From Imaging to Instrument: A CT-Based Classification of the Nasofrontal Beak and Frontal Ostium Clearance with Prospective Surgical Correlation

Abstract

Background/Objectives: Frontal sinus surgery remains one of the most challenging procedures in endoscopic sinus surgery due to the complex and variable anatomy of the frontal recess. While several radiological parameters have been proposed to guide surgical planning, there is no standardized anatomical classification that informs both approachability and techniques. This study aimed to develop a reproducible CT-based classification of the nasofrontal beak (NFB) and frontal ostium clearance (FOC), and to validate its surgical applicability through a two-phase design. Methods: A combined retrospective–prospective cohort study was conducted. In Phase 1, CT scans from 1383 patients (2595 sinuses) were analyzed to measure NFB and FOC lengths via a novel technique. The cutoff points were defined through histograms, kernel density estimation, and K-means clustering. In Phase 2, 100 patients (191 sinuses) who underwent frontal sinus surgery were prospectively assessed. Radiological classifications were compared with intraoperative findings, and instruments were selected on the basis of anatomical classes. Agreement between radiological and surgical classifications was analyzed, and surgical outcomes were recorded. Results: NFB and FOC were each classified into three groups: Class A (≤6 mm), Class B (6–12 mm), and Class C (≥12 mm) for NFB, and Class A (≥12 mm), Class B (6–12 mm), and Class C (≤6 mm) for FOC. There was excellent agreement between radiological and intraoperative NFB classes. Class-specific techniques included no modifications for Class A NFB, frontal sinus punches for Class B, and straight/curved drill for Class C. Only one intraoperative lamina papyracea injury and two cases of postoperative neo-ostium narrowing were reported. Significant sex-based anatomical differences and minor side-to-side asymmetries were also observed. Conclusions: This novel classification provides a reproducible, internally validated system for stratifying frontal sinus anatomy preoperatively, with good radiological–surgical concordance for the nasofrontal beak. By mapping the anatomical class onto instrument selection, it offers a practical planning adjunct. As a single-center derivation without an external cohort or comparator arm, it demonstrates anatomical–surgical concordance and feasibility rather than clinical superiority, and warrants multicenter external validation.

1. Introduction

Frontal sinus surgery represents one of the most complex procedures in otorhinolaryngology, owing to the proximity of the frontal recess to the anterior skull base, the orbit, and the anterior ethmoidal artery, and to intricate anatomical variations that differ not only between patients but also between sides [1,2,3]. The frontal sinus drains through the narrow, hourglass-shaped frontal recess, the boundaries of which are defined by a variable constellation of frontoethmoidal cells, the agger nasi cell, the bulla ethmoidalis, and the nasofrontal beak. A broad spectrum of diseases may involve this region, including chronic rhinosinusitis with and without nasal polyposis, mucoceles, osteomas and other benign neoplasms, inverted papilloma, sinonasal malignancy, and trauma or cerebrospinal fluid leak. Surgical management is correspondingly graded: from simple endoscopic frontal sinusotomy (Draf I–IIa) for limited disease, through extended unilateral clearance (Draf IIb), to the endoscopic modified Lothrop/Draf III procedure for advanced, recurrent, or neoplastic disease. Selecting and safely executing the appropriate procedure depends critically on a precise preoperative understanding of frontal recess and ostial anatomy, since inadequate clearance predisposes to disease persistence and neo-ostium restenosis, whereas overly aggressive dissection risks injury to the skull base, orbit, and anterior ethmoidal artery. Identification of the frontal recess and ostium is therefore essential, with surgical extensiveness determined by pathology and anatomical variants [3,4]. Multiple CT measurements aid preoperative evaluation, including frontal sinus size and shape, intersinus septum orientation, and the morphology of the nasofrontal beak (NFB) [5,6]. The NFB is the midline thickening of the nasal process of the frontal bone, presenting as the anterior border of the frontal sinus ostium [7,8]. Several radiological parameters and classifications have been proposed to characterize this region, including the International Frontal Sinus Anatomy Classification (IFAC), which describes the presence and identity of frontoethmoidal cells, and radiological grading schemes of the frontal ostium. However, existing systems are oriented toward identifying anatomical structures rather than quantifying the bony working corridor available to the surgeon, and few translate directly into intraoperative instrument selection. A reproducible, measurement-based parameter that links preoperative imaging to operative decision-making remains lacking. Owing to its surgical importance and varying anatomy, this study introduces a CT-based classification of the nasofrontal beak based on its anteroposterior diameter and the frontal ostium clearance (FOC), and prospectively examines its anatomical–surgical concordance in a two-phase design, presenting class-specific techniques and approaches aimed at improving frontal sinus ostium identification. The term “classification” here refers to a measurement-based operative stratification of frontal sinus anatomy according to surgically relevant CT-derived parameters, rather than a new formal anatomical taxonomy.

