Accuracy and reliability of magnetic resonance imaging in orthodontic diagnosis and treatment planning—a systematic review and meta-analysis

Summary

Background

Magnetic resonance imaging (MRI) is a non-ionizing imaging technique. Using MRI in dentistry may potentially lower the general radiation dose of the examined population, provided MRI can replace various radiation-based images. Furthermore, novel MRI imaging modalities for three-dimensional and two-dimensional cephalometrics have recently been developed for orthodontic diagnosis.

Objectives

This systematic review aimed to determine the diagnostic accuracy and reliability of MRI in orthodontic diagnosis and treatment planning.

Search methods

An electronic search was conducted on 20 November 2022 in the following databases: PubMed, LILACS, Web of Science, EMBASE, Scopus, and Cochrane. The search was updated on 30 August 2023. Furthermore, a grey literature search was performed in Google Scholar and Open-Grey.

Selection criteria

This review included descriptive, observational, cohort studies, cross-sectional, case-control studies, and randomized/non-randomized trials related to the research question. The study excluded studies related to patients with syndromes, chronic diseases, craniofacial anomalies, or bone diseases.

Data collection and analysis

The included studies were quality assessed using the “Joanna Brigg’s Critical Appraisal Tool for diagnostic test accuracy”. The GRADE approach for non-randomized studies was used for strength-of-evidence analysis.

Results

Eight of the 10 included studies compared MRI with either cone beam computed tomography or lateral cephalogram and found a high intra- and inter-rater agreement for landmark identification. The risk of bias was high in four studies, moderate in three, and low in three studies. Homogeneity was lacking among the included studies in terms of MRI imaging parameters and sample characteristics. This should be taken into consideration by future studies where uniformity with respect to these parameters may be considered.

Conclusions

Despite dissimilarity and heterogeneity in the sample population and other methodological aspects, all the included studies concluded that MRI enjoyed considerable intra- and inter-examiner reliability and was comparable to current diagnostic standards in orthodontics. Furthermore, the studies agreed on the innovative potential of MRI in radiation-free diagnosis and treatment planning in orthodontics in the future.

Registration

CRD number: CRD420223XXXXX

Introduction

Diagnostic imaging is essential in determining the presence and extent of various orthodontic malocclusions and pathologies. Two-dimensional (2D) imaging has been used for cephalometric analysis for decades [1]. 2D imaging modalities are performed on various age groups, including children, adolescents, and adults. As 2D modalities suffer from various intrinsic limitations—including distortion and superimposition of multiple structures and landmarks—diagnosis has often been compromised [2]. The introduction of cone beam computed tomography (CBCT) in orthodontics has overcome these limitations and CBCT has become a popular tool in orthodontic imaging, diagnosis, and treatment planning [3]. At present, the CBCT imaging modality is used for 2D and three-dimensional (3D) imaging to assess skeletal and dental relationships in orthodontics. However, cephalograms and CBCTs come at the cost of ionizing radiation exposure, which is detrimental to the younger population [4].

Magnetic resonance imaging (MRI) has become an indispensable imaging modality in medicine and orthopedics [5]. It has been used to diagnose various soft and hard tissue disorders and is currently being investigated for its utility in diagnosing a wide range of diseases [6]. The main advantage of MRI is that it relies on non-ionizing radiation, which does not impose radiation hazards on the exposed population [7]. In dentistry, MRI may, therefore, potentially lower the general radiation dose to the examined population, provided the modality can replace several radiation-based images [7–10].

Various studies have reported that medical MRI units present mineralized tissues with the desired accuracy [9–11]. Recently, new studies have tested its utility in dentistry to diagnose various dental and skeletal pathologies [9, 10, 12]. Furthermore, novel MRI imaging modalities have been developed for 3D and 2D cephalometrics for orthodontic diagnosis [13–16]. Several studies have used MRI to assess the accuracy and reliability of MRI for landmark identifications using 2D cephalometric analysis and 3D volumetric analysis of hard and soft tissue. Even so, the literature provides no definitive answer as to the accuracy and reliability of MRI in orthodontic diagnosis and treatment planning. Therefore, this systematic review aimed to assess current knowledge on diagnostic accuracy and reliability of MRI in orthodontic diagnosis and treatment planning.

