Spatial Discrepancies in CBCT-Based Localization of Impacted Mandibular Third Molars Authors: Zhizheng Li, Yanping Zou, Tiantian Wang, Xinjia Tang, Jiale Huang, Shuai Peng, Jihong Zhao Published: 2026 ABSTRACT AbstractIntroduction and aimsTo quantitatively evaluate the accuracy of cone-beam computed tomography (CBCT) in localizing impacted mandibular third molar (IMTM), using surgical guides as an image transfer modality.MethodsPatients requiring extraction of low-level horizontally impacted mandibular third molars underwent standardized CBCT. For 10 deeply impaction cases (Pell & Gregory Class C), personalized guides were designed and printed based on CBCT models to demarcate crown boundaries. Intraoperative mesiodistal and vertical discrepancies between guide-indicated and actual crown positions were measured. For an additional 60 cases without guides, only the horizontal discrepancy between the IMTM and the adjacent second molar was measured.ResultsActual IMTM positions extended mesially beyond the second molar distal surface (mean: 4.51 ± 1.52 mm), with significantly greater deviations in Class C than Class B (P < .05). Surgical guide-based measurements confirmed that actual crown positions were mesial and inferior to CBCT indications. Mean mesiodistal deviation was 4.94 ± 1.45 mm. Significant vertical discrepancies were also observed (mean superior: 5.21 ± 1.38 mm; mean inferior: 4.62 ± 0.95 mm).ConclusionIn the mandibular retromolar region, CBCT-based localization may deviate from actual anatomy by several millimeters in mesiodistal and vertical dimensions, with deeper impactions showing greater discrepancies. Heightened vigilance is warranted when planning surgeries based solely on CBCT for deeply impacted IMTMs or retromolar foreign bodies.Clinical RelevanceDuring mandibular retromolar region imaging, CBCT demonstrates mesiodistal and vertical deviations, with these error increasing as the target approaches the mandibular angle and inferior mandibular border.Key words: Cone-beam computed tomography, Impacted third molar, Surgical guide, Localization discrepancy Introduction Since its introduction from the field of angiography in the 1980s, cone-beam computed tomography (CBCT) has become one of the principal imaging modalities in oral and maxillofacial surgery.1, 2, 3, 4 Compared to conventional multidetector spiral CT, CBCT offers distinct advantages, including lower radiation doses, reduced costs, shorter acquisition time, and fewer metal artifacts, making it widely applicable for diagnosing dental and jaw diseases, preoperative evaluation, and surgical planning.5, 6, 7, 8, 9 Recently, surgeons across multiple centres have observed a degree of inconsistency between the spatial coordinates suggested by CBCT-reconstructed images and the actual anatomy exposed intraoperatively.10,11 This is particularly evident in cases of low-level horizontally impacted mandibular third molars (IMTMs) or foreign bodies located deep within the retromolar region or near the inferior border of the mandible. Such spatial discrepancy can compromise preoperative judgments regarding surgical difficulty, risk, and approach, thereby increasing surgical risks, reducing operational precision, and potentially leading to complications. While previous studies have evaluated the geometric accuracy and measurement precision of CBCT, most concluded that global linear and angular errors remain within clinically acceptable limits.12, 13, 14, 15, 16, 17, 18, 19 However, these studies predominantly utilized internal or anatomical landmarks as references, and measurements were typically conducted near the geometric centre of the cone beam. For the mandibular retromolar region—an area often distant from the imaging centre, anatomically unique, and highly critical for surgical decision-making—has not been sufficiently evaluated with respect to reconstruction accuracy. To address this, this study focused on IMTMs whose crowns appeared, on CBCT, to be immediately adjacent to the second molar (M2). The mesial displacement of the IMTM crown relative to the distal surface of the M2 was measured intraoperatively to assess CBCT's accuracy in representing retromolar structures. Furthermore, CBCT-reconstructed images were transferred to 3D-printed surgical guides, enabling precise, quantitative comparison between radiographic data and actual intraoperative anatomy. This approach allowed a quantitative assessment of the 3-dimensional localization accuracy of CBCT in the mandibular retromolar region, with the aim of providing evidence to support clinical decision-making, preoperative evaluation, and risk prediction. Methods Patients A total of 70 patients diagnosed with low-level, horizontally IMTMs requiring surgical extraction were included. The cohort consisted of 60 cases who underwent intraoperative measurements only, and 10 cases underwent surgical guided procedures. Patients were consecutively treated at the Hospital of Stomatology, Wuhan University, between August 2025 and March 2026. The inclusion criteria were as follows: (1) CBCT demonstrated that the IMTM was in horizontal position and the crown was located below the bone surface (Pell-Gregory classification Class B or Class C); (2) the IMTM is indicated for surgical removal. Cases would be excluded if (1) adjacent teeth containing metal or other materials can produce image artifacts; (2) the