2. Materials and Methods

2.1. Study Design and Participants

This combined retrospective–prospective cohort study consisted of two phases. Phase 1 (January 2020–December 2023) retrospectively analyzed CT scans from 1383 patients (2595 frontal sinuses), enabling measurement and classification of NFB anteroposterior length and FOC length. Phase 2 (January 2024–January 2025) prospectively enrolled 100 consecutive patients (191 frontal sinuses) with frontal sinus pathology requiring surgery. Radiological classifications were applied preoperatively and validated intraoperatively, with specific surgical techniques and instruments tailored to each class. All patients underwent a six-month postoperative follow-up. Inclusion criteria required patients aged 20 years or older to ensure skeletal maturity and stable frontal sinus morphology [9], with no previous sinonasal surgery. Patients with bilateral frontal sinus agenesis were excluded. Phase 1 comprised consecutive paranasal-sinus (PNS) CT scans acquired at a single institution using a uniform protocol. To ensure that the measured parameters reflected native, undistorted bony anatomy, scans demonstrating disease extending into or remodeling the frontal sinus or frontal recess (e.g., frontal mucocele, frontal sinus opacification with bony erosion, or frontally extending neoplasia) were excluded. The retained indication spectrum therefore consisted predominantly of chronic rhinosinusitis with or without polyposis and non-frontal sinonasal disease, together with scans performed for unrelated indications. This deliberate exclusion of frontal-distorting pathology is a methodological safeguard for an anatomical classification, although it necessarily limits direct extrapolation of the derived cutoffs to cases of advanced frontal disease (see Limitations). An overview of patient eligibility is shown in Figure 1.

2.2. CT Acquisition and Measurement Protocol

All CT scans were acquired at a single institution using a uniform non-contrast paranasal-sinus protocol on multidetector scanners with submillimetric slice thickness (≤1 mm) and high-resolution bone algorithm reconstruction. In the prospective phase, preoperative CT was obtained within a standard preoperative window; because the measured parameters are bony, they are not expected to change over this interval. Axial datasets were reformatted into coronal and sagittal multiplanar reconstructions (MPR) for precise anatomical evaluation. All measurements were performed on parasagittal reconstructed images using bone window settings.

The standardized measurement protocol involved simultaneous display of coronal and sagittal MPR to identify the frontal recess region, identification of the NFB as the anterior bony boundary of the frontal sinus outflow tract on coronal images, and selection of the parasagittal plane corresponding to maximal anteroposterior projection of the NFB. This plane was intentionally selected to represent the limiting surgical clearance encountered during endoscopic frontal sinus access, rather than an averaged anatomical dimension. Three reference lines were then constructed: Line X from the anterior table of the frontal sinus to the most inferoposterior point of the NFB; Line Y along the anterior wall of the NFB following its natural axis; and Line Z from the intersection of Lines X and Y through the NFB tip posteriorly to the skull base. The anteroposterior NFB length was recorded as the distance between the intersection point and the NFB tip. The FOC, representing the available anteroposterior surgical space between the NFB and skull base, was recorded as the distance along Line Z from the NFB tip to the skull base. It is important to clarify that the FOC is measured specifically at the level of the frontal ostium—the infundibular entrance from the frontal sinus into the frontal recess—and is intended as a focused surrogate for the narrowest antero-posterior bony corridor an instrument must traverse, rather than as a description of the frontal recess as a whole. For readers accustomed to viewing the recess and sinus on sagittal images, this corresponds to the constriction immediately at the ostial isthmus. All reference lines and measurements are illustrated in Figure 2.