Materials and methods

This systematic review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA) guidelines [17] and the Cochrane Handbook for Systematic Reviews [18]. Before commencement, the study protocol was registered in the International Prospective Register of Systematic Reviews (PROSPERO CRD number CRD42022377204).

Study identification

The search strategy is described in detail in Supplementary File 1. It included relevant words and MeSH terms to delimit the population (P)—patients of any age or gender requiring any form of orthodontic treatment, intervention (I)—2D and 3D analysis based on MRI, comparator (C)—2D and 3D diagnosis and treatment by conventional cephalogram and/or CBCT or none, and outcome (O) of the research question—accuracy and reliability of MRI.

An electronic search was conducted on 20 November 2022 in PubMed, LILACS, Web of Science, EMBASE, Scopus, and the Cochrane library without any restrictions of language or publication year. The search was updated on 30 August 2023, and again on 26 February 2024 (Supplementary File 1). The search strategy was modified to suit the requirements of each database. Grey literature was searched in Google Scholar and Open-Grey. Two authors (SS and AA) independently performed hand searching in journals relevant to orthodontics and dentomaxillofacial radiology (European Journal of Orthodontics, American Journal of Orthodontics and Dentofacial Orthopedics and Dentomaxillofacial Radiology) (Supplementary File 1). Furthermore, the reference lists of all included articles were revised manually to identify additional relevant literature. The search results are presented in a PRISMA chart (Fig. 1). The searches were saved electronically in EndNote online (Clarivate Analytics), which was also used to remove duplicates.

During the first step of scrutiny, titles, and abstracts were evaluated based on the pre-determined inclusion and exclusion criteria. The study included descriptive, observational, cohort studies, cross-sectional, case-control studies, and randomized/non-randomized trials related to the research question. The following study types were excluded: case reports, case series, letter to the editor, comprehensive, and systematic reviews. Furthermore, we excluded studies related to patients with syndromes, chronic diseases, or craniofacial anomalies, patients having diseases affecting the skeletal system and ossification of bone, studies with inadequate information and those related to evaluation of the temporomandibular joint or obstructive sleep apnea (Supplementary File 1). The second step of scrutiny included a full-text evaluation of the selected articles, after which the remaining articles were included in the qualitative synthesis. The studies excluded after full-text assessment and the reasons for their exclusion are listed in Supplementary File 2. Two reviewers (SS and AA) performed each step independently. In case of disagreement, a consensus was established by consulting a senior reviewer (PBS).

Data extraction

A self-designed, pre-piloted data extraction sheet was used for data extraction. Data were extracted based on the following broad categories: (i) study characteristics, (b) sample characteristics, (c) characteristics of the MRI machine used, (d) characteristics of the comparison group, (e) various parameters compared, and (f) results of the analysis. All these categories are described in detail in various tables and appendices throughout the systematic review. The authors of the studies were contacted for additional details when as needed. Data were extracted independently by two reviewers (SS and AA) and, in case of disagreement, a senior reviewer was consulted (RSN).

Quality assessment

The quality of the included studies was assessed using the Joanna Brigg’s Critical appraisal tool for diagnostic test accuracy [19]. The GRADE approach for non-randomized studies was used for the analysis of the strength of evidence of each outcome [20]. These assessments were performed by two reviewers (SS and AA) independently. In case of disagreement a senior reviewer (PBS) was consulted.

Qualitative synthesis and meta-analysis

The data derived from the included studies were tabulated and discussed in the research group to develop the most appropriate strategy for its presentation. Furthermore, the possibilities of performing the meta-analysis were explored. All qualitative variables were converted into the same measuring unit for adequate qualitative or quantitative comparison. The outcomes that could be derived from two or more studies were further evaluated and subjected to meta-analysis in a random-effects model in STATA-14 (Stata Statistical Software, Release 14, StataCorp LP, College Station, TX, USA). Where comparisons were made, the effect measures for linear and angular variables were envisaged to be a standard mean difference (MD). In contrast, for each diagnostic method, the linear and angular values were reported in millimeters and degrees, respectively. Furthermore, an attempt was made to analyze the agreement scores for different modalities by using inter-class correlation (ICC) values. Heterogeneity was assessed using I² values and significance was recorded. For the assessment of publication bias, we planned to create funnel plots from meta-analyses, which included a minimum of ten studies.

