Introduction

Diagnostic imaging plays a pivotal role in dentistry, particularly in diagnosis and treatment planning. Among various radiographic modalities, multislice computed tomography (MSCT) and cone beam computed tomography (CBCT) are widely utilized. CBCT has gained popularity due to its lower radiation dose, reduced cost, and compact design, making it more accessible for routine clinical use [1-3]. Conversely, MSCT remains the gold standard for quantitative bone density measu­rement because of its standardized and reproducible Hounsfield units (HUs) [4, 5].

Unlike MSCT, which provides calibrated HU values reflecting tissue density, CBCT generates gray values (GVs) that represent grayscale intensities, but lack standardization across devices and acquisition protocols. This variability poses challenges for the direct comparison and clinical interpretation of CBCT data [1, 6]. Understanding these differences is crucial for evaluating the diagnostic accuracy and clinical utility of CBCT in bone quality assessment.

Bone mineral density (BMD) is a key determinant of dental implant success, influencing primary implant stability and osseointegration. Preoperative assessment of BMD assists clinicians in planning implant placement and selecting appropriate implant types [4]. While MSCT-based HU measurements are well-established in this context, the potential for CBCT-derived GVs to serve as reliable indicators of bone quality warrants thorough investigation [2, 7].

If a high correlation exists between CBCT GVs and CT HUs, CBCT data can reliably identify natural tis­sues and dental materials. A recent systematic review by Selvaraj et al. [7] highlighted a positive correlation between CBCT GVs and CT HUs. The current review aimed to provide robust scientific evidence regarding the diagnostic value of CBCT GVs in evaluating bone density and phantom materials. By evaluating more recent research beyond the scope of previous reviews, this study compared CBCT and CT to determine their correlation and the feasibility of using GVs for accurate diagnosis and treatment planning.

Material and methods

The research protocol for this systematic review was registered in the International Prospective Register of Systematic Reviews (PROSPERO) under the ID No. CRD42018085940.

The primary research question was: Do the GVs of CBCT correlate with the HUs of MSCT in assessing bone density?

Inclusion criteria were based on the PIRD framework and defined as follows: Population (P): Patients or specimens undergoing imaging for bone density assessment; Index test (I): GVs of CBCT; Reference test (R): HUs of MSCT; Diagnosis (D): Assessing bone density.

A comprehensive search was conducted across three major international databases, i.e., PubMed, Web of Science, and Scopus, to identify relevant studies published until September 30, 2024. References of considered studies were also manually screened to ensure completeness. For clarity and transparency, the full search strategy, with databases, key words, Boolean operators, and date ranges is summarized in Table 1. The process of the study selection is illustrated in the PRISMA flow diagram in Figure 1.

Eligible studies included in vivo, in vitro, and ex vivo experiments, which investigated correlations between GVs from CBCT and HUs from MSCT, with no restrictions in terms of year, language, or publication setting.

Excluded were review articles, case reports or series, letters to editor, editorials, opinions, book chapters, textbooks, conference abstracts, patents, studies utilizing bone density measurement methods other than CBCT and MSCT, evaluations of pathological bone conditions, and articles presenting qualitative findings only.

Identified studies were imported into EndNote 20 software for cataloging and duplicate removal. Two independent reviewers (T.E. and K.R.) screened the remaining studies, applying inclusion and exclusion criteria to evaluate inter-rater reliability (k = 0.81, indicating substantial agreement). Titles and abstracts were reviewed independently, with potentially eligible full-text articles further assessed. Discrepancies were resolved through discussion or by a third reviewer (A.Sh.).

To ensure consistency and reproducibility in data extraction, a standardized data extraction form was developed and pilot-tested prior to formal data collection. This form contained predefined fields covering study characteristics, imaging parameters, regions of interest, and outcome measures. Data extraction was independently performed by two reviewers, with disagreements resolved by consensus or third-party adjudication.

Data extracted from eligible studies included study identification (author, year, country), sample characteristics (study population, sample size, test modality, gold standard, region of interest), and results and conclusions (sensitivity, specificity, positive predictive value, AUC, mean difference, Pearson’s correlation coefficients, Spearman’s correlation test, beta coefficients).

The Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2) checklist was employed to assess the risk of bias of the included studies (Figure 2) [3, 8-10, 12-17, 25-29, 31, 32]. For this review, the QUADAS-2 checklist was modified. “Subject” was used instead of “patient,” as actual patients were not part of most studies. Studies rated as “high” or “unclear” in one or more domains were considered to have a high risk of bias or concerns regarding applicability. Conversely, studies rated “low” across all domains were deemed to have a low risk of bias or low applicability concerns. Risk of bias was independently evaluated by two reviewers (T.E. and K.R.), with disagreements resolved by a third author (A.Sh.).

Heterogeneity among studies was measured qualitatively by examining variations in study design, imaging modality (in vivo, ex vivo, or in vitro), anatomical regions examined, CBCT and MSCT protocols, and statistical approaches used for reporting correlations. These differences were considered too substantial to justify pooling the data quantitatively.

