Introduction

In an animal study, it has been shown that Gd-DTPA absorbed through the round window membrane distributed mainly into perilymphatic space following intratympanic gadolinium-diethylene-triamine pentaacetic acid (Gd-DTPA) injection [1]. Enlarged endolymphatic space in patients with Meniere’s disease has been successfully recognized as the area with low signal intensity on 3D fluid-attenuated inversion-recovery (FLAIR) images obtained after intratympanic injection of Gd-DTPA [2]. Enlarged endolymphatic space was partly surrounded by perilymphatic fluid with high signal on 3D-FLAIR images.

Delineation of the boundary between cochlear endolymphatic space and surrounding bone was not clear, as the bone and endolymphatic space showed low signal intensity on 3D-FLAIR images.

To visualize endolymphatic space in the labyrinth as high signal while maintaining the separation from perilymphatic fluid space, a 3D inversion-recovery turbo spin echo sequence (3D-IR TSE) with an inversion time shorter than that of 3D FLAIR (to suppress the signal of perilymph fluid with higher Gd-DTPA concentration) might successfully suppress the signal of perilymph such that only endolymph would have positive signal, allowing the depiction of the border between bone and endolymphatic space.

This method would require the acquisition of two separate sequences to obtain the endolymphatic and perilymphatic anatomy. Mutual anatomical relationships between endo-and perilymphatic space could not be appreciated without the fusion of two separately obtained images. Instead of this, we assumed that the real reconstruction of inversion recovery data might, with a single sequence, be able to separately visualize endolymph, perilymph and bone in a clinically acceptable scan time. Features of the real reconstruction of inversion recovery data have been reported previously [3–5].

The purpose of this study was to evaluate whether 3D inversion-recovery turbo spin echo (3D-IR-TSE) with real reconstruction could separate the signals of perilymph (positive value), endolymph (negative value) and bone (near zero) by setting the inversion time between the null point of Gd-containing perilymph fluid and that of the endolymph fluid without Gd.

Materials and methods

Patients

Thirteen patients with clinically suspected endolymphatic hydrops (nine Meniere’s disease, two delayed endolymphatic hydrops [6, 7] and two acute low-tone sensorineural hearing loss, age 24-74 years, mean age 39.5 years, five men and eight women) underwent intratympatic administration of gadolinium-diethylene-triamine pentaacetic acid-bis (methylamide) (Gd-DTPA-BMA; Omniscan, Daiichi Pharmaceutical Co. Ltd., Tokyo, Japan). In two patients with delayed endolymphatic hydrops, intratympanic injection was performed for both the left and right ears. Thus, 15 ears were included in this study. These patients were scheduled for intratympanic injection therapy, with gentamicin to control severe vertigo or with steroid for the treatment of sensorineural hearing loss [8–11].

Clinical diagnosis was based on the patients’ history and various otological tests such as audiograms, electrocochleograms and vestibular-evoked myogenic potentials (VEMP).

Written informed consent was obtained from all patients. This study was approved by the institutional review board of our university hospital.

Intratympanic gadolinium injection

The detailed methods for intratympanic gadolinium injection have been reported previously [2].

Gd-DTPA BMA was diluted eightfold with saline (v/v 1:7). The diluted Gd-DTPA BMA was injected intratympanically through the tympanic membrane using a 23-G needle and a 1-ml syringe after the patient was placed in the supine position with his/her head turned approximately 30° away from the sagittal line toward the other ear. The diluted Gd-DTPA-BMA was injected until a backflow of fluid into the external ear was observed under a microscope. The amount of diluted gadolinium injected was 0.4 to 0.5 ml. After the injection, the patient remained in the supine position for 60 min with his/her head turned approximately 60° away from the sagittal line toward the other ear.

MR imaging

All scans were performed on a 3-T MRI scanner (MAGNETOM Trio; Siemens Medical Solutions, Erlangen, Germany) using a receive-only 12-channel phased-array coil. Twenty-four hours after intratympanic injection of diluted Gd-DTPA BMA, T1-weighted 3D-FLASH (fast low-angle shot) and conventional 3D-FLAIR (fluid-attenuated inversion recovery) imaging was performed. In addition, T2-weighted 3D-CISS (constructive interference in the steady state) imaging was performed to obtain reference images of labyrinthine fluid-space anatomy.