Results

Study characteristics

Table 1 presents the study characteristics of the ten included studies [21–30]. All studies were published within the period of 11 years from 2011 [30] to 2021 [21]. Eight studies were conducted in Europe [21, 22, 24–29] and two were conducted in Asia [23, 30]. The studies compared the diagnostic reliability and accuracy of MRI in orthodontic diagnosis with gold standard methods such as 3D cephalometric analysis, CBCT [22, 25, 26, 30], or 2D cephalometric analysis [21, 23, 28, 29]. One of the studies explored the dimensional accuracy of MRI images after image registration and fusion of CBCT and MRI using various software programs [30]. Another study presented the utility of a novel gradient echo MRI sequence popularly known as “Black bone” in cephalometric diagnosis [29]. One of the most recent studies introduced a novel scan protocol with an ultra-short echo time [21]. Five of the included studies [21, 22, 24, 28, 29] were funded by various different groups, among which three were funded by the Dietmar Hopp Foundation [22, 24, 28]. One of the studies was non-funded [26]. Finally, four studies did not report on funding [23, 25, 27, 30].

Agreement between the reviewers

Excellent agreement was found between the reviewers who performed the screening of the articles in two stages (Cohen’s Kappa—0.96 for the first and 0.92 for the second stage). Good agreement was found between the two reviewers during data extraction (Cohen’s Kappa range 0.76–0.89), risk of bias analysis (Cohen’s Kappa range 0.82–0.86), and GRADE assessment (Cohen’s Kappa range 0.84–0.88).

Risk of bias

Agreement was good between the two reviewers during the assessment of risk of bias of the included studies (Cohen’s Kappa 0.80–0.92) (Fig. 2). The risk of bias was found to be high for four studies [21, 23, 26, 30], and moderate for three studies [27–29]. The main paucity observed in the included studies concerned sampling, confounding factors related to the assessment of the index tests, interpretation of the results, and the interval between the index test and reference test. A low risk of bias was recorded for three studies [22, 24, 25].

Sample characteristics

The characteristics of the study populations of all the included studies are described in Supplementary File 3. Six of the studies had a sample size exceeding 10 [22–26, 28]. The largest sample was that of the study by Heil et al. (n = 20) [28]. All of the studies included adults as a sample population except for one study, which included adolescents [28]. Participant age ranged from 13.95 ± 5.34 years [28] to 44 ± 16.2 years [25]. Seven studies followed well-defined inclusion and exclusion criteria [22–29]. Details of the modality comparison are provided in Supplementary File 3.

Overall, eight studies compared MRI landmark identification with CBCT or lateral cephalograms [21–23, 25, 26, 29, 30]. Two studies performed inter-rater and intra-rater reliability testing for 3D and 3D MRI images [24, 27]. Among the studies that compared MRI cephalometric landmark assessment with CBCT, four compared 3D cephalometric landmark identification in MRI with that of CBCT [24–27], whereas two studies used multiplanar images from CBCT and compared these with multiplanar images of MRI for landmark identification [22, 30]. The remaining four studies compared conventional lateral cephalograms with lateral cephalograms derived from MRI [21, 23, 28, 29]. In addition, the study by Grandoch et al. compared various anatomical structures in the craniofacial region [25]. The details of raters were mentioned in six of the studies [21, 22, 24, 26–28], where the largest number of raters were found in the study by Abkai et al. 2021, that is, 40 orthodontists with 15 years of experience in cephalometric analysis [21]. Only experienced radiologists served as assessors in two studies [22, 24], whereas only experienced orthodontists rated results in the study by Maspero et al. [26]. One of the studies included both radiologists and an orthodontist as raters [28]. Only one rater was used in the study by Juerchott et al.; a radiologist with 5 years of experience [27].