Findings from the included studies were extracted and categorized (Table 2). Due to significant heterogeneity among the included studies, a meta-analysis was not feasible.

Results

The electronic search identified 2,092 studies. A total of 1,036 duplicate articles and 1,021 irrelevant articles were excluded during the title and abstract screening phase, leaving 35 articles for eligibility assessment. Of these, 23 studies were excluded for not meeting inclusion criteria, resulting in 12 studies included in the qualitative analysis. Additionally, a manual search of reference lists identified 5 more studies, bringing the total to 17 studies for qualitative synthesis (Figure 1).

The majority of studies showed low risk of bias in the domains of index test, reference standard, and flow and timing. However, Naitoh et al. [8] reported a high risk of bias in the flow and timing domain due to variable evaluation intervals, which may compromise the reliability of the findings (Figure 2). This inconsistency in measurement timing can bias findings and affect the comparability of data, highlighting the importance of conducting evaluations at regular intervals to reduce bias and ensure accurate conclusions.

Out of the 17 included studies, 10 were conducted in vitro, 4 ex vivo, and 3 in vivo. The in vivo studies consistently demonstrated strong linear correlation between CBCT GVs and MSCT HUs [8-10]. Within the in vitro studies, most confirmed strong correlations using regression or Pearson’s analysis, though some reported moderate or weak relationships influenced by factors, such as region of interest (ROI), FOV size, or material composition.

Correlation strength was categorized consistently across studies as weak (r < 0.3), moderate (r = 0.3-0.7), and strong (r > 0.7) [11]. The summary of the study charac­teristics, imaging protocols, and reported correlation strengths is presented in Table 2 [3, 8-10, 12-17, 25-29, 31, 32].

Additional findings provided specific insights into these influencing factors. For instance, one study found that smaller FOVs (8 × 8 cm) resulted in lower GV errors compared to larger ones (12 × 15 cm) [12], while another observed significant variation in gray values across CBCT systems and materials (with exceptions noted in small-FOV systems) [13]. Bastami et al. [9] demonstrated a strong linear correlation between GVs derived from CBCT and HUs obtained from MSCT in rabbit calvarial defects grafted with various biomaterials. Similarly, Kamaruddin et al. [14] demonstrated significant Pearson’s correlations using an anthropomorphic phantom, despite insignificant differences in mean values. Other research confirmed strong correlations across homogeneous dental materials [15], and revealed inconsistencies in CBCT gray values due to repositioning errors [16]. Silva et al. [17] found that CBCT tended to overestimate gray intensity compared to MSCT in mandibular bone.

Discussion

This systematic review assessed the reliability of CBCT GVs in comparison to MSCT HUs for evaluating BMD. CBCT has gained increasing use in dental diagnostics due to its lower radiation exposure, reduced cost, and greater accessibility. Although MSCT remains the gold standard for quantitative bone density assessment, CBCT may serve as a practical alternative when imaging conditions are well-controlled.

Strong correlations between CBCT GVs and MSCT HUs have been consistently reported in in vivo, ex vivo, and in vitro studies, especially in the context of pre-implant planning and bone quality assessment. For example, Bastami et al. [9] showed a strong agreement in evaluating graft materials in rabbit skull defects, and Naitoh et al. [8] demonstrated the feasibility of CBCT for long-term bone quality monitoring in patients. These findings indicate that CBCT-derived grayscale values may be clinically applicable when used under optimized protocols.

Several technical and procedural factors influence the accuracy of CBCT grayscale values. FOV size, for instance, plays a critical role in minimizing scatter and image distortion. Studies have shown that smaller FOVs (e.g., 8 × 8 cm) result in more accurate GV readings, while larger FOVs introduce more noise and variability. Yadegari et al. [12] highlighted significant GV deviations associated with larger FOVs (12 × 15 cm), reinforcing the importance of tailoring FOV size to specific diagnostic requirements.

Software tools and ROI selection methods also affect measurement consistency. Manual ROI placement is subject to operator variability, while standardized and automated techniques improve reproducibility. For example, studies employing circular ROI placement in cancellous bone [8] or 3D registration algorithms [8], reported enhanced measurement reliability. This emphasizes the need for precise, replicable ROI protocols to ensure accurate density evaluations.

The anatomical site of measurement significantly affects GV and HU values. Studies have consistently shown substantial variation in jawbone density, which can impact primary implant stability. For example, a study analyzing 260 implant sites found highest mean HU values in the anterior mandible (862.8 ± 203.4 HUs), followed by the anterior maxilla (594.2 ± 95.2 HUs), posterior mandible (528.4 ± 115.6 HUs), and posterior maxilla (438.1 ± 110.2 HUs) [18, 19]. These findings emphasize the biomechanical advantages of the anterior mandible for implant placement. Moreover, differences in correlation strength between CBCT and MSCT values have been reported based on anatomical location, with some studies stratifying maxillary and mandibular sites and observing significant differences [10, 13, 15]. These variations likely stem from differences in trabecular structure and bone quality as well as regional susceptibility to imaging artifacts.

