünewald et al., 1994) and the technique is therefore very suited to detect subtle changes in patients.
To date, analysis of T2 relaxometry data has been carried out by manual placement of regions of interest (ROIs) over predefined areas of anatomy such as the hippocampus. However, the search space is then obviously limited to those areas chosen for ROI placement. In practice, it is unlikely that all the regions that might be affected by disease are assessed. It has well established that structural abnormalities in HS are not limited to the hippocampus Bronen et al., 1991, Lencz et al., 1992. An unbiased general search of the whole brain volume is, therefore, highly desirable. Moreover, the ROI-based approach is susceptible to intra-rater and inter-rater variability. This study introduces an alternative method of analysis that aims to overcome these issues. It applies voxel-based statistical parametric mapping techniques commonly used for the analysis of functional neuroimaging data to the analysis of the T2 relaxation data in an objective and automated manner. This procedure is evaluated in a study of patients with HS in which the results are compared with those obtained with the conventional ROI analysis.
Section snippets
Subjects
Nineteen patients (mean age 33 ± 13 years, 12 women) with typical HS determined from our comprehensive epilepsy surgery program at Austin Health were studied. Eight patients in this group had been diagnosed with left HS and the remaining 11 patients had right HS. In all patients, conventional MR imaging was initially performed at 1.5 T (GE LX) using a routine epilepsy protocol which did not include T2 relaxometry. Diagnosis of HS was based on three features: reduced hippocampal volume,
Conventional approach (ROI analysis)
ROIs were drawn on the T2-weighted images (2nd echo, TE = 58 ms) in four bilateral predefined areas including the hippocampal head, white matter of the anterior temporal lobe, amygdala, and white matter of the frontal lobe, as described previously by our group Briellmann et al., 1998, Jackson et al., 1993. The ROIs were copied onto the T2 maps. Abnormal T2 was defined in patients as a signal increase of more than 2 × SD of the mean of the controls where SD is the standard deviation (Briellmann
Conventional ROI analysis
As expected, all 19 HS patients had a significant increase in T2 values in the ipsilateral hippocampus (Table 1). Four subjects had additional slight increase of the contralateral hippocampal T2 signal. Ten subjects had increased signal in the white matter of one or both anterior temporal lobes, four subjects had amygdaloid signal changes and two subjects had frontal lobe white matter signal changes. Patients with right-sided HS appeared to have more pronounced signal increase in the
Discussion
This study introduces a voxel-based approach for the assessment of T2 relaxometry data. It demonstrates that the automated method based on statistical parametric mapping, developed to assess neuroimaging activation data and later adapted to assess morphometric changes can also be used to analyse quantitative tissue characteristics such as changes in relaxation times. An analogous voxel-based approach has been implemented for the analysis of diffusion tensor imaging (DTI) maps (Rugg-Gunn et al.,
Conclusion
The technique of VBR is capable of detecting T2 abnormalities in patients with HS. With use of modified statistical designs, other parameters of interest such as cognitive function or disease severity can be included in the analysis. The technique is promising as it may help to detect tissue abnormalities in brain areas, which are not usually investigated using predefined ROIs. Therefore, it may be used in the future to assess widespread signal abnormalities associated with focal abnormalities
Acknowledgements
The authors are very grateful for the support of Neurosciences Victoria (NSV) and the National Health Medical Research Council (NHMRC), Australia. This work was also supported by NHMRC grants 144105 and 226100.
References (23)
- et al.
Why voxel-based morphometry should be used
NeuroImage
(2001) Voxel-based morphometry should not be used with imperfectly registered images
NeuroImage
(2001)- et al.
A voxel-based morphometric study of ageing in 465 normal adult human brains
NeuroImage
(2001) - et al.
A comprehensive study of gray matter loss in patients with Alzheimer's disease using optimized voxel-based morphometry
NeuroImage
(2003) - et al.
Voxel-based morphometric analysis of gray matter in first episode schizophrenia
NeuroImage
(2002) - et al.
MR T2 relaxometry in Alzheimer's disease and age-associated memory impairment
Neurobiol. Aging
(1996) - et al.
Distributional assumptions in voxel-based morphometry
NeuroImage
(2002) - et al.
Voxel based morphometry—The methods
NeuroImage
(1997) - et al.
Hippocampal sclerosis in temporal lobe epilepsy demonstrated by magnetic resonance imaging
Ann. Neurol.
(1991) - et al.
Hemicranial volume deficits in patients with temporal lobe epilepsy with and without hippocampal sclerosis
Epilepsia
(1998)