European Radiology

, Volume 23, Issue 1, pp 3–11

Anterior temporal lobe white matter abnormal signal (ATLAS) as an indicator of seizure focus laterality in temporal lobe epilepsy: comparison of double inversion recovery, FLAIR and T2W MR imaging

Authors

  • Emiko Morimoto
    • Department of Diagnostic Imaging and Nuclear MedicineKyoto University Graduate School of Medicine
    • Department of Diagnostic Imaging and Nuclear MedicineKyoto University Graduate School of Medicine
  • Tomohisa Okada
    • Department of Diagnostic Imaging and Nuclear MedicineKyoto University Graduate School of Medicine
  • Akira Yamamoto
    • Department of Diagnostic Imaging and Nuclear MedicineKyoto University Graduate School of Medicine
  • Nobuyuki Mori
    • Department of RadiologyTenri Hospital
  • Riki Matsumoto
    • Department of NeurologyKyoto University Graduate School of Medicine
  • Akio Ikeda
    • Department of NeurologyKyoto University Graduate School of Medicine
  • Nobuhiro Mikuni
    • Department of NeurosurgerySapporo Medical University
  • Takeharu Kunieda
    • Department of NeurosurgeryKyoto University Graduate School of Medicine
  • Dominik Paul
    • Siemens AG Healthcare Sector
  • Susumu Miyamoto
    • Department of NeurosurgeryKyoto University Graduate School of Medicine
  • Ryosuke Takahashi
    • Department of NeurologyKyoto University Graduate School of Medicine
  • Kaori Togashi
    • Department of Diagnostic Imaging and Nuclear MedicineKyoto University Graduate School of Medicine
Neuro

DOI: 10.1007/s00330-012-2565-4

Cite this article as:
Morimoto, E., Kanagaki, M., Okada, T. et al. Eur Radiol (2013) 23: 3. doi:10.1007/s00330-012-2565-4

Abstract

Objectives

To investigate the diagnostic capability of anterior temporal lobe white matter abnormal signal (ATLAS) for determining seizure focus laterality in temporal lobe epilepsy (TLE) by comparing different MR sequences.

Methods

This prospective study was approved by the institutional review board and written informed consent was obtained. Three 3D sequences (double inversion recovery (DIR), fluid-attenuated inversion recovery (FLAIR) and T2-weighted imaging (T2WI)) and two 2D sequences (FLAIR and T2WI) were acquired at 3 T. Signal changes in the anterior temporal white matter of 21 normal volunteers were evaluated. ATLAS laterality was evaluated in 21 TLE patients. Agreement of independent evaluations by two neuroradiologists was assessed using κ statistics. Differences in concordance between ATLAS laterality and clinically defined seizure focus laterality were analysed using McNemar’s test with multiple comparisons.

Results

Pre-amygdala high signals (PAHS) were detected in all volunteers only on 3D-DIR. Inter-evaluator agreement was moderate to almost perfect for each sequence. Correct diagnosis of seizure laterality was significantly more frequent on 3D-DIR than on any other sequences (P ≤ 0.031 for each evaluator).

Conclusions

The most sensitive sequence for detecting ATLAS laterality was 3D-DIR. ATLAS laterality on 3D-DIR can be a good indicator for determining seizure focus localization in TLE.

Key Points

Magnetic resonance imaging is widely used to investigate temporal lobe epilepsy.

Numerous MR sequences can show anterior temporal lobe white matter abnormal signal.

ATLAS on 3D-DIR can frequently indicate seizure focus laterality in TLE.

3D-DIR is more sensitive about ATLAS laterality than T2WI or FLAIR.

Keywords

EpilepsyMagnetic resonance imagingBrainTemporal lobeSequence inversion

Abbreviations and acronyms

ATLAS

anterior temporal lobe white matter abnormal signal

CSF

cerebrospinal fluid

DIR

double inversion recovery

EEG

electroencephalography

FLAIR

fluid-attenuated inversion recovery

MR

magnetic resonance

PAHS

pre-amygdala high signals

TLE

temporal lobe epilepsy

T2WI

T2-weighted imaging

2D

two-dimensional

3D

three-dimensional

Introduction

Temporal lobe epilepsy (TLE) is the most common form of epilepsy [1], and is the most likely to be medically refractory and highly responsive to surgical treatment [13]. Accurate localization of a seizure focus is essential for surgery [4], but presurgical examinations occasionally fail to provide seizure focus localization. Intracranial electroencephalography (EEG) with subdural electrodes offers a high degree of sensitivity for seizure focus localization [5], but electrode placement carries certain risks and is usually conducted unilaterally [6]. Seizure focus laterality should therefore be assessed with high precision prior to subdural electrode placement. Magnetic resonance (MR) imaging can provide detailed anatomical information non-invasively, but may not always detect abnormalities related to seizure foci [7, 8].