2.3. Method Equivalence Analysis

To assess whether coronal-guided plane selection altered measurement outcomes, 200 randomly selected Phase 1 patients (right frontal recess only) were evaluated by two independent reviewers using two approaches: (1) direct selection of the parasagittal plane showing maximal NFB anteroposterior projection, and (2) a coronal-guided standardized approach identifying the first coronal slice demonstrating patent communication between the frontal recess and nasal cavity, with sagittal plane correlation through the center of the frontal recess/ostium pathway (Figure 3). Agreement was assessed using Spearman correlation and Bland–Altman analysis.

2.4. CT Classification Development

Phase 1 measurements (2595 frontal sinuses, 1383 patients) were analyzed via histograms and kernel density plots, revealing inflection points at approximately 6 mm and 12 mm for both variables. Unsupervised K-means clustering analysis independently validated these cutoffs. For NFB length, three clusters centered at approximately 7.2 mm, 10.0 mm, and 12.8 mm were identified. FOC length showed three clusters centered at 4.1 mm, 7.9 mm, and 13.4 mm. These cluster centroids supported classification into three classes: 0–6 mm (Class A NFB, Class C FOC), 6–12 mm (Class B for both), and ≥12 mm (Class C NFB, Class A FOC). The density plots and K-means clustering appear in Figure 4.

2.5. Surgical Intervention

In Phase 2, 100 eligible patients (191 frontal sinuses) underwent endoscopic sinus surgery performed under general anesthesia in the reverse Trendelenburg position using a rigid 4K video endoscope with four angled scopes (0°, 30°, 45°, and 70°). Preoperative preparation included nasal decongestion via Merocel soaked in adrenaline-saline solution (Figure 5a). Following uncinectomy (Figure 5b), middle meatal antrostomy (Figure 5c), removal of agger nasi cells when present (Figure 5d), and middle turbinate axilla resection when indicated (Figure 5e), the NFB and FOC were identified and measured intraoperatively using a surgical ruler with 1 mm resolution (Figure 5f), then categorized using the same CT cutoff values. Preoperative CT scans were reviewed as part of routine surgical planning without performing formal measurements or assigning classes; CT-based classification was performed separately according to the predefined protocol.

Subsequent surgical steps were tailored to the NFB and FOC classes. A consistent pattern was observed in which Class A NFBs frequently corresponded with Class A FOCs, whereas Class B and Class C NFBs similarly aligned with their respective FOC classes. When classes did not align, the NFB class guided subsequent steps. In cases with Class A NFB and FOC, the NFB was left unaltered and the frontal sinus was well visualized with a 70° scope (Figure 6a). When both were Class B, the posterior two-thirds of the NFB were shortened via a frontal sinus punch (70°, small) (Figure 6b,c). In Class C cases, a straight or curved 45° drill removed the posterior two-thirds and part of the anterior one-third of the NFB (Figure 6d,e). Class-specific surgical techniques are summarized in

Table 1. Summary of the novel classification and the associated surgical techniques.

Nasofrontal Beak Anteroposterior Length		Frontal Ostium Clearance Length
Class A	≤6 mm	Class A	≥12 mm
Class B	6–12 mm	Class B	6–12 mm
Class C	≥12 mm	Class C	≤6 mm
Surgical Techniques (common pairs §)
Class A NFB, Class A FOC: The NFB was left unaltered; the operation continued based on pathology.
Class B NFB, Class B FOC: The posterior two-thirds of the NFB was shortened using a frontal sinus punch (70°, small).
Class C NFB, Class C FOC: A straight or curved (45°) drill was used to remove the posterior two-thirds and part of the anterior one-third of the NFB.

^§ When FOC and NFB classes do not align, the NFB classification guides the surgical approach.

. Complication parameters recorded included cerebrospinal fluid leakage, anterior ethmoid artery injury, and lamina papyracea injury. Postoperative follow-up (two weeks, then monthly for six months) assessed neo-ostium stenosis, hyposmia, and anosmia.