The parameters for 3D-FLASH were as follows: repetition time (TR) of 4.3 ms, echo time (TE) of 1.97 ms, flip angle of 10 degrees with RF spoiling, matrix size of 256 × 256, and 96 axial 0.8-mm-thick slices covering the posterior fossa with a 16-cm square field of view. The number of excitations was two, giving a total scan time of 2 min 51 s. The parameters for 3D-CISS were as follows: TR of 11.42 ms, TE of 5.71 ms, flip angle of 50 degrees, matrix size of 320 × 320, and 48 axial 0.8-mm-thick slices with a 16-cm square field of view. The number of excitations was one, and the scan time was 3 min 42 s.

The parameters for 3D-FLAIR and 3D-IR TSE were as follows: TR of 9,000 ms, TE of 134 ms, flip angle of 180 degrees (constant) for the turbo-spin-echo refocusing echo train, echo train length of 23, matrix size of 384 × 384, and 12 axial 2-mm-thick slices covering the labyrinth with a 16-cm square field of view, acquired using the GRAPPA parallel imaging technique with an acceleration factor of 2 [12]. The number of excitations was one, and the scan time was 14 min.

In a previous pilot study, a TI of 1,000 ms was selected for 3D inversion-recovery imaging of endolymphatic space, nulling the signal of Gd-containing perilymph. As the suppression of fluid without Gd could be achieved with a TI of 2,500 ms on 3D-FLAIR images, a TI of 1,700 ms (near the midpoint between 1,000 ms and 2,500 ms) was selected to assign positive longitudinal magnetization to perilymphatic fluid, negative longitudinal magnetization to endolymphatic fluid and zero magnetization to compact bone and air.

Endolymphatic hydrops on MR images

Endolymphatic hydrops was determined subjectively by an experienced neuroradiologist who has 20 years of experience in inner-ear MR imaging using the following criteria: Endolymphatic hydrops in the cochlea is thought to be positive if the negative signal area on 3D-real IR in the cochlear peripheral area is bulging toward the scala vestibuli. Endolymphatic hydrops in the vestibule is thought to be positive if the negative signal area on 3D-real IR in the vestibule is more than a quarter of the entire vestibule. These tentative criteria were established based on previous histological research in humans and animals [13–15].

Results

In all patients, low signal of endolymphatic space in the labyrinth on 3D-FLAIR was observed in the anatomically appropriate position, and it showed negative signal on 3D-real IR. The low signal area of surrounding bone on 3D-FLAIR showed near zero signal on 3D-real IR (Figs. 1, 2, 3). Gd-containing perilymphatic space showed high signal on 3D-real IR images. The spatial relationship between endolymphatic space and perilymphatic space can be well appreciated on 3D-real IR images. In nine Meniere’s disease patients and two patients with acute low-tone sensorineural hearing loss, the endolymphatic space was enlarged. In the two patients with acute low-tone sensorineural hearing loss, endolymphatic space in the upper turn of the cochlea seemed to be more enlarged, while perilymphatic space of the scala vestibuli seemed to be narrower compared to that of the lower turn. In two cases of delayed endolymphatic hydrops, both the left and right ears showed endolymphatic hydrops.

Fig. 1
figure 1

A 31-year-old man with acute low-tone senrorineural hearing loss in the right ear (average, 31 dB). All images were obtained 24 h after the intratympanic injection of Gd-DTPA. (a) 3D-FLAIR (9,000/134/2,500) shows enlarged endolymphatic space in the cochlea (arrows) as low signal areas; however, the boundary between endolymphatic space and surrounding bone is unclear. Mild enlargement of the endolymphatic space in the vestibule (short arrows) is observed. (b) A 3D-real IR sequence (9,000/134/1,700) visualizes severely enlarged endolymphatic space in the cochlea (arrows) and mildly enlarged endolymphatic space in the vestibule (short arrows) as negative signal intensity values, while the surrounding bone area has near zero signal intensity. This image allows the separation of perilymph space (high signal intensity), endolymph space and surronding bone on a single image. In this acute low-tone senrorineural hearing loss patient, perilymphatic space in the upper cochlear turn seems to be narrower than in the lower turn due to relatively severe endolymphatic hydrops in the upper turn

Full size image
Fig. 2
figure 2

A 33-year-old man with delayed endolymphatic hydrops. All images were obtained 24 h after the intratympanic injection of Gd-DTPA. (a) 3D-FLAIR (9,000/134/2,500) shows enlarged endolymphatic space in the cochlea (arrows), but not in the vestible. The boundary between endolymphatic space and surrounding bone is unclear. (b) A 3D-real IR sequence (9,000/134/1,700) visualizes severely enlarged endolymphatic space in all cochlear turns (arrows) as negative signal intensity values, while the surrounding bone area has near zero signal intensity. This image makes possible the delineation of the scala media (negative signal intensity; black on this image) from perilymph space (scala tympani and scala vestibuli with positive high signal intensity; white on this image) and surronding bone (near zero signal; gray on this image) on a single image