Despite dissimilarity and heterogeneity in the sample population and other methodological aspects, all the included studies reported a considerable intra- and inter-rater reliability of MRI, concurred that this modality was comparable to current conventional standards, that is, 3D CBCT, multiplanar reconstruction (MPR), and lateral cephalometric radiograph (LCR) images. The included studies agreed on its innovative potential for radiation-free diagnosis and treatment planning in orthodontics. (Table 1)

MRI scanner characteristics

MRI scanner characteristics are listed in Table 2. Six the 10 studies used a 3-tesla magnet MRI system [21, 22, 24, 26–28] and three used a 1.5-tesla MRI system [23, 25, 29]. One study failed to specify which MRI system was used [29]. Five of the studies had MRI devices from Siemens Health Care (Magnetom Trio/Magnetom Avanto) [22–24, 27, 28], whereas three were from Phillips Health Care (Phillips Achieva) [21, 25, 26]. Eley et al. used an MRI device manufactured by GE Medical Systems [29]. T1- and T2-weighted images were obtained in three studies, T1-weighted images in four studies, and T2-weighted images in one study. The study by Abkai et al. from 2021 introduced a method, which utilized a novel ultra-short echo modality, providing MRI projections within a shorter time interval to acquire cephalometric images [21].

The MRI examinations were conducted using various MRI channel surface coils. Two studies used a 15-channel MRI coil [22, 24], whereas there were variations between the unit of MRI coil used in other studies, such as an 8-channel MRI coil [21], a 4-channel MRI coil with modifications [25], a 16-channel MRI [27] coil, or 6-channel MRI coils [28] (Table 2). Furthermore, some studies failed to mention which type of MRI surface coil was used [23, 26, 29, 30]. 2D images were obtained by five studies [21, 23, 28–30], and 3D images by six studies [22, 24–27, 30]. The study by Tai et al. evaluated both 2D and 3D images from MRI [30]. MRI acquisition time ranged from 3 s [21] to 7 min and 1 s [22, 24, 27]. Five studies did not describe patient positioning during acquisition of MRI images [21, 23, 25, 26, 28], whereas the remaining five studies stated that MRI was acquired in centric occlusion [22, 24, 27, 29, 30], while the study by Tai et al. [30] described that their patients were seated on a chair while the MRI images were taken. Table 2 describes the echo time, repetition time, bandwidth, number of averages, echo train length, Field of view (FOV), number of sections, MRI voxel size and slice thickness, and patient positioning during imaging.

Characteristics of comparison modalities (CBCT and lateral cephalogram units)

Three of the included studies comprised detailed descriptions of the CBCT units used [22, 25, 26]. Considerable variability was recorded between these studies with respect to CBCT scanner settings and image acquisition parameters, CBCT unit, tube voltage, scanning time, CBCT FOV and CBCT voxel size (Supplementary File 4). The imaging softwares used were also dissimilar among the studies. Three of the included studies provided details of the cephalometric units in their study [21, 28, 29]. The method of derivation of lateral cephalograms using MRI was different in all of them.

Four studies performed the segmentation of craniofacial structures using 3D MRI images [22, 24, 27, 30]. Three of these studies were by the same author and employed the same segmentation software (Amira software, Version 6.4.0, Thermo Fisher Scientific, MA, USA) [22, 24, 27] (Supplementary File 5) for segmentation of craniofacial structures. Various types of statistical analyses were conducted in the included studies to assess the data and determine the utility of MRI. The analyses used are detailed in Supplementary File 5.

Comparison of various linear, angular, and soft tissue measurements

Supplementary File 6 presents a matrix of the various outcomes that were objectively compared among the included studies. Four studies calculated the MD among 34 linear, 33 angular, and 7 soft tissue variables [22, 26, 28, 30]. Among the linear measurements (measurements in mm), Maspero et al. reported the lowest MD of −0.08 for the distance between the right Gonion point (Go R) to Sella (S) and the highest for the distance between Go R to right condylion point (Co R) (0.53). Five of the measurements showed negative MD, whereas others were positive [26]. The MD was lowest for the distance between posterior nasal spine to Anterior nasal spine (ANS) (0) and highest for the distance between left condylion point (Co L) to point A (A) (−0.93) in the study by Juerchott et al. (a) [22], whereas Heil et al. reported the lowest values for overjet and Distance between point Pogonion (Pg) to Nasion-point B(NB) line (–0.05) and the highest values for overbite (0.61) [28]. Among the angular values in Maspero et al., the lowest MD was observed for the angle between Basion point (Ba), Sella, and Nasion (N) (0.03°) and the highest for ANB (0.47°) [26]. This was reported by Juerchott et al. (a) [22] as Frankfort’s Horizontal (FH) plane.N-Pg (−0.07°) and FH plane Sella Gnathion (S-Gn) angle (0.88°), and Heil et al. [28]. as 0.09° for upper incisor to Nasion to Point A (U1.N-A) and −1.33° for interincisal angle (). Most of the angular MD values were positive in the study by Maspero et al., whereas about half of the total reported values in the other two studies were negative [26]. Soft tissue variables in three planes were only reported by Tai et al. with the lowest MD being recorded in soft tissue-chin thickness (0.18) and the highest in distance between the soft tissue glabella and subnasale (0.81) [30].

