Is MRI a viable alternative to CT/CBCT to identify the course of the inferior alveolar nerve in relation to the roots of the third molars?

The inferior alveolar nerve (IAN) is next to the lingual nerve, one of the most endangered anatomical structures during the surgical removal of mandibular third molars (MTM). Any impairment (i.e., temporary or persistent) of the IAN after MTM removal is reported with an overall incidence of 3.6%; a persistent neurosensory deficit after 6 months is observed in 0.9% [1]. The probability of IAN impairment after MTM removal depends on the spatial relation between the mandibular canal (MC) and the MTM [2] and lack of a bony MC wall increases the risk of IAN impairment [3]. Hence, preoperative imaging is essential and should include at least a conventional radiograph [4]. In cases with a close relation between the MC and MTM, the exact distance and spatial relation might not be correctly assessed by 2D images making additional 3D imaging (i.e., computed tomography (CT) or cone beam computed tomography (CBCT)) necessary [4, 5]. However, CT/CBCT diagnostics relies on the presence of a bony MC wall for locating the IAN. A histomorphometric analysis indicated that the integrity of the MC wall is dependent on the quality/density of the surrounding trabecular bone [6], which explains that in approximately every fifth patient the IAN/MC appears difficult to be identified based on CBCT images [7].

In this context, magnetic resonance imaging (MRI) has been recently proposed as a radiation-free 3D imaging method for MTM [8]. MRI benefits by directly depicting the neurovascular bundle (NVB) [9] and is therefore independent of the integrity of the MC wall [10, 11]. An ex vivo study has proven by superimposition of MRI and CT scans that the geometric accuracy of MRI to display the IAN is comparable to the CT technique [12]. However, the shape and volume of the NVB were underestimated in a more recent study tracing the IAN in CBCT and MRI [13], and in 7% of the cases with 1 T MRI, the relation between the MC and impacted MTM was not assessable due to magnetic susceptibility artifacts [14]. Nevertheless, by using 3 T MRI, which has been demonstrated being superior for detection of the courses of brain nerves compared to 1.5 T MRI [15], the accurate visualization of the mandibular branch of the trigeminal nerve has been corroborated [16, 17].

The increasing acquisition and use of 3D radiographic imaging within dentistry (e.g., for implant treatment planning [18] and preoperative assessment of MTM [19, 20]) is raising the overall ionizing radiation exposure of the population, i.e., an additional increase of the cancer incidence of 0.46 per year was estimated due to applying CBCT imaging prior to MTM removal [21]. Therefore, MRI could be a reasonable alternative for the planning of various surgical procedures within dentistry [22]. However, despite the increasing number of studies reporting on the possibilities of IAN visualization by MRI, only a few studies [8, 14, 23] have addressed whether MRI is a reliable diagnostic method to evaluate the course of the IAN in relation to the roots of the MTM. These studies, which have used MRI images alone [14] or MRI with either panoramic radiographs [8] or CBCT [23], have only considered the vertical relation between the MTM and the MC on a panoramic overview. However, from a clinical point of view, the whole 3D volume including the bucco-lingual relation is important and might affect the surgical approach [24]. Thus, the aim of this study was to assess the reliability of judging the spatial relation between IAN/MC and the roots of the MTM based on orthoradial slices of MRI or CT/CBCT images.

Taking the high frequency of MTM removal, the estimated increase of the cancer incidence due to applying CBCT imaging prior to MTM removal [21], and the questionable benefit to actually reduce the risk for neurosensory disturbance by CBCT compared to panoramic radiographs [35] into account, renders MRI as an interesting radiation-free 3D imaging method. The present study assessed the reliability of judging the spatial relation between IAN/MC and the roots of the MTM based on orthoradial slices of MRI or CT/CBCT images. A slightly lower inter- and intrarater agreement has been recorded for MRI compared to CT/CBCT images (i.e., κ of 0.74 vs. 0.81–0.84, respectively) when determining the course of the IAN/MC, but the recorded κ values still indicate a good reliability for the judgement of MRI images. Additionally, the agreement between the imaging modalities appeared independent of the examiner’s experience and MRI might offer advantages in selected cases with an accessory IAN. Herein, in about 8% of the cases, an accessory IAN was described in MRI images, which was not recorded in CT/CBCT images. The judgement of the direct contact between the IAN/MC and the roots of the MTM showed a higher disagreement between the imaging modalities (i.e., in almost 20% of the cases), which might be due to the fact that CT/CBCT and MRI is actually depicting the MC or the IAN, respectively.