A notable methodological gap is the absence of micro-computed tomography (micro-CT) as a reference standard in the included studies. Micro-CT provides high-resolution, three-dimensional imaging, and allows accurate quantification of bone morphology, with parameters, such as bone volume, trabecular thickness, and structural indices [20]. Although limited to ex vivo analysis due to its invasive nature, micro-CT is essential for establishing accurate benchmarks for evaluating other modalities. While some recent studies report acceptable agreement between CBCT and micro-CT for certain morphometric parameters, particularly when high-resolution voxel settings are applied, the lack of direct micro-CT comparisons in this review limits the validation of CBCT for fine-structure analysis [20, 21].

Although linear correlations between grayscale values and bone density were commonly reported, the clinical utility of GV-based thresholds for predicting implant stability remains unclear. Primary implant stability depends on multiple factors, including implant geome­try, surgical technique, and bone density distribution. GV and HU values, therefore, should be interpreted as indirect indicators rather than definitive predictors. Some studies have reported significant correlations between CBCT-derived grayscale values and implant stability quotient (ISQ) [22], although inter-scanner variability and calibration issues remain concerns. Recent literature supports the idea that grayscale values when integrated with other clinical parameters, may aid in preoperative planning, but cannot independently predict implant outcomes [23, 24].

CBCT’s inherent technological limitations, including beam hardening, scatter, and noise can cause overestimation of GV values in dense structures, such as cortical bone and enamel [3, 25]. To address these discrepancies, several studies have proposed mathematical conversion models to align CBCT GVs with MSCT HUs. For example, Cassetta et al. [26] introduced a 0.7 conversion ratio, while Ostovarrad et al. [27] and Parsa et al. [28] proposed exponential and linear regression equations, respectively. These approaches aim to improve cross-modality consistency, but require further validation across devices and patient scenarios.

Substantial heterogeneity was observed among the included studies, such as differences in CBCT models, voxel sizes, FOVs, anatomical sites, and examiner expertise. For instance, Varshowsaz et al. [29] reported signi­ficant GV discrepancies in high-density regions, while Shokri et al. [13] emphasized the impact of imaging conditions on measurement accuracy. Although many studies demonstrated strong average correlation coefficients, this variability raises questions about the overall reliability of GVs as consistent diagnostic indicators. CBCT measurements remain highly sensitive to acquisition protocols, and without standardized calibration, high correlation values alone cannot justify the substitution of MSCT with CBCT for BMD analysis [25, 30].

This review is subject to several limitations. Variability in study methodology, lack of standardized imaging protocols, and differences in reporting of correlation indices, limited both data comparability and meta-analysis feasibility. Some studies lacked sufficient statistical details, which restricted robust synthesis. Future investigations should adopt uniform evaluation protocols, and include broader populations and anatomical sites to enhance generalizability.

Collectively, current evidence highlights the emerging potential of CBCT in bone density evaluation within dental and maxillofacial imaging. Nevertheless, the clinical applicability of grayscale values remains constrained by methodological heterogeneity, device-related variability, and the absence of universally accepted calibration standards. The lack of validation against high-resolution reference modalities, such as micro-CT, further limits the interpretability and standardization of findings. Future investigations should focus on implementing standardized imaging protocols, developing robust cross-platform calibration models, and incorporating gold standard comparisons, to enhance the diagnostic reliability and translational value of CBCT-based assessments.

Conclusions

Under controlled imaging conditions, CBCT demonstrates strong potential for bone density estimation, espe­cially in low-density regions, due to its significant correlation with MSCT HUs. However, MSCT remains the gold standard for quantitative assessment, particularly in high-density structures, where CBCT accuracy may be compromised by artifacts and scatter. Therefore, clinicians should carefully balance CBCT’s practical benefits, such as reduced radiation exposure and cost, against its current limitations when selecting an imaging modality for preoperative planning or bone quality evaluation.

Disclosures

Author contributions: Conceptualization: A.Sh.; Metho­dology: A.Sh., T.E.; Questionnaire adaptation: A.Sh., T.E.; Investigation and data collection: T.E., K.R. and F.M.; Formal analysis: T.E., A.I.D.; Data curation: T.E., F.M.; Writing of original draft: T.E., F.M.; Writing – review and editing: T.E., F.M., A.I.D., A.Sh.; Supervision: A.Sh.; Project administration: A.Sh. All authors have read and agreed to the published version of the manu­script.

Funding: This research received no external funding.

Institutional Review Board statement: Not applicable.

Informed consent statement: Not applicable. This systematic review was based exclusively on previously published literature and did not involve human participants, primary data collection, or direct patient contact.

Data availability statement: All the data generated or analyzed in this study are included in this manuscript. The data generated in this study may be requested from the corresponding author.

Acknowledgments: This paper was derived from a thesis with the reference number: 9904172384, which was conducted at the Research and Technology Center
of Hamadan University of Medical Sciences, Iran.

Conflicts of interest: The authors declare no conflicts
of interest.

AI use statement: No artificial intelligence tools were used in the preparation of this manuscript.

References