Full size image
Fig. 3
figure 3

A 42-year-old man with Meniere’s disease. All images were obtained 24 h after the intratympanic injection of Gd-DTPA. (a) 3D-FLAIR (9,000/134/2,500) shows severely enlarged endolymphatic space in the cochlea (arrows) and in the vestibule (short arrows). The boundary between endolymphatic space and surrounding bone is unclear. (b) A 3D-real IR sequence (9,000/134/1,700) visualizes severely enlarged endolymphatic space in all cochlear turns (arrows) and in the vestibule (short arrows) as negative signal intensity values, while the surrounding bone area has near zero signal intensity. (c) 3D-CISS (11.42/5.71/flip angle 50 degree) image shows the combination of endolymphatic and perilymphatic space

Full size image

No side effect relating to intratympanic administration of Gd-DTPA was observed.

Discussion

Separate visualization of perilymph and endolymph fluid space by MR imaging has been tried by several researchers [16]. Direct visualization of Reissner’s membrane using high spatial resolution imaging was successful in animals [17] and human cadavers [18, 19]; however, clear visualization in living human subjects has not been successful due to the limited spatial resolution of clinical MR imaging units [20, 21].

Intravenous administration of Gd-DTPA in healthy human volunteers resulted in a slight signal increase in the labyrinth after 4 h [22]. However, the separation between the endolymphatic space and perilymphatic space was not clear, probably due to an insufficient concentration of Gd in the perilymphatic space.

Intratympanic injection of Gd-DTPA and the utilization of 3D FLAIR at 3 T made the visualization of endolymphatic hydrops possible in vivo [2]. Intratympanically administered Gd-DTPA distributed mainly into perilymphatic fluid space, and not into endolymphatic space. However, it was difficult to differentiate the low signal of endolymphatic space on 3D FLAIR from surrounding bone. Especially endolymphatic space in the vestibule is difficult to delineate when it is enlarged [2].

To delineate endolymphatic space precisely and to allow the quantification of endolymphatic-space volume in the future, endolymphatic space needs to be visualized separately, not only from perilymphatic space, but also from bone and air. By changing the inversion time, endolymphatic space and perilymphatic space might be separately visualized as positive signal. This will allow the volume quantification of each space in the future.

Quantification of each space is an important goal in the future for the objective diagnosis of endolymphatic hydrops and for monitoring treatment efficacy. Even with current spatial resolution however, it will take 30 min to obtain both endolymphatic images and perilymphatic images separately. Furthermore, the spatial relationship between the two spaces cannot be appreciated without the fusion of two separately obtained images. Non-uniform distribution of Gd-DTPA in the perilymphatic space also would make it difficult to uniformly suppress the signal of perilymphatic fluid with a single inversion time.

The present method with 3D inversion recovery using real reconstruction divides the signal magnitude of the endo- and perilymphatic fluid into positive and negative; thus, the precise value of the inversion time is not as important.

One of the limitations of the present method is the relatively low spatial resolution in the slice direction due to limited scan time and signal-to-noise ratio. To improve the signal-to-noise ratio, further study is necessary to investigate strategies that reduce acquisition time (and improve imaging efficiency), such as shorter TR and longer echo-train length. A shorter acquisition time would allow us to obtain higher spatial resolution or a larger number of excitations.

Another limitation of this study is the lack of histological confirmation for the presence of endolymphatic hydrops. It is virtually impossible to obtain histological proof in human patients; therefore, animal experiments would be important to validate the results of this study. Clinical diagnosis of endolymphatic hydrops is based on patients’ history, symptoms and various otological tests. However, proof of endolymphatic hydrops in a particular patient is rarely obtained. Besides the animal experiments, the accumulation of follow-up MR-imaging studies using present methods, showing a correlation between the progression of image findings and clinical records, would be necessary to verify the utility of the present method using a 3D inversion-recovery sequence with real reconstruction.

Some invasiveness and relatively long scan time might prevent the wide spread of this method in the clinical setting, especially in the period of acute vertigo attack of Meniere’s disease. Rupture of Reissner’s membrane during the attack might cause the contamination of endo- and perilymph fluid, and severely decrease the significance of this method.

In conclusion, by optimizing the inversion time, endolymphatic space, perilymphatic space and surrounding bone or air can be separately visualized on a single image using a 3D inversion-recovery sequence with real reconstruction. This method might open the door for the objective evaluation of endolymphatic space disease in clinical settings.