patient was pregnant; (3) pre-existing limited mouth opening (upper and lower middle incisor distance ≤ 35 mm); (4) the patient refused a CBCT examination; (5) other factors that lead to patients’ refusal of treatment or patient non-compliance. This study complied with the Declaration of Helsinki and was approved by the Regional Ethical Review Board of the Ethics Committee of the Hospital of Stomatology, Wuhan University (WKDQ2025B99). This study was registered under Chinese Clinical Trials Registry on 20/8/2025, with the registration number: ChiCTR2500115417. All patients provided written informed consent prior to participation. CBCT examination All CBCT scans were performed using the same device (Newtom VGI, Quantitative Radiology) with standardized parameters: 110 kV, 2.23 mA, 0.3 mm voxel size, 15 cm (diameter) × 12 cm (height) field of view, 3.6 s acquisition time, and approximately 6.07 mAs exposure. Patients were positioned in an upright seated posture. Standardized positioning was achieved using the machine’s alignment system and laser guides: the Frankfort horizontal plane was aligned with the horizontal laser beam, and the midsagittal plane with the vertical laser beam. A 360° rotation scan was performed to acquire images. Images were exported in DICOM 3.0 format. Surgical guide design Virtual models and surgical guides were designed using Mimics Research 21 and Magics 21 (Materialise, Leuven, Belgium). During the conversion of DICOM data into 3-dimensional models, all cases were processed by the same experienced investigator using a consistent standard threshold, strictly following the standard medical segmentation workflow recommended by Mimics software to maximize resolution consistency. The surgical guides were fabricated using a Desktop Photocurable 3D Printer (Shape 1+ Dental, RAYSHAPE) with standard resin (SGV2, RAYSHAPE). The printing layer thickness was 50 μm, and the overall fabrication accuracy was within ±75 μm. Strictly following the manufacturer’s curing guidelines, the printed guides were post-cured in water under ultraviolet light at 30°C for 3 minutes and disinfected preoperatively with 0.5% povidone-iodine solution. The square window of the guide was designed based on the projections of tangents to the most anterior, superior, and inferior points of the IMTM crown, viewed perpendicular to the long axis of the mandible. The guide seating and retention components extended mesially to include the crowns of 4 adjacent teeth (Figure 1). A mixed tooth-bone support design was adopted, with the 4 adjacent teeth providing primary seating and stability, the exposed mandibular bone surface offering additional support. For each case, the mandibular model and the impacted third molar model were also printed separately and test-fitted with the surgical guide in vitro to verify alignment with the digital design before use in surgery. Furthermore, prior to surgery, each guide was seated intraorally, only guides that achieved stable and unique seating were used for the surgical procedure and study measurements. Representative case of surgical guide design. (A) CBCT image of an impacted tooth #38. (B) Surgical guide design: the cyan model represents the completed surgical guide, the red model represents tooth #38, and the yellow model represents tooth #37. (C) Printed surgical guide. Surgical procedure and discrepancy measurement All procedures were performed by the same experienced surgeon team using a standardized protocol. A conventional retromolar angular incision was made, followed by routine flap elevation. After soft tissue flap elevation, the surgical guide was seated. A bone window was created according to the guide design to expose the IMTM. The bone window was further enlarged as needed to accommodate the surgical procedure. After complete exposure of the maximum crown dimension, the guide was reseated to measure the distance discrepancies between the guide-indicated position and the actual crown position using a surgical probe. Lfront was defined as the absolute distance between the most anterior edge of the actual crown and the most anterior edge of the guide window; Lup was defined as the absolute distance between the most superior edge of the actual crown and the most superior edge of the guide window; and Lbottom was defined as the absolute distance between the most inferior edge of the actual crown and the most inferior edge of the guide window. The IMTM was then extracted following routine procedures to complete the surgery (Figure 2). Surgical procedures for IMTM extraction under guide assistance. (A) Flap elevation to expose the bone surface. (B) Seating of the surgical guide. (C) Bone window creation according to the guide design. (D) Exposure of the IMTM without visualization of the most anterior crown margin. (E) Mesial extension of the bone window to expose the most anterior crown margin. (F) Tooth sectioning. (G-I) Sequential removal of the root, crown, and dental follicle. (J) Reseating of the surgical guide showing the actual IMTM position mesial to the location indicated by CBCT. (K) Wound closure. (L) Extracted IMTM and dental follicle. Intraoperative deviation measurements without guide assistance In the 60 cases without surgical guides, routine surgical procedures were performed. After exposing the most anterior point of the IMTM crown, the vertical