Previous reports have described anterior temporal lobe white matter abnormal signal (ATLAS) ipsilateral to the seizure focus on T2-weighted imaging (T2WI), and this finding has been considered as an indicator of seizure laterality [915]. ATLAS can be observed in 33 % of medically refractory TLE patients as an increased signal in the anterior temporal lobe white matter or a loss of grey–white matter demarcation [9], and is not observed in normal subjects [16]. However, Townsend et al. [17] showed that the T2 relaxation time of white matter in the anterior temporal lobe among TLE patients was increased bilaterally as often as unilaterally and ipsilaterally to the seizure focus. No previous studies have clarified how to evaluate bilateral ATLAS for determination of the seizure focus. Some methods have to be applied to visualize subtle signal changes suggesting ATLAS and to analyse bilateral ATLAS to enable correct diagnosis of seizure laterality.

Double inversion recovery (DIR) is a new MR sequence that better visualizes subtle signal changes in the grey and white matter by nullifying signals from both cerebrospinal fluid and white matter [18, 19]. Owing to its high contrast, DIR has been applied to epilepsy to detect hippocampal sclerosis at 3 T [20] or abnormal signal changes related to seizures at 1.5 T [21, 22], but no previous studies have focused on ATLAS visualized using DIR in TLE patients. In addition, normal findings at the anterior temporal lobe with DIR in particular have not been investigated, and they must be clarified in terms of defining ATLAS.

The purposes of this study were (1) to define normal MR findings of anterior temporal lobe white matter and (2) to study the diagnostic capability of ATLAS for determining seizure focus laterality in TLE patients by comparing two-dimensional (2D) and three-dimensional (3D) MR sequences, including 3D-DIR.

Materials and methods

This study was approved by the institutional review board. All subjects gave written informed consent prior to enrolment in the study.

MR image acquisition

All examinations of normal volunteers and patients with epilepsy were conducted using a 3-T whole-body MR system (MAGNETOM Trio, A Tim System; Siemens Healthcare, Erlangen, Germany) with a 32-channel head coil. The MR imaging protocol consisted of three single-slab 3D turbo spin-echo sequences (DIR, T2WI and FLAIR) and two 2D turbo spin-echo sequences (T2WI and FLAIR). All 3D sequences were implemented using a turbo spin-echo sequence with variable flip angle (work in progress provided by Siemens [23]). The acquisition parameters are summarized in Table 1. Axial (parallel to the anterior commissure–posterior commissure line) and oblique coronal (perpendicular to the long axis of the hippocampus) images were acquired using two 2D sequences. All 3D images were acquired in the sagittal orientation and reformatted to axial and oblique coronal images with a section thickness of 1.3 mm on a workstation (Multi-Modality Workplace; Siemens Healthcare, Erlangen, Germany).
Table 1

Sequence parameters for 2D-T2WI, 3D-T2WI, 2D-FLAIR, 3D-FLAIR and 3D-DIR

 

2D-T2WI

3D-T2WI

2D-FLAIR

3D-FLAIR

3D-DIR

Repetition time (ms)

3,200

3,200

12,000

5,000

7,500

Effective echo time (ms)

79

409

96

409

308

Inversion time (ms)

2,760

1,800

450, 3,000

Flip angle (°)

120

T2 variable

120

T2 variable

T2 variable

Bandwidth (Hz/Px)

180

781

182

781

789

Parallel imaging factor

2

2

2

2

2

Averaging

1

1

1

1

1

Turbo factor

11

141

19

141

109

Slice turbo factor

2

2

2

Matrix size

372 × 448

256 × 256

260 × 320

256 × 256

192 × 192

Field of view (mm2)

183 × 220

255 × 255

180 × 220

255 × 255

255 × 255

Phase resolution (%)

80

100

70

100

100

Number of slices

35

128

39

128

128

Slice thickness (mm)

3

1.3

3

1.3

1.3

In-plane resolution (mm)

0.6 × 0.5

1 × 1

1.0 × 0.7

1 × 1

1.3 × 1.3

Acquisition time (min)

1.44

3.26

2.48

5.22

8

Normal volunteer study

Subjects

Twenty-one healthy subjects (7 women, 14 men; mean age 34 years; range 24–65 years) were enrolled for a normal control study. No subjects had any history of epileptic attack, severe head trauma or other diseases of the central nervous system.