In this context, CT/CBCT relies on the presence of a MC wall for locating the IAN, i.e., in cases with a minimal amount of bone surrounding the IAN, it can be difficult to identify the MC in CBCT images [7]. In contrast, MRI displays directly the NVB including the IAN. To date, there is no standard term regarding what is actually visualized by MRI. Nasel et al. [9] have previously stated that the neural and vascular structures within the MC could not be distinguished and thus referred to the term “NVB.” Since then, “NVB” [36‐39], “IAN” [8, 17, 40, 41], “mandibular nerve” [10, 12], or “MC” [11, 14, 23, 29] were used synonymously in various publications. The cross-sectional area of the NVB in the region of the MTM measured by histomorphometry is about 13.45 ± 2.23 mm²; thereof, the IAN and the inferior alveolar artery (IAA) represent 32.4 and 4.5% of the area, respectively [39]. The IAN itself is composed of a larger mental (2/3) and a smaller incisive (1/3) branch [42]. In regard to the structural dimensions within the NVB (i.e., with the IAN being 6 to 7 times larger than the IAA), it is reasonable to assume that the hyperintense signal in MRI is mostly expressed by the IAN. However, the NVB as total does not necessarily fill the whole volume of the MC (Fig. 5). Additionally, the structures within the NVB might change their spatial relation to each other as it has been shown for the IAA, i.e., the IAA changes its position within the MC on average about 4 times being at the MTM most often in a cranial position [43]. Altogether, this might explain at least partly the higher disagreement observed herein between the imaging modalities for the judgement of a direct contact between the IAN/MC and the roots of the MTM. For example, the IAN might be located within the MC more distant to the roots and not displaying a direct contact in the MRI images, although no bone tissue is displayed between the MC and the roots in CT/CBCT images. Alternatively, the IAN might be located close to the roots and the hyperintense signal interferes with the judgement of the presence/absence of hypointense bone tissue between the IAN and the roots in MRI images. Finally, the superior visualization of hard tissue in CT/CBCT images (e.g., the border between cancellous bone and the root surface) might also play a role.

Any disagreement between the imaging modalities for the course of the IAN/MC was primarily among adjacent regions, i.e., primarily buccal/lingual vs. inferior and in some cases interradicular vs. inferior (Table 2). Specifically, the disagreement can be explained for the most part by the “borderline” cases with the IAN/MC matching 2 groups/classes (e.g., lingual and inferior). It was defined herein to allocate the IAN/MC to the aspect, which harbored the greater cross-sectional area; however, this implies naturally a certain risk for inter- and intrarater disagreement. Further, the fact that either the MC or the IAN are displayed in CT/CBCT or MRI images, respectively, probably contributes to a potential disagreement among adjacent groups/classes, i.e., as discussed above, that the IAN does not necessarily fill the whole volume of the MC and therefore might be depicted in a slightly different position in MRI images compared to CT/CBCT images. Nevertheless, only in very few cases (i.e., in 0.7% of all judgements) that the IAN/MC was clearly “misjudged” between the imaging modalities (i.e., a confusion between buccal/lingual, buccal/interradicular, or lingual/interradicular), which underlines MRI as viable alternative to CT/CBCT. Additionally, the confusion buccal vs. lingual appeared only twice by the examiner with the least experience. Finally, CT/CBCT was considered herein as a “gold standard” as it is the main 3D imaging technique being used if a close relation is suspected on panoramic radiographs [4]. However, one should keep in mind that the “misjudgment” might actually not necessarily occur when judging the MRI images. In fact, MRI is the most commonly used technique to directly visualize the IAN [22] and to precisely localize the IAN including pathologic processes [9, 44‐46].

To the best of our knowledge, a direct comparison to previous studies is not possible, as this is the first comparison of MRI and CT/CBCT images for a description of the relation of IAN/MC to MTM based on a 3D volume data set. In general, this relation is ideally expressed by both the course of the IAN in relation to the roots of the MTM (i.e., buccal, lingual, inferior, interradicular) and the presence/absence of bone tissue between the IAN and the roots of the MTM [28]. For example, a previous study [8] compared amongst other parameters the vertical position of the MTM in relation to the IAN based on MRI and panoramic radiographs. They reached an agreement in only 56% of the cases and the interrater agreement was poor (κ_w = 0.142). However, from a clinical point of view, MRI will not replace the panoramic radiograph as primary diagnostic method, but—as shown herein—MRI could replace CT/CBCT in cases, where the panoramic radiograph indicated the need for an additional 3D imaging technique to display in detail the vertical and bucco-lingual relation between the IAN/MC and the roots of the MTM without additional ionizing radiation exposure.