distance (Dmesial) between the actual anterior border of the IMTM crown and the distal surface plane of the second molar was measured using a probe. Statistical analysis All data are presented as the mean ± standard deviation. The data analysis was performed using IBM SPSS Statistics version 27.0 (IBM Corp.). The statistical data were analysed using the dependent t-test, one-way ANOVA for comparisons among 3 groups. P < .05 was considered statistically significant. Results All surgeries were successfully completed without any reported complications. Although the surgical guides consistently exhibited deviations relative to the actual positions of the impacted teeth, they effectively served as auxiliary tools for localization of fully impacted third molars. Mesiodistal deviation without guide assistance In all 60 cases where CBCT suggested that the IMTM crown was adjacent to the distal of M2, intraoperative findings revealed that the actual crown intraoperative findings revealed that the crown extended mesially beyond the M2 distal plane to varying degrees, appearing on the buccal side of M2. Dmesial ranged from 1.43 to 8.60 mm (mean: 4.51 ± 1.52 mm). When classified according to the Pell & Gregory classification based on impaction depth (Table 1), Dmesial ranged from 1.43 to 6.45 mm (mean 3.43 ± 0.99 mm, n = 21) in Class B group, and from 2.81 to 8.60 mm (mean 5.08 ± 1.44 mm, n = 39) in Class C group. In the vertical dimension, greater vertical impaction depth (Class C) was associated with significantly larger deviations (P < .0001, Figure 3A). Horizontally, Dmesial also correlated with spatial classification, with Class III group showing significantly greater deviations than Class I and II (P = .0004, Figure 3B). Table 1Distance from the most mesial edge of the IMTM crown to the distal plane of M2.Table 1 dummy alt textSample sizeDmesial (mm)P valueVertical depth classB213.43 ± 0.99<.0001C395.08 ± 1.44Horizontal space classI233.90 ± 0.80.0004II204.24 ± 1.49III175.64 ± 1.67All cases604.51 ± 1.52 Distance from the most mesial edge of the IMTM crown to the distal plane of M2. Comparison between subgroups based on the Pell & Gregory classification. (A) Class C IMTMs exhibited significantly greater CBCT image deviation compared with Class B IMTMs. (B) Class III group showing significantly greater deviations than Class I and II. Guide-based objective mesiodistal deviation Based on above findings, we enrolled 10 Class C IMTM cases. Surgical guides were used as carriers to transfer CBCT images into the oral cavity for direct, objective comparison and measurement against the actual anatomical positions. In all 10 cases, the intraoperative findings revealed that the bone window created according to the guide design failed to expose the most anterior edge of the crown, indicating that the actual IMTM crown was positioned mesial to the location indicated by the guide. Lfront values ranged from 2.78 to 7.65 mm, with a mean of 4.94 ± 1.45 mm. Guide-based objective vertical deviation The vertical discrepancy was evaluated using 2 parameters Lup and Lbottom. Lup values were unobtainable in some cases due to the loss of reference points after bone window extension. Additionally, based on surgical considerations, complete exposure of the most inferior aspect of the IMTM crown was not required for all patients, and thus Lbottom was not measured in some cases. Intraoperative observations indicated that, in most patients, the bone window created according to the guide design was located inferior to the actual IMTM position (Table 2). Lup values ranged from 2.68 to 6.39 mm, with a mean of 5.21 ± 1.38 mm (n = 5). Lbottom values ranged from 4.03 to 5.70 mm, with a mean of 4.62 ± 0.95 mm (n = 4). Table 2Discrepancies between the actual IMTM position and the surgical guide-indicated position.Table 2 dummy alt textLfront (mm)Lup (mm)Lbottom (mm)Case no.n = 10n = 5n = 414.96\\25.95\\34.104.864.0342.785.835.7055.692.68\66.40\\77.656.305.3583.306.393.3893.61\\104.93\\Mean ± SD4.94 ± 1.455.21 ± 1.384.62 ± 0.95 Discrepancies between the actual IMTM position and the surgical guide-indicated position. Discussion Theoretically, the relative positions of adjacent structures within the same scan (such as the second and third molars) should remain constant. However, our findings demonstrate that the actual position of the IMTM was consistently mesial and slightly superior to that indicated by CBCT, with the deviation increasing for Class C or deeper IMTMs. Clinically, IMTMs that appear distal to the M2 on CBCT-based localization are often found on the buccal side of the M2 during surgery. The discrepancies between Lup and Lbottom preliminarily suggest potential image discrepancies in the vertical morphology of CBCT. However, the small sample size precludes both definitive conclusions and precise quantification. Such discrepancies have often been overlooked or attributed to differences in patient positioning during CBCT acquisition and the surgeon’s intraoperative viewing perspective (typically from the right anterior aspect).10,20, 21, 22, 23 In this study, we employed surgical guides as a transfer medium for CBCT images and standardized both the viewing perspective and the surgical workflow to minimize bias arising from perspective differences. Several mechanisms may contribute to the discrepancies observed in this study. First, the position