Image analysis

Two neuroradiologists (M.K. and E.M. with 16 and 7 years of experience in neuroradiology, respectively) independently evaluated all images to detect any signal differences from surrounding white matter of the anterior temporal lobe.

Patient study

Patients

To select clinically definite TLE patients for the study, 51 epilepsy patients (25 women, 26 men; mean age 33 years; range 2 months–79 years) were prospectively enrolled and examined with brain MR imaging at our institution between September 2009 and October 2010. Seizure foci were determined by the consensus of board-certified epileptologists A.I. and R.M., with 26 and 17 years of clinical experience in epileptology, respectively) at temporal, frontal, parietal and occipital lobes in 28, 6, 5 and 1 case, respectively. Foci were considered generalized or uncategorized in 3 and 8 cases, respectively. For focus localization, all available epilepsy examinations were used, including routine EEG studies according to the International 10-20 System with additional anterior temporal electrodes (T1 and T2) (n = 51), long-term video EEG monitoring with scalp electrodes (n = 16), fluorine 18-fluorodeoxyglucose positron emission tomography (n = 42), single photon emission computed tomography (n = 6), magnetoencephalography (n = 4) and routine MR imaging including 2D-FLAIR and 2D-T2WI (n = 51). MR images were used only to determine focal lesions suggestive of an epileptic focus, without evaluating ATLAS. Detailed neurological history and physical examinations were also referenced. Four patients were also examined using intracranial EEG with subdural electrodes. Among 28 patients with TLE, 7 patients were excluded because of inadequate MR image quality due to motion artefacts (n = 1) or presence of apparent abnormal lesions in the anterior temporal lobe other than hippocampal sclerosis, such as tumour (n = 2), inflammatory changes (n = 1), ectopic grey matter (n = 1) and postoperative changes (n = 2). The remaining 21 TLE patients (11 women, 10 men; mean age 39 years; range 16–79 years) were clinically diagnosed with right, left or bilateral seizure foci, in 5, 11 and 5 patients, respectively. Seizure foci in all 5 bilateral TLE patients were predominantly one-sided (right in 4 patients and left in 1 patient) on EEG and other examinations. These patients were thus sorted into right or left seizure focus groups according to the laterality of the major seizure focus. Finally, among the 21 TLE patients, 9 patients and 12 patients were considered to have right foci and left foci, respectively.

Image analysis

MR images of the 21 TLE patients were evaluated for laterality of ATLAS. Two neuroradiologists (M.K. and E.M.), blinded to clinical information, independently evaluated the laterality of ATLAS for each image set using a 5-point scale, as follows: 1 = definitely right; 2 = probably right; 3 = no laterality or no ATLAS; 4 = probably left; and 5 = definitely left. For evaluation, both acquired and reformatted images were used. Images of the five sequences for the 21 patients were independently selected and randomly presented to the two evaluators to prevent recall bias. ATLAS was defined as increased signal at the anterior temporal lobe white matter or a loss of the grey–white matter demarcation. Any signal differing from surrounding white matter at the anterior temporal lobe in the volunteer study was defined as normal appearance and was not regarded as ATLAS. In this session, the two neuroradiologists carefully evaluated only ATLAS laterality without referring to other MR findings.

