Data analysis
All data were processed at a vendor-provided ADW 4.6 workstation (GE Medical Systems).
To evaluate the contrast enhancement in the cochlear fluid quantitatively, regions of interest (ROI)s were separately set manually in the basal turn of the cochlea, brainstem and cerebrospinal fluid space of the cerebellopontine angleon 3D-FLAIR images for all patients (Fig.1). The SIR was calculated using the following equation:
SIR = SIcochlea/ SI brainstem (Eq.1)
, where SIcochlea represents the signal intensity of the cochlear basal turn, SIbrainstemis the brainstem. Additionally, the signal intensities in the bilateral cochlear regions (i.e., with and without affection) were, respectively, evaluated using a defined CNR equation:
CNR = ( SIcochlea- SIcerebrospinal fluid)/σbrainstem (Eq.2)
, where SIcochlea represents the signal intensity of cochlear basal turn, SIcerebrospinal fluid is the cerebrospinal fluid space, and σbrainstem is the standard deviation of the signal intensity of the brainstem.
To obtain the optimal scan angle, the anterior skull base was used as the baseline in sagittal direction (zero degree). The 3D-FLAIR images of 22 patients were reconstructed so that the relative maximum areas of the saccule, utricle and lateral semicircular canal can be displayed at the same level. The MR images were qualitatively analyzed twice by two experienced head and neck radiologists blinded to the clinical findings. In total,six measurements were performed and the mean level was calculated.
The degree of EH in the vestibule and cochlea was assessed by visual comparison of the relative areas of the non-enhanced endolymphatic space versus the contrast-enhanced perilymph space on 3D-FLAIR at 6 hours post-injectionin the axial plane, separately for the cochlea and the vestibule. The degree of cochlear hydrops was categorized as none, grade I, or grade II according to the criteria previously described by Baráth et al (15).
However, to evaluate the degree of the vestibular hydrops, we used the four-stage grading system by Bernaerts et al. (16).Vestibular EH and cochlear EH were reviewed twice by two experienced readers.
The SIR and CNR values were systematically compared between images acquired at different post-injection time using a Student’s t-test in SPSS 20.0 (IBM, Chicago, IL). The Student’s t-test was also applied to estimate the difference of the relative maximum areas of the saccule, utricle and lateral semicircular canal at different scan degrees applied. Additionally, the Cohen’s kappa statistic was used to estimate the interobserver agreement on detecting and grading EH. The significance threshold was set as P= 0.05.
Results
With double-dose injection, the 3D-FLAIR images acquired at 6 hours post-injection showed the strongest image contrast in the cochlea of 44 ears for twenty-two patients (Fig.2). The SIR and CNR values were separately found in the cochlea of the affected 22 ears 4 hours, 6 hours, 8 hours and 10 hours after agent injection (Table 1). Comparing the SIR and CNR values between 4 hours and 6 hours, and between 6 hours and 8 hours, respectively, significantly higher values were shown at 6 hours in both the affected and unaffected ears (P <0.05).
The optimal angles with the anterior skull base in sagittal view were obtained, ranging from 6.20to13.6 degrees (mean:10.74±2.24 degrees) and being parallel to the gyrus rectusnearly. The mean level of 10.74 degreecan thus be considered the optimal scan angle from the anterior skull base (Fig.3a,b), if compared to the scanning in parallel to the anterior skull base (P <0.01; Fig.3 c,d).
Cochlear hydrops was present in 14affectedears (grade I in 10 and grade II in 4 cases), and vestibular hydrops was detected in 17affectedears (grade I in 3, grade II in 10 and grade III in 4 instances). Cohen к test for interobserver agreement was 0.97 for cochlear findings and 0.98 for the vestibule (normal and abnormal).Seventy-seven percent (17/22) of clinically diseased ears had EH shown by MR imaging, whereas 9% (2/22) of clinically normal ears showed EH with MR imaging.