Altogether, MRI presents several advantages compared to CT/CBCT. Besides the obvious ones, such as radiation-free recording and the independency of the presence of a bony MC wall, MRI might be advantageous in displaying an accessory IAN and/or the lingual nerve. Herein, in 7 out of 87 cases, an accessory IAN was described in the MRI images, which was not visible in the CT/CBCT images (Fig. 4). However, based on the present data, one cannot exclude the possibility of false positive results in the judgement of MRI images for an accessory IAN. Future studies including histological examinations would be required to determine the truth regarding the presence/absence of accessory IAN. In general, the actual lack of a “gold standard” should be considered when interpreting the results herein, i.e., one does actually not know whether the judgement based on MRI or CT/CBCT images is the truth. Further, a recent study explored the feasibility of MRI images to detect and follow the lingual nerve [47]. Visualization of the lingual nerve was not part of the present study and the best sequence to depict the lingual nerve requires future studies. In general, visualization of the lingual nerve prior to MTM removal sounds tempting, as a temporary neurosensory disturbance (up to 13 weeks) is reported in 2% of the cases [48] and the lingual nerve is not visible in CT/CBCT images. However, one should also keep in mind the limitations of MRI with the most important ones being the general accessibility and the fact that it is a more expensive and time-consuming examination. Specifically, the cost-effectiveness might be difficult to prove, especially as even CT/CBCT still lacks the proof of a benefit to actually reduce the risk for neurosensory disturbance compared to panoramic radiographs only [35].

When evaluating the results of the present study, several limitations should be taken into consideration. First, for both imaging modalities, the orthoradial MPR have been produced prior to any assessment by different persons. Specifically, 2 experienced radiologic technologists were responsible for the CT/CBCT scans and the dental student together with the radiologist for the MRI scans. It can be discussed whether orthoradial MPR performed by different persons may affect the reliability; however, as all examiners judged the same reconstructions, any effect should be minor. In this context, if the examiners would have prepared the MPR themselves, it might have lowered the reliability for both MRI and CT/CBCT. However, this might primarily affect the “borderline” cases with the IAN/MC matching 2 groups/classes. Second, the inclusion of a dental student could be considered as a weakness, yet it allowed us to get an idea about the effect of the examiner’s experience. Specifically, it was interesting to see that the agreement between the judgements of both imaging modalities appeared independent of the examiner’s experience for the outcome parameters (i.e., course of the IAN/MC and direct contact between the IAN/MC and the MTM). Hence, at least for this specific topic, already a short calibration session prior to committing this study enabled a dental student to read the MRI and CT/CBCT images with an agreement comparable to the 2 other examiners (i.e., an oral surgeon and a radiologist). This in turn allows a careful conclusion that one can probably expect a high learning curve when starting to incorporate MRI as an alternative to CT/CBCT for assessing the relation between IAN and MTM in daily practice. Third, 68 MTM were assessed herein by CT and only 19 cases by CBCT. As CBCT provides a slightly higher resolution compared to CT [49, 50], an assessment based on CBCT images only might increase the reliability of the “standard” method. Finally, the present study did neither aim to develop new sequences for head and neck MRI nor to compare different sequences. Instead, the investigation was based on sequences of the standard protocol, which are usually available in radiological institutions. Hence, it would be an interesting topic of future studies whether the depiction of the IAN could be further improved, which in turn might improve the reliability of judgements based on MRI images.

In conclusion, the present study demonstrated the suitability of MRI with fat-saturated axial PD T2 TSE and coronal PD TSE sequences for the assessment of the IAN course in relation to the roots of the MTM prior to surgery as a radiological method free of ionizing radiation. Specifically, although the judgement of MRI images showed a slightly lower inter- and intrarater agreement in terms of determining the course of the IAN/MC compared to a judgement based on CT/CBCT images, the reliability can be considered as good. Additionally, the agreement appeared overall independent of the examiner’s experience and MRI might have advantages in seldom cases with a second, accessory IAN.