Statistical analysis

The normal volunteer group and TLE patient group were compared using Welch’s t test and Pearson’s χ2 test for age and sex, respectively. Inter-evaluator agreement was evaluated using κ statistics. By defining scores 1 and 2 as indicative of seizure foci on the right, and scores 4 and 5 as indicative of seizure foci on the left, we examined the concordance rate between ATLAS laterality and the clinically defined side of the seizure focus. Score 3 was considered indeterminate and was sorted into a false group in any case. The proportion of patients with score 3 was calculated as the indeterminate rate for each evaluator. In typical analyses, sensitivity and specificity are calculated, but all cases in this study were clinically diagnosed as possessing epileptic foci, and no cases without epilepsy were included. Our results thus could not be analysed using a standard two-by-two table for sensitivity and specificity, as such calculations require the assumption that the analysis is designed for evaluating the presence or absence of a single disease [24]. Following these considerations, we defined “correct diagnostic rate” as the proportion of correct detections of seizure focus laterality, either to the left or right, in all cases. Differences in concordance between ATLAS laterality and clinically defined seizure laterality were compared in pairwise analyses between 3D-DIR and other sequences using McNemar’s test. A value of P < 0.05 after Bonferroni correction for multiple comparisons was considered significant. Data were analysed using a statistical package (IBM SPSS Statistics version 18.0; SPSS, Chicago, USA).

Results

Normal volunteer study

No normal volunteer showed abnormal signals in the anterior temporal lobe on any of 2D-T2WI, 3D-T2WI, 2D-FLAIR or 3D-FLAIR. On 3D-DIR, however, pre-amygdala high signals (PAHS) were detected in all normal volunteers (Fig. 1). PAHS were found in the area bounded anteriorly, medially and cranially by the cortex, laterally by the outer edge of the inferior horn of the lateral ventricle, posteriorly by the amygdala and caudally by the floor of the inferior horn of the lateral ventricle on oblique coronal images. No other signals differing from surrounding white matter except PAHS were seen on 3D-DIR in any volunteer.
https://static-content.springer.com/image/art%3A10.1007%2Fs00330-012-2565-4/MediaObjects/330_2012_2565_Fig1_HTML.gif
Fig. 1

A 26-year-old healthy male volunteer. Oblique coronal (perpendicular to the long axis of the hippocampus) image (a) and axial image (b) on 3D-DIR show faint pre-amygdala high signals (PAHS) in the white matter bilaterally (arrowheads)

Patient study

No significant difference in age (Welch’s t test, P = 0.27) or sex (Pearson’s χ2 test, P = 0.35) was found between TLE patients and normal volunteers. Representative signal changes and the extent of ATLAS in TLE patients are shown (Figs. 2 and 3). Inter-evaluator agreement of laterality scores was moderate to almost perfect for each sequence (2D-T2WI, 0.59; 3D-T2WI, 0.78; 2D-FLAIR, 0.62; 3D-FLAIR, 0.84; 3D-DIR, 0.96).
https://static-content.springer.com/image/art%3A10.1007%2Fs00330-012-2565-4/MediaObjects/330_2012_2565_Fig2_HTML.gif
Fig. 2

A 39-year-old woman clinically diagnosed with right temporal lobe epilepsy. Oblique coronal images on 3D-DIR (a), 2D-T2WI (b), 3D-T2WI (c), 2D-FLAIR (d), and 3D-FLAIR (e) reveal ATLAS in the white matter of the right temporal lobe (arrows). In this case, images from 3D-DIR and 3D-T2WI were determined as score 1 (definitely right) and the other images were determined as score 2 (probably right) by evaluator 1. The 2D-T2WI images were determined as score 2 (probably right) and the other images were determined as score 1 (definitely right) by evaluator 2. However, ATLAS could be detected most easily on 3D-DIR among these images

https://static-content.springer.com/image/art%3A10.1007%2Fs00330-012-2565-4/MediaObjects/330_2012_2565_Fig3_HTML.gif
Fig. 3

A 29-year-old woman clinically diagnosed with left temporal lobe epilepsy. Oblique coronal images on 3D-DIR (score 5 [definitely left]) (a), on 2D-T2WI (score 3 [indeterminate]) (b), on 3D-T2WI (score 4 [probably left]) (c), on 2D-FLAIR (score 3 [indeterminate]) (d), and on 3D-FLAIR (score 4 [probably left] by evaluator 1 and score 3 [indeterminate] by evaluator 2) (e) are shown. 3D-DIR showed bilateral ATLAS (arrow and arrowheads). In the right temporal lobe, small ATLAS was apparent in the subcortical white matter (arrow). In the left temporal lobe, ATLAS (arrowheads) was more prominent around PAHS than that on the right side