2.6. Interobserver Agreement and Statistical Analysis

Interobserver reliability was assessed using 50 randomly selected CT scans independently evaluated by eight raters. The ICC (two-way random-effects model, ICC(2,k)) for absolute agreement was 0.988 (95% CI: 0.983–0.993), indicating excellent agreement. The remaining Phase 1 scans were distributed evenly among observers.

Data were analyzed using R (v4.5.0). Continuous variables were non-normally distributed (Shapiro–Wilk test) and reported as median and interquartile range. Associations between categorical variables were assessed via chi-square test. Spearman correlation determined correlations between continuous variables; Bland–Altman plots assessed measurement reliability. Statistical significance was set at p < 0.05. Given the retrospective nature of Phase 1, no a priori power calculation was performed; the sample size (1383 patients, 2595 sinuses) substantially exceeded typical frontal sinus studies. Phase 2 was designed as a prospective validation cohort to assess agreement between imaging-derived measurements and intraoperative findings rather than to test a specific hypothesis. For the prospective phase, agreement between CT-based and intraoperative classes was quantified using complete three-by-three confusion matrices, from which per-class sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and overall accuracy were derived for both the NFB and FOC classifications. Ordinal agreement beyond chance was assessed using weighted Cohen’s kappa (both linear and quadratic weighting), reflecting the ordered nature of Classes A, B, and C.

3. Results

A total of 1483 patients were enrolled (1383 in Phase 1: 2595 frontal sinuses; 100 in Phase 2: 191 frontal sinuses). The median age was 38 years (IQR: 30–48). Right NFB length: 9.64 mm (IQR: 7.97–11.26); left: 9.59 mm (IQR: 8.18–10.91). Right FOC: 6.09 mm (IQR: 4.18–8.27); left: 6.55 mm (IQR: 4.34–8.85). Combined median NFB length was 9.62 mm (IQR: 8.07–11.06) and FOC was 6.28 mm (IQR: 4.33–8.64). Categorical variables appear in

Table 2. Categorical variables.

Variables	Phase 1—n (%)	Phase 2—n (%)	Total
Sex
Male	828 (59.9%)	57 (57.0%)	885 (59.7%)
Female	555 (40.1%)	43 (43.0%)	598 (40.3%)
Frontal Sinus Side
Right	1305 (50.3%)	98 (51.3%)	1403 (50.4%)
Left	1290 (49.7%)	93 (48.7%)	1383 (49.6%)
NFB Length Classes
Class A (≤6 mm)	123 (4.7%)	10 (5.2%)	133 (4.8%)
Class B (6–12 mm)	2100 (80.9%)	146 (76.4%)	2246 (80.6%)
Class C (≥12 mm)	372 (14.3%)	35 (18.4%)	407 (14.6%)
FOC Length Classes
Class A (≥12 mm)	198 (7.6%)	23 (12.0%)	221 (7.9%)
Class B (6–12 mm)	1182 (45.5%)	68 (35.6%)	1250 (44.9%)
Class C (≤6 mm)	1215 (46.8%)	100 (52.4%)	1315 (47.2%)

. A subtle side-to-side asymmetry was present, with slightly longer beak lengths and smaller ostium diameters on the right.

3.1. Findings from Phase 1

A total of 2595 frontal sinuses in 1383 patients were retrospectively analyzed. Spearman correlation tests assessed associations between patient age and NFB length (ρ = −0.037, p = 0.060) or FOC length (ρ = 0.065, p < 0.001). Although the age–FOC correlation was statistically significant, both measurements showed very weak age correlation. Both measurements demonstrated strong statistical significance for sex associations and cross-correlations between NFB and FOC (

Table 3. Association between NFB and FOC classes and sex in Phase 1 patients.