Reliability of MRI as a diagnostic tool in orthodontics

Five of the included studies provided details of MRI reliability as compared to CBCT or lateral cephalograms. These data are presented in Table 3 [21, 22, 24, 26, 28]. Four of these studies assessed intra-class correlation for evaluating the intra- and inter-rater reliability [22, 24, 26, 28]. All four studies found a high level of agreement between MRI and the conventional standard modality used (i.e. CBCT or lateral cephalogram). Further details on ICC and intermodal agreement are provided in Table 4.

The highest inter-rater reliability was found in the study by Heil et al. [28] with excellent agreement amounting to an average (±standard deviation [SD], range) of 0.98 ± 0.014 (95% confidence interval [CI] −0.938; 0.997). Similarly, a high CBCT intra-rater agreement for rater I and rater II were found at 0.977 ± 0.019 (95% CI −0.926, 0.996) and 0.975 ± 0.017 (95% CI −0.937; 0.992), respectively. This result is similar to that of the MRI group, which recorded an inter-rater agreement of 0.929 ± 0.106 (0.467–0.996) and an intra-rater agreement of 0.975 ± 0.016 (95% CI −0.935; 0.997) for rater I and 0.961 ± 0.065 (95% CI −0.692; 0.998) for rater II. ICC values also showed high agreement in all four studies. Maspero et al. found equal precision for the coefficient of variation except for the distance between the ANS to the Menton (M) and the distance between the Go R and the Co R, which were found to be significantly less precise in MRI than CBCT [26]. However, the numerical values for reliability were comparatively inferior to CBCT, and high intra- and inter-rater reliability were found in the MRI group with a mean of 0.881 ± 0.071 (95% CI −0.849; 0.91) for rater I, 0.912 ± 0.064 (95% CI −0.884; 0.939) for rater II, and 0.833 ± 0.08 (95% CI −0.798; 0.868) for inter-rater reliability. The study by Juerchott et al. (a) compared linear measurements, angular measurements, and mean Euclidean distances for MRI versus CBCT [22]. The study found an excellent agreement and high inter- and intra-rater reliability as shown in Table 3.

The study by Abkai et al. assessed the mean relative distance for linear measurements and reported 2.4–2.7 for MRI and 1.6 for LCR [21]. Similarly, for angular measurements, an average deviation of 1.2° ± 1.1° was found as compared with LCR. The study by Juerchott et al. (b) assessed the inter- and intra-rater reliability of MRI, where the mean Euclidean distance (mm) for various skeletal and dental landmarks was evaluated in three planes, i.e., x, y, and z. The authors reported an intra-rater reliability for MRI of 0.87 mm (0.41–1.63) for rater I, 0.94 mm (0.49–1.28) for rater II, and 1.10 mm (0.52–2.29) for inter-rater reliability [24].

Meta-analysis

Due to variability in the assessment methods and outcome variables, the meta-analysis was attempted only for the data derived from the studies with homogenous protocols. Supplementary File 7 presents the pooled effect size, 95% CI, weight, p values, and the heterogeneity of the MDs in the linear and angular measurements between MRI and CBCT/lateral cephalogram that were reported for two or more studies. Among the linear measurements, the lowest pooled MD was 0.018 mm (95% CI −1.2 to 1.237) for Gonion Left to Menton (Go L-Me). The highest pooled MD was recorded for Nasion to Menton (N-Me) 0.428 mm (95% CI −0.620 to 1.475). Among the angles, the lowest pooled MD was seen for SN-PNS-ANS −0.057° (95% CI −1.062 to 0.947), whereas the highest pooled MD was reported for U1.N-A −0.684° (95%CI −1.763 to 0.394). The I² values for all these meta-analyses was 0.0% and all the MD were statistically non-significant (p-value greater than 0.05).