Results of concordance between ATLAS laterality and the clinically defined side of the seizure focus are summarized in Table 2. Correct diagnostic rates of 3D-DIR (0.71 and 0.76 for evaluators 1 and 2, respectively) were higher than those of any other sequences (2D-T2WI, 0.24 and 0.19; 3D-T2WI, 0.33 and 0.38; 2D-FLAIR, 0.19 and 0.14; 3D-FLAIR, 0.33 and 0.33 for evaluators 1 and 2, respectively). Correct diagnosis of seizure laterality was significantly more frequent on 3D-DIR than on any other sequences, including the other two 3D sequences (P = 0.004–0.031 and 0.001–0.031 for evaluators 1 and 2, respectively; see Tables 3 and 4 for details) using McNemar’s test after Bonferroni correction for multiple comparisons. Indeterminate rates (i.e. rates of score 3) were lower with 3D-DIR (0.24 and 0.19 for evaluators 1 and 2, respectively) than with any other sequences (2D-T2WI, 0.67 and 0.76; 3D-T2WI, 0.52 and 0.48; 2D-FLAIR, 0.76 and 0.81; 3D-FLAIR, 0.57 and 0.62 for evaluators 1 and 2, respectively, calculated from Table 2).
Table 2

Comparison of ATLAS laterality and clinically defined side of the seizure focus for each sequence with each evaluator

Evaluator 1

Evaluator 2

Clinical diagnosis

2D-T2WI

Clinical diagnosis

2D-T2WI

Right

Left

Indeterminatea

Total

Right

Left

Indeterminatea

Total

Right

4

0

5

9

Right

4

0

5

9

Left

2

1

9

12

Left

1

0

11

12

Total

6

1

14

21

Total

5

0

16

21

Clinical diagnosis

3D-T2WI

Clinical diagnosis

3D-T2WI

Right

Left

Indeterminatea

Total

Right

Left

Indeterminatea

Total

Right

4

0

5

9

Right

3

0

6

9

Left

3

3

6

12

Left

3

5

4

12

Total

7

3

11

21

Total

6

5

10

21

Clinical diagnosis

2D-FLAIR

Clinical diagnosis

2D-FLAIR

Right

Left

Indeterminatea

Total

Right

Left

Indeterminatea

Total

Right

4

0

5

9

Right

2

0

7

9

Left

1

0

11

12

Left

1

1

10

12

Total

5

0

16

21

Total

3

1

17

21

Clinical diagnosis

3D-FLAIR

Clinical diagnosis

3D-FLAIR

Right

Left

Indeterminatea

Total

Right

Left

Indeterminatea

Total

Right

4

0

5

9

Right

4

0

5

9

Left

2

3

7

12

Left

1

3

8

12

Total

6

3

12

21

Total

5

3

13

21

Clinical diagnosis

3D-DIR

Clinical diagnosis

3D-DIR

Right

Left

Indeterminatea

Total

Right

Left

Indeterminatea

Total

Right

7

0

2

9

Right

7

0

2

9

Left

1

8

3

12

Left

1

9

2

12

Total

8

8

5

21

Total

8

9

4

21

aIndeterminate: the cases categorized as score 3 (there was neither laterality nor ATLAS)

Table 3

Analysis using 2 × 2 table for 3D-DIR and each other sequence with each evaluator in terms of the concordance of ATLAS laterality and clinically defined side of the seizure focus

Evaluator 1

Evaluator 2

2D-T2WI

3D-DIR

2D-T2WI

3D-DIR

False

True

Total

False

True

Total

False

6

10

16

False

5

12

17

True

0

5

5

True

0

4

4

Total

6

15

21

Total

5

16

21

3D-T2WI

3D-DIR

3D-T2WI

3D-DIR

False

True

Total

False

True

Total

False

6

8

14

False

5

8

13

True

0

7

7

True

0

8

8

Total

6

15

21

Total

5

16

21

2D-FLAIR

3D-DIR

2D-FLAIR

3D-DIR

False

True

Total

False

True

Total

False

6

11

17

False

5

13

18

True

0

4

4

True

0

3

3

Total

6

15

21

Total

5

16

21

3D-FLAIR

3D-DIR

3D-FLAIR

3D-DIR

False

True

Total

False

True

Total

False

6

8

14

False

5

9

14

True

0

7

7

True

0

7

7

Total

6

15

21

Total

5

16

21

“True” was defined as a case with score 1 or 2 in clinically defined right seizure focus patients and as a case with score 4 or 5 in clinically defined left seizure focus patients. “False” was any other patient, including cases with score 3