		Male	Female	p-Value
NFB Classification	Class A	63 (51.2%)	60 (48.8%)	<0.001
	Class B	1222 (58.2%)	878 (41.8%)
	Class C	273 (73.4%)	99 (26.6%)
FOC Classification	Class A	100 (50.5%)	98 (49.5%)	0.017
	Class B	717 (60.7%)	465 (39.3%)
	Class C	741 (61.0%)	474 (39.0%)

, Figure 7). Method equivalence analysis demonstrated excellent agreement between original and coronal-guided approaches. For NFB length, Spearman correlation was strong (ρ = 0.961, p < 0.001) with a mean bias of −0.008 mm and 95% limits of agreement from −1.32 to +1.30 mm. FOC showed strong correlation (ρ = 0.964, p < 0.001) with mean bias of −0.025 mm and limits of agreement from −1.40 to +1.35 mm, confirming that the coronal-guided approach enhances standardization without materially affecting measurement values.

3.2. Findings from Phase 2

Phase 2 evaluated 100 patients (191 frontal sinuses). Medians: CT NFB anteroposterior length 9.12 mm (IQR: 7.18–11.25), intraoperative NFB 8 mm (IQR: 6–10), CT FOC 5.62 mm (IQR: 3.36–10.52), intraoperative FOC 5 mm (IQR: 3–7). Bland–Altman analysis and Spearman correlation between CT and intraoperative measurements appear in Figure 8. Mean bias for NFB was +1.22 mm (95% limits: +0.59 to +1.86); for FOC, +0.39 mm (95% limits: −6.22 to +6.99). These findings indicate close agreement for NFB and wider variability for FOC, consistent with the limitations of intraoperative endoscopic measurement.

Strong concordance was observed between radiological and surgical grading (

Table 4. Association between nasofrontal beak (NFB) classes: CT vs. surgical.

		Surgical Class A	Surgical Class B	p-Value
CT Classification	Class A	10 (100%)	0 (0%)	<0.001
	Class B	49 (33.6%)	97 (66.4%)
	Class C	0 (0.0%)	12 (34.3%) *

* Remaining 23 (65.7%) confirmed as Class C intraoperatively.

and

Table 5. Association between frontal ostium clearance (FOC) classes: CT vs. surgical.

		Surgical Class A	Surgical Class B	p-Value
CT Classification	Class A	85 (85%)	15 (15%)	<0.001
	Class B	39 (57.4%)	25 (36.8%)
	Class C	3 (13.0%)	13 (56.5%)

Note: Remaining Class C cells: Class C FOC 7 (30.4%) intraoperatively.

). All CT Class A NFBs were confirmed intraoperatively as Class A. Two-thirds of Class B NFBs were confirmed as Class B, with the remaining cases classified as Class A. Most Class C NFBs corresponded to intraoperative Class C. Overall, the CT-based NFB classification demonstrated a robust ability to predict intraoperative anatomy. Quantitatively, the NFB classification achieved an overall accuracy of 68.1% (130/191) with a quadratic-weighted Cohen’s kappa of 0.55 (linear-weighted kappa 0.46). Per-class performance was as follows: Class A sensitivity 100% and specificity 72.9% (NPV 100%); Class B sensitivity 66.4%, specificity 73.3% (PPV 89.0%); and Class C sensitivity 65.7%, specificity 100% (PPV 100%) (

Table 6. Per-class diagnostic performance of the CT-based NFB classification against intraoperative findings (overall accuracy 68.1%; quadratic-weighted κ = 0.55, linear-weighted κ = 0.46). PPV, positive predictive value; NPV, negative predictive value.

CT Class	Sensitivity	Specificity	PPV	NPV	n (CT)
Class A (≤6 mm)	100.0%	72.9%	16.9%	100.0%	10
Class B (6–12 mm)	66.4%	73.3%	89.0%	40.2%	146
Class C (≥12 mm)	65.7%	100.0%	100.0%	92.9%	35

and

Table 7. Per-class diagnostic performance of the CT-based FOC classification against intraoperative findings (overall accuracy 61.3%; quadratic-weighted κ = 0.50, linear-weighted κ = 0.39). PPV, positive predictive value; NPV, negative predictive value.