of the target region within the scan field is considered the most critical factor. The mandibular retromolar region, particularly in cases of low-level horizontally impacted IMTMs, is typically located far from the geometric centre of the CBCT cone-beam. Regions near the periphery of the scan field are more susceptible to image stretching and distortion.12,17,24, 25, 26 Such discrepancies between CBCT localization and intraoperative anatomy is often overlooked in conventional measurement studies centred on anatomical landmarks.27, 28, 29, 30, 31 Second, the partial volume effect is an inherent physical limitation of CT imaging.32, 33, 34, 35, 36 When adjacent tissues of different densities occupy the same voxel, the delineation of boundaries becomes blurred, leading to “expansion” or “shrinkage” of 3-dimensional reconstructed models, which may influence the observed discrepancies. Third, the role of image reconstruction algorithms represent a potential contributing factor. 25,37,38 Whether the observed discrepancies are algorithm-related remains uncertain and would require further investigation by researchers specializing in this area. Nevertheless, it is evident that studies on CBCT image reconstruction algorithms have not specifically addressed this issue. We hope that the present study will draw attention to this matter among relevant experts. From a technical perspective, adjusting the scan centre or optimizing reconstruction algorithms may partially improve geometric accuracy.39, 40, 41, 42 However, given the substantial variability among devices and patients, these approaches are unlikely to fully resolve the issue. For complex cases requiring high localization precision, spiral CT may offer superior accuracy based on accumulated research reports.15,16,43 However, the present study did not directly compare spiral CT and CBCT in the retromolar region. This study has limitations. First, the sample size of guide group was relatively small, and no control group (eg, Class A or B IMTMs) was included. Second, the inherent accuracy and reproducibility of the surgical guide were not independently validated, and the guide design perspective (perpendicular to the long axis of the impacted tooth and the local bone surface) may not have fully aligned with the actual surgical perspective in some cases. Additionally, although potential measurement errors arising from guide positioning inaccuracies, segmentation threshold variability, and manual probing errors cannot be completely excluded, the magnitude of the observed deviations far exceeds any plausible magnitude of these errors, and they would not theoretically affect the main conclusions of this study. Future studies could employ artificial jaw models to systematically simulate various impaction patterns and scanning parameters, enabling more controlled separation and quantification of different error sources. Finally, this study was conducted using a single CBCT device and scanning protocol. The pattern of discrepancies may vary across different devices, scanning parameters, reconstruction algorithms, and patient positioning.12 This study provides quantitative evidence of systematic spatial discrepancies between CBCT-based localization and intraoperative findings of IMTMs in the mandibular retromolar region. When planning surgical procedures based on CBCT images, particularly for extraction of low-level horizontally impacted mandibular third molars or exploration of foreign bodies in the retromolar region, surgeons should be aware of the potential discrepancy between the imaged location and the actual anatomical position. This discrepancy, which may reach several millimetres in the mesiodistal dimension, and tends to increase with proximity of the IMTM to the mandible angle and inferior border. Clinicians should exercise caution when relying on the “absolute position” provided by CBCT and may consider designing bone windows with a mesial and apical safety margin. This study aims to highlight a critical clinical issue, rather than to dismiss the overall value of CBCT. Author contributions Jihong Zhao and Zhizheng Li conceived and designed the study. Zhizheng Li and Yanping Zou designed the 3D-printed surgical guides. Jihong Zhao, Tiantian Wang, Xinjia Tang, Jiale Huang and Shuai Peng performed the surgeries. Zhizheng Li and Yanping Zou analysed and interpreted the intraoperative measurement data, drafted the original manuscript. Jihong Zhao critically revised the manuscript. All authors gave their final approval and agree to be accountable for all aspects of the work. Data availability All data generated or analysed during this study are included in this published article and its supplementary information files. Conflict of interest None disclosed. Acknowledgements We thank Dr. Xiang Li from the Oral and Maxillofacial Radiology Department, School and Hospital of Stomatology, Wuhan University, for his assistance with CBCT scan protocols and technical support. No funding was received for this study. Footnotes Supplementary material associated with this article can be found in the online version at doi:10.1016/j.identj.2026.109694. Supplementary materials REFERENCES Associated Data Supplementary Materials Data Availability Statement All data generated or analysed during this study are included in this published article and its supplementary information files.