The forest plot (Fig. 3A) depicts the pooled intermodal bias for angular measurements in MRI. It was found to be 0.11°C (95% CI −0.32; 0.53), whereas it was 0.14 mm (95% CI −0.23; 0.51) for linear measurements (Fig. 3B). Heterogeneity was absent in these analyses and the p values were > 0.05 (Fig. 3A and B). The pooled ICC value for MRI observations was found to be 0.91 (95% CI 0.776; 1.05, I² = 97.5%, P = 0.000) and that for CBCT was found to be 0.96 (95% CI 0.94; 0.97, I² = 0.0%, P = 0.836) (Fig. 3C and D) The funnel plots could not be created as the number of studies in each meta-analysis was below 10.

Strength of evidence (GRADE)

Supplementary file 8 presents the assessment for the strength of evidence (GRADE) generated for outcomes of the included studies. A very low grade of evidence was observed for inter-rater and intra-rater agreement of MRI for landmark identification. Similarly, a very low grade of evidence was seen for agreement between MRI with CBCT, agreement between MRI and lateral cephalogram for landmark identification, and intermodal agreement for lines and angles.

Discussion

In orthodontics, a handful of reports have compared the reliability and accuracy of MRI with that of CBCT and lateral cephalometric evaluation [14–16]. This systematic review aimed to conduct a comprehensive analysis of these comparisons and identify any paucities that may be addressed in future research.

Within the limitations, this systematic review provides evidence to support that MRI has a high intra- and inter-rater reliability for landmark identification during diagnosis in orthodontics. Across the literature, data support that MRI is an acceptable diagnostic tool that is comparable to the current diagnostic standards in orthodontics, for example, lateral cephalograms and CBCT. The main findings of the present systematic review were: (1) the intra-rater reliability of MRI is comparable to those of CBCT and lateral cephalograms, (2) the inter-rater reliability of MRI is comparable to those of CBCT and lateral cephalograms, and (3) MRI may be a valuable diagnostic tool in orthodontics. However, future research is needed before MRI may be considered an evidence-based alternative to the currently established modalities [31].

The variability with respect to assessed landmarks, angles, and linear measurements was evident from the 74 comparisons, which were identified in the four studies that had calculated MD [22, 26, 28, 30]. Most linear and angular measurements showed MDs of less than 0.5 mm and 0.5 °C, respectively (Supplementary File 6). This was reflected in a high inter-rater reliability with excellent agreement between MRI and the gold standard methods. Due to the high variability, meta-analytical evaluation of data is seldom recommended. The same was encountered in the present review, but we used the random-effects model of meta-analysis to reconfirm the interpretations derived from the descriptive analysis.

The included studies displayed considerable variability with respect to study methods, including sample size, type of MRI used, compared factors, landmarks and parameters studied, rater qualifications for reliability testing, statistical methods, and other characteristics pertaining to MRI, CBCT, and lateral cephalogram imaging. This variability was one of the main reasons for the high or moderate risk of bias assessed among 70% of the studies and may be attributed to the novelty of the field and the preliminary nature of these research studies. Assessment of the strength of evidence using GRADE revealed that the strength of evidence for various outcomes of the systematic review was very low (Supplementary File 8). The main reasons for this were inadequacy of sample population, unclear sample characteristics, lack of homogeneity of various methodological aspects, for example, use of a 3-tesla or 1.5-tesla magnet in the studies, different exposure parameters, lacking rater blinding and, as mentioned above, the high risk of bias in some studies.