Discussion
In this study, we explored the optimal imaging parameters, including the optimal scan time and scan angle, of 3D-FLAIR in labyrinthine imaging for patients with vertigo and sensorineural hearing loss. The optimal scan time was found at 6 hours after injecting the contrast agent at double-dose and the best scan angle ranged from 6.20 to13.6 degrees (mean:10.74±2.24 degrees) was also obtained.
As the first main finding, we found that 3D-FLAIR can visualize endolymphatic hydrops 4 hours, 6 hours, 8 hours and 10 hours after double dose administration, and the significantly higher CNR and SIR values were shown at 6 hours in both the affected and unaffected ears. As the CNR value reflects the signal difference between the endolymphatic and perilymphatic regions, a significantly higher CNR in 3D-FLAIR images at 6 hours allowed a more robust distinguishing between the endolymphatic and perilymphatic regions if compared with the images acquired at other time points. That is to say, 6 hours after injecting the contrast agent can be considered an optimal scan time at double-dose. Nakashima et al (7) performed MR imaging 4 h after intravenous injection of a double dose of gadolinium in patients with Ménière’s disease at 3T. This findingwas however different fromthe 4 hours at a single-dose injection reported previously (17).The reason might be explained that similar to that the extent of cochlear enhancement varied with the concentration of the drug (6), the peak time of the drug concentration in the cochlea also varied.
Additionally, the obtained optimal scan angle of 3D-FLAIRwas about 10.74±2.24 degrees from the anterior skull basein sagittal view. With this, the relative maximum areas of the saccule, utricle and lateral semicircular canal can be displayed at the same level. As the bilateral semicircular canals are fully visible at the same level, we can visualize the saccule and utricle simultaneously, and judge whether EH herniation occurs in the lateral semicircular canal.3D images of the membranous labyrinth directly reconstructed from temporal bone sections suggest that in EH changes of saccular morphology are more sensitive than those of the utricle (18). Accordingly, Attyé et al. described the inversion of the saccule-to-utricle area ratio (SURI) as a qualitative marker (19). In addition, the vestibular semiquantitative, three-stage EH grading system proposed byBaráth et al. (15)and four-stage EH grading system proposed by Bernaerts et al. (16) were commonly used.With these semi-quantitative MR grading systems,the area of the utricle and saccule is requested to be obtained. It is thus important to display the relative maximum area of both at the same level.Besides these, Gurkov et al. (20)reported that EH herniation into the lateral semicircular canal (SCC) seemed to have a morphological correlation of an impaired caloric response in patients with MD.Therefore, it is beneficial to observe the degree of EH by simultaneously displaying the lateral SCC on the same level. When scanning the inner ear, we routinely place the scan line parallel to the lateral SCCbased on the image at coronal view.The scanning angle needed to be adjusted appropriately on the sagittal image, aiming to show the largest relative areas of the utricle and saccule, and lateral SCC on the same horizontal plane. Different scan angles in the sagittal positioning phase show different relative areas of the utricle and saccule, and lateral SCC accordingly.Therefore, a proper scanning angle is very important.The optimal scan time and scan angle can help us to accurately determine EH.
By histopathology, Rauch et al. (21) and Merchant et al. (22)found EH in 5% and 26% of temporal bones in patients without previous symptoms of MD. The former figure corresponds to an incidence of 9% in clinically “silent” ears in our series. EH was found on the clinically affected side in 17/22 ears (77%). Histopathologically, Fraysse et al. (23) described results in 93% of 21 affected ears.Our rate of EH is slightly lower than the previous ones, because the patients we included were not all MD patients. All nine definite MD patients (4) included showed EH (100%). The rate is higher than the results obtained in the former study (23).
One major limitation of this study is that the contralateral normal ears of the vertiginous patients were used as controls. Due to ethical reasons it is not allowed to observethe peak time of the drug concentration in the cochleaof a healthy populationwith gadolinium injection.