This was a retrospective descriptive and analytical study, conducted at the Radiology Department of the Bouaké Regional Hospital Center (CHR), over a four-month period from March to June 2026.

The study covered the cranial CT images of 150 Ivorian patients of both sexes, aged 20 years or over in order to ensure complete sphenoid pneumatization. A non-probability exhaustive (consecutive) sampling method was used: all images obtained during the period and meeting the criteria below were included, without random selection or prior sample-size calculation. Of the 184 cranial CT examinations assessed for eligibility during the study period, 34 were not included: 15 patients younger than 20 years, 7 examinations with slice thickness greater than 3 mm, 5 with low resolution or artifacts in the study area, 4 with the sphenoid sinus not or poorly visualized, and 3 in patients with a history of sphenoid surgery; 150 examinations were retained for analysis (Figure 1).

Figure 1. Study flow diagram: 184 CT examinations assessed for eligibility, 34 not included (by criterion) and 150 retained for analysis.

Inclusioncriteria: Ivorian patients of both sexes, aged 20 years or over, who had undergone a cranial CT scan with bone window (with or without contrast injection), and presenting images of usable morphometric quality.

Non-inclusioncriteria: slice thickness greater than 3 mm, low-resolution images or those with artifacts in the study area, sphenoid sinus not visualized or poorly visualized, history of sphenoid sinus surgery.

Images were acquired on a General Electric 64-slice scanner (140 kV, 144 mAs), with 1.5- to 3-mm slices in the axial plane, reconstructed in the coronal and sagittal planes, then processed with AW Volume Share 7 software and analyzed in bone window. Each examination was reviewed jointly by a neurosurgeon and a radiologist, who recorded a single consensus rating at the time of reading; the scans were not rated independently beforehand, so inter-observer agreement could not be estimated. The type of pneumatization was classified according to the Hammer and Radberg system [12], supplemented by the sellar/postsellar distinction of later classifications [13][14] (Figure 2), and the morphology of the sella turcica according to the Axelsson classification [16].

Figure 2. Types of sphenoid sinus pneumatization (sagittal section; Hammer and Radberg classification). The extent of the aerated cavity (blue) progresses from front to back; the red lines mark the anterior wall (tuberculum) and posterior wall (dorsum) of the sella turcica. The percentages correspond to the present series (n = 150).

Neurovascular variants were assessed on each side on bone-window reconstructions. Protrusion was defined as bulging of more than half of the circumference of the ICA or optic nerve canal into the sphenoid sinus cavity. Dehiscence was defined as the absence of an identifiable bony covering between the neurovascular structure and the sinus air cavity; dehiscence could therefore be recorded in the absence of protrusion.

A standardized form recorded, for each patient: demographic data (sex, age); sinus characteristics (main type of pneumatization and variants, presence and characteristics of the septation, ICA and optic nerve variants, sella turcica morphology); and four sphenoid distances measured on the reconstructions (Figure 3): nasal orifice-sphenoid sinus distance, anterior edge-posterior edge of the sinus distance, sinus superior edge (or anterior skull base)-sphenoid ostium distance, and sinus floor-sphenoid ostium distance.

Figure 3. Schematic representation of the four sphenoid distances measured on the sagittal CT reconstruction (sphenoid sinus, ostium and nasal cavity).

The analysis followed the SAMPL guidelines. Data were entered in Microsoft Excel 2019 and analyzed with SPSS version 25. Qualitative variables are expressed as counts and percentages. Normality of each distance within each sex was examined with the Shapiro-Wilk test; it was not rejected for any distance in either sex (all p ≥ 0