Table 4

Results of McNemar’s test on 3D-DIR and each other sequence for each evaluator in terms of concordance with ATLAS laterality and clinically defined side of the seizure focus

 

P value

P value after correction*

Evaluator 1

Evaluator 2

Evaluator 1

Evaluator 2

3D-DIR vs 2D-T2WI

0.002

<0.001

0.008

0.002

3D-DIR vs 3D-T2WI

0.008

0.008

0.031

0.031

3D-DIR vs 2D-FLAIR

0.001

<0.001

0.004

0.001

3D-DIR vs 3D-FLAIR

0.008

0.004

0.031

0.016

*P values after Bonferroni correction for multiple comparisons

No cases were seen in which correct laterality was not detected by 3D-DIR but was detected by other sequences. In only 1 case, 3D-DIR detected ATLAS contralateral to the clinically defined side of the seizure focus according to both evaluators. Although 2D-FLAIR did not show ATLAS in that case, all other sequences also indicated ATLAS on the contralateral side. When differences between 2D and 3D scans were examined, 3D scans showed better laterality detection than 2D scans by 2 and 4 cases for T2WI and by 3 and 4 cases for FLAIR in evaluators 1 and 2, respectively.

Discussion

This is the first study to compare the detectability of seizure laterality by evaluating ATLAS in TLE patients among different sequences at 3 T. The present results indicate that 3D-DIR can detect seizure focus laterality in TLE patients according to ATLAS laterality with significantly higher concordance with final clinical diagnosis than 2D-T2WI, 3D-T2WI, 2D-FLAIR or 3D-FLAIR. Lower indeterminate rates with 3D-DIR compared with other sequences could also indicate a higher efficacy of 3D-DIR than other sequences in detecting faint signal changes of ATLAS and evaluating their extent. Although seizure focus localization is necessary for surgical treatment, no single examination is currently sufficient to determine focus laterality. The combination of several examinations thus remains necessary. ATLAS on 3D-DIR can non-invasively provide additional information suggesting seizure focus laterality, which will give MR imaging an important role to play in the diagnosis and management of TLE patients.

Several reasons are considered to contribute to the superiority of 3D-DIR. One is the high contrast of DIR. The grey matter is well delineated and little background signal is present at the white matter, facilitating detection of abnormal signals. Another reason is 3D acquisition with thin slices, which decreases partial volume effects of the grey matter and helps to differentiate it from ATLAS. However, superiority of 3D acquisition was observed in only a few cases on T2WI and FLAIR in our study. This was probably because T2WI and FLAIR could detect only large and apparent ATLAS due to reduced contrast between ATLAS and normal white matter. DIR is reportedly superior to other sequences such as T2WI and FLAIR in detecting white matter lesions in multiple sclerosis and malformations of cortical development [19, 21, 25].

Previous studies have shown that 3D-DIR at 1.5 T can detect abnormal signals in patients with partial epilepsy [21, 22], but the detectability of abnormal signals on 3D-DIR was higher in the present study. Salmenpera et al. [22] showed that 3D-DIR at 1.5 T revealed signal changes (ipsilateral, 14 %; bilateral, 7 %; contralateral, 5 %) without distinguishing between grey and white matter in TLE patients. The better results in the present study are considered attributable to the higher signal-to-noise ratio related to higher magnetic strength [26] and reduced partial volume effects due to thinner slice thickness.

In this study, ATLAS laterality could be detected ipsilateral to the side of clinical diagnosis in less than 24 % and 19 % on 2D-T2WI and 2D-FLAIR, respectively. In previous studies, ATLAS was detected in 33 % on 2D-T2WI [9] and in 75 % on 2D-FLAIR [27] among medically refractory TLE patients. The difference might be attributable to differences in patient populations. Most previous reports have studied patients who are refractory to medication and scheduled for surgery. On the other hand, we included TLE patients who can be managed with medication alone. Patients in this study were considered to have less prominent signal changes compared to those enrolled in previous studies [9, 27]. Such subtle changes were better detected by 3D-DIR, which suggests the wide applicability of 3D-DIR to less severe cases of TLE.