CT Class	Sensitivity	Specificity	PPV	NPV	n (CT)
Class A (≥12 mm)	30.4%	97.6%	63.6%	91.1%	23
Class B (6–12 mm)	36.8%	77.2%	47.2%	68.8%	68
Class C (≤6 mm)	85.0%	53.8%	66.9%	76.6%	100

). The low Class A PPV (16.9%) reflects the small number of Class A beaks (n = 10) and the conservative tendency to reclassify short beaks “up” intraoperatively, and should be interpreted in light of this limited denominator. On the other hand, CT-based and intraoperative FOC classifications demonstrated more modest agreement (overall accuracy 61.3%; quadratic-weighted kappa 0.50, linear-weighted kappa 0.39), with per-class sensitivities ranging from 30% to 85% and correspondingly wider variability. The full confusion matrices and per-class diagnostic metrics for both classifications are presented in Table 4 and Table 5. These results confirm that the NFB axis is the more reliable and reproducible component of the system, consistent with our recommendation that it take precedence when the two classes diverge. One patient with radiological Class C NFB and Class C FOC sustained a lamina papyracea injury; the periosteum remained intact, requiring no further intervention. Postoperatively, no patients experienced hyposmia or anosmia. Two patients exhibited neo-ostium narrowing at the final follow-up, both with Class C NFB (one with Class C FOC, one with Class B), both radiologically. The endoscope angulation required for adequate visualization also tracked the NFB class: Class A beaks were generally well visualized with 0° to 30° optics, whereas Class B and especially Class C beaks more frequently required 45° and 70° endoscopes, reinforcing the practical, instrument-oriented intent of the classification.

4. Discussion

Although the literature acknowledges the importance of anatomical landmark in frontal recess surgery, few studies have quantified surgical difficulty and complication probability through measurements [10,11,12,13,14]. Gheriani et al. graded the frontal ostium radiologically using two vertical parasagittal lines, yielding four difficulty levels [15]. However, this approach inadequately accounts for NFB length. Several studies (Makihara et al., Craig et al., Park et al.) measured NFB length and frontal ostium diameter using parasagittal views without reference lines, reducing reproducibility [16,17,18]. Unlike approaches relying on averaged or arbitrarily selected parasagittal planes, the present method deliberately targets the limiting anteroposterior clearance within the frontal recess, reflecting the worst-case scenario during endoscopic instrumentation—a limitation repeatedly emphasized in foundational frontal recess surgery literature [19]. Additionally, these prior studies lack specific surgical applicability.

Given the absence of a standardized NFB classification with tailored surgical techniques, this study accomplished both goals. Phase 1 measured NFB and FOC lengths via a new surgically applicable method (Figure 2) in 2595 frontal sinuses. Statistical analysis (Figure 4) definitively determined cutoffs: Class A (≤6 mm), Class B (6–12 mm), and Class C (≥12 mm) for NFB, and Class A (≥12 mm), Class B (6–12 mm), and Class C (≤6 mm) for FOC. The FOC cutoffs differ from Gheriani et al.’s frontal ostium grade (7.5 mm) [15], though they are close to the observed cluster centroids (4.1, 7.9, 13.4 mm). Importantly, the 6 mm and 12 mm thresholds were not selected arbitrarily or merely for ease of recall; they emerged concordantly from histogram inflection points, kernel-density structure, and independent unsupervised K-means clustering of a large sample (2595 sinuses), and were subsequently rounded to clinically operable integer values. This data-driven derivation reduces—though does not eliminate—the risk of overfitting inherent to single-center cutoff selection. Method equivalence analysis confirmed near-identical measurements between original and coronal-guided approaches, validating that the proposed metrics reflect stable anatomical relationships rather than slice-selection artifacts.

Phase 2 performed radiological and intraoperative classification for 100 patients (191 frontal sinuses). Bland–Altman plots between CT and surgical measurements showed acceptable agreement and correlation (Figure 8), confirming real-life applicability. Both NFB and FOC classifications demonstrated statistically significant associations between CT and intraoperative findings (Table 4 and Table 5), though concordance strength differed. NFB classification showed stronger radiological-surgical correspondence, whereas FOC showed only moderate agreement, likely reflecting the limitations of intraoperative measurement within confined endoscopic fields. These findings support prioritizing the NFB class when classifications do not align, as the NFB represents a more stable and reliably assessed anatomical determinant.

Comparing current measurements with published data, the median NFB length (9.62 mm) exceeds the values reported by Burkart and Zimmer (0.8 ± 0.3 cm) and Park et al. (7.8 ± 1.8 mm). Median FOC (6.28 mm) is lower than previously reported frontal ostium diameters [18,20]. These differences may reflect not only anatomical variability across populations but also methodological differences in how measurements were acquired and defined. Statistically significant sex-based variation was observed, with males more frequently associated with Class C NFB and FOC (Table 3), suggesting greater bony prominence and narrower drainage pathways in males, contrasting with Burkart and Zimmer’s findings [20]. Subtle side-to-side asymmetry was noted, consistent with previously reported variations [15,21].

Contemporary literature increasingly emphasizes tailoring instrumentation to anatomical constraints [8,22,23,24]. While many publications explore Draf procedure selection based on underlying pathology, few systematically examine how anatomical variations influence instrument choice. This study assigned specific instruments to each class: no advanced instrumentation for Class A, frontal sinus punches for Class B, and straight or curved drills for Class C. This structured allocation provided a transparent, anatomy-based framework for preoperative planning and instrument selection. Although only one intraoperative lamina papyracea injury and two cases of postoperative neo-ostium narrowing were observed, we emphasize that the study included no comparator arm; these uncontrolled counts cannot be attributed to the classification and may equally reflect operator experience. Accordingly, we make no claim of reduced complications or improved clinical outcomes and present the surgical phase as a demonstration of anatomical–surgical concordance and feasibility rather than of clinical superiority.

The proposed classification is not a new anatomical taxonomy and does not replace the International Frontal Sinus Anatomy Classification (IFAC), which describes frontoethmoidal cell anatomy. Rather, it complements existing nomenclature by translating radiological anatomy into a pragmatic, instrument-oriented metric that directly informs intraoperative decision-making. Both systems may be used synergistically, with the IFAC guiding anatomical understanding and this metric supporting procedural planning.

This study has limitations. In the prospective phase, intraoperative measurements were performed by a single experienced surgeon, as accurate endoscopic frontal recess assessment requires advanced expertise. Although preoperative CT evaluations were conducted without formal measurements to minimize anticipation bias, the intraoperative assessment could not be fully blinded. The surgical component should be interpreted as evaluating anatomical–surgical concordance and internal construct validity rather than independent external validation. Nevertheless, the consistent radiological-surgical concordance observed for NFB categorization supports the internal validity and feasibility of the proposed classification. Future multicenter studies with multiple surgeons and blinded assessment are required to establish external validity and inter-operator reproducibility. Several further limitations warrant emphasis. First, the classification was both derived and tested within a single institution; no independent external cohort was available, so the generalisability of the 6 mm and 12 mm cutoffs and the risk of overfitting cannot be excluded, notwithstanding their support by unsupervised clustering. Second, intraoperative linear measurement within the confined frontal recess is inherently imprecise; this is reflected in the wide limits of agreement for FOC (−6.99 to +6.22 mm) compared with the narrow limits for NFB, and measurements were obtained by a single surgeon without intraoperative interobserver replication. Third, navigation-based and postoperative-CT confirmation of the achieved clearance were not performed and would strengthen future validation. Fourth, two linear parameters cannot capture the full complexity of frontal recess anatomy; the NFB/FOC metric is intended to complement, not replace, comprehensive cell-based assessment and the IFAC. The sagittal angulation of the outflow tract is a logical further parameter for future refinement. Taken together, the present work should be regarded as a single-center derivation with internal prospective concordance testing, requiring multicenter external validation before broad clinical adoption.

5. Conclusions

This study introduces a CT-derived, measurement-based stratification of the nasofrontal beak and frontal ostium clearance, tested prospectively against intraoperative findings within a single center. The NFB classification demonstrated good radiological–surgical concordance (overall accuracy 68.1%; quadratic-weighted kappa 0.55), whereas the FOC classification showed more modest agreement, supporting prioritization of the NFB axis when the two diverge. The system provides a reproducible, instrument-oriented adjunct for preoperative planning rather than a demonstration of clinical superiority or a replacement for comprehensive anatomical assessment. As a single-center derivation, it requires multicenter external validation before broad clinical adoption, and the integration of additional parameters such as outflow-tract angulation may further refine its utility.