Current limitations of MRI

The major concerns for the use of MRI are its high cost of equipment, large size of the equipment, large installation area, and lack of availability of trained radiotechnologists operating the MRI machine for orthodontic purpose. The identification of cephalometric points using MRI images requires further training of the operating team. The availability of MRI machines across orthodontic clinics and education centers for routine cephalometrics could be challenging due to the abovementioned reasons. Furthermore, it has been associated with claustrophobia, requirements for operation and interpretation, long scan times, and image artifacts due to orthodontic metal brackets [5, 12]. Maspero et al. suggested that an open gantry method may eliminate claustrophobic experiences among patients [26]. The included studies used 3-tesla or 1.5-tesla MRI systems. Reducing the power of the magnet limits the clarity of images, but also reduces the exposure time and size of the machine. With advancements in the image segmentation and enhancement software, it may be possible to strike a balance between making the equipment suitable for most dental operators and generating images with a clarity that is optimized for orthodontic diagnosis. Abkai et al. introduced a novel ultra-short echo modality, providing one “shot“ MRI projections and thereby reducing exposure time [21]. Similar attributes have been displayed by a recent modification of MRI into black bone MRI [29]. The study by Maspero et al. also suggested that imaging with MRI for orthodontic patients using aligners or ceramic brackets may be done to completely eliminate concerns regarding artifacts [26]. MRI is not a new modality in the medical field and various methods have been invented and are being tested in orthopedics to eliminate the concerns regarding the use of metal structures in MRI. These methods may possibly be adapted and modified for dental use [9–11, 32, 33, 34, 35]. Recently, Sennimalai et al. conducted a review listing the studies assessing the use of MRI in orthodontics [36]. However, it was a scoping review and hence lacked systematic inclusion of studies, quantitative evaluation of study outcomes, and assessment of strength of evidence. The present systematic review followed the recommended methods of evidence-based medicine with strategies for quality assurance.

Future perspectives

Various controversial areas remain in contemporary orthodontics. These are primarily related to an incomplete understanding of the biomechanics and effects of different protocols. The option to apply radiation-free imaging may potentially increase the number of times that dental and skeletal structures may be visualized during and after treatment with the added benefit of exploring soft tissue landmarks. Though CBCT is an excellent diagnostic tool for orofacial structures and airways, the high radiation dosage limits its frequent use, particularly in adolescents undergoing orthodontic treatments. One recent advance in orthodontics is the segmentation of 3D structures using CBCT and 3D printing using CBCT. Four of the included studies performed segmentation using 3D MRI images and found that the accuracy of 3D printing was comparable to that of CBCT as established by Juerchott et al. in their studies [22, 24, 27]. Recently, a study attempted the diagnosis of canine impaction by segmentation using 3D MRI images [14]. This was a feasibility study without any kind of reliability testing, which was, therefore, not included in the qualitative and quantitative analysis of this review. Even so, segmentation using 3D MRI images may potentially open a new path towards reducing radiation exposure to our patients during diagnosis of canine impactions and may potentially bring an end to the current dispute in orthodontics about whether patients should be exposed to radiation (CBCT) for diagnosis of impacted canines or not [4].

The main limitations of this review are several aspects of methodological variability, variations in the positioning of the patients during MRI imaging, and small sample size. However, the results of this study provide an evidence-based direction for the journey towards radiation-free orthodontic diagnosis. Furthermore, the high cost of apparatus and lack of global availability of MRI for dentistry and orthodontic purposes currently pose challenges to its routine application.

Conclusion

This systematic review found that, within limitations, the literature exhibited a comparable intra- and inter-rater reliability and accuracy of MRI and standard imaging modalities (3D CBCT, and LCR images) in orthodontics. Current evidence supports the potential of radiation-free diagnosis and treatment planning in orthodontics. However, more studies with larger sample sizes and strong methodological considerations as highlighted in this article may facilitate qualitative analysis and add stronger evidence related to the reliability of this innovative technique.

Acknowledgement

We would like to acknowledge the contribution of Dr Nitesh Tewari, Additional Professor, Pediatric Dentistry, and Dr Ashish Dutt Upadhyaya, Scientist, Clinical Research Unit, All India Institute of Medical Sciences, New Delhi, India for their help with meta-analysis in the project.

Conflict of interest

None declared.

Funding

None declared.

Presentation at a meeting

International Association of Dental Research (IADR) 2023, Bogota, Colombia. Awarded as Winner of “IADR Innovation Award for Excellence in Orthodontics Research” 2023.

Ethical approval statement

Formal ethics approval was not required for this paper.

Data availability

The data underlying this article are available in the article and in its online supplementary material.

References