In all sequences other than 2D-FLAIR, ATLAS was only found contralateral to the clinically defined side of the seizure focus in 1 case in this study. In previous studies, patients have occasionally shown ATLAS bilaterally or contralateral to the side of seizure focus [10, 17]. An animal model study confirmed that unilateral kindling stimulation could decrease neuronal density in the contralateral temporal lobe in the same way as in the ipsilateral temporal lobe, because of anatomical connections between these regions [28]. Secondary damage to the contralateral side might be induced in TLE patients and contribute to contralateral ATLAS. The patient showing ATLAS only contralateral to the seizure focus in this study might have had an area more vulnerable to seizure damage in the contralateral temporal lobe white matter than in the ipsilateral temporal lobe white matter, or another lesion unrelated to epileptogenesis might have been present.

The pathological changes that correspond to ATLAS have not yet been clearly identified. One study has shown significantly lower average myelin staining in TLE patients with ATLAS than in TLE patients without ATLAS [15]. Other studies have detected the histopathological findings in ATLAS as gliosis [11, 12], corpora amylacea [12, 13], dilated perivascular spaces [12], minor inflammatory changes [12], oligodendroglial cell clusters [12, 13], heterotopic neurons [13, 14] and microdysgenesis [11]. Any of these pathological changes would contribute to ATLAS, but are not specific for ATLAS, because similar histological abnormalities were observed even in TLE patients without ATLAS [11, 13] and patients without epilepsy [12]. Schijns et al. [14] suggested that the area showing temporal lobe changes including ATLAS was at least not part of the epileptogenetic zone, because patients achieved good outcomes with selective amygdalo-hippocampectomy without resection of those changes.

To evaluate ATLAS, it is important to be aware of the presence of normal PAHS observed on 3D-DIR in all healthy volunteers. While the anatomical characteristics of PAHS remain unclear, neuronal connections reportedly exist between the amygdala and two other regions: the anterior temporal cortex and the olfactory bulb [29, 30]. Some differences have been speculated to exist in myelination or components of neural connections from surrounding white matter, which may contribute to the faint signal change on DIR [12, 15].

Several limitations must be considered when interpreting the results of this study. Firstly, no histological confirmation was obtained, although significantly high detectability of ATLAS on 3D-DIR was clearly demonstrated. This was because many patients in this study had less severe TLE that did not require surgical treatment. Previous reports have detected various pathological changes in the anterior temporal lobe related to TLE [1113, 15], but pathological changes related to ATLAS indicated by DIR have not been discussed. Future investigations of radiological and pathological correlations may reveal the histopathological basis of ATLAS on DIR. Secondly, the signal intensities of ATLAS are relatively faint and the quality of MR imaging is thus crucial. In addition to daily quality checks, signal heterogeneities that typically appear in images as regions of increased and decreased signal at 3 T were minimized after careful adjustment and use of a proper filter provided in the scanner [31]. In this study, the evaluators paid close attention to white matter signals to exclude residual effects of any such artefacts. Thirdly, pulsation of cerebrospinal fluid (CSF) has been reported to cause prepontine artefacts in the temporal lobe on coronal imaging, when the phase-encoding gradient is applied in left–right orientation [32]. In the present study, the phase-encoding direction of 2D sequences (T2WI, FLAIR) was right–left. Prepontine artefacts due to pulsation of CSF can arise in the temporal lobes on these 2D sequences. However, such artefacts do not influence the evaluations used in this study, because prepontine CSF artefacts typically appear in the dorsal temporal lobes, whereas we evaluated only the anterior temporal lobes, where CSF artifacts do not arise. Moreover, the inversion pulse for nullifying CSF applied in 3D acquisition largely reduces artefacts caused by CSF pulsation [33].

In conclusion, 3D-DIR was the most sensitive sequence for detecting ATLAS among the 2D and 3D sequences tested. ATLAS on 3D-DIR could provide a good indicator of seizure focus laterality in TLE.

Acknowledgements

We wish to thank Mr. Katsutoshi Murata and Mr. Masato Uchikoshi, of Siemens Japan K.K., for their contributions to the optimization of scan parameters.

Copyright information

© European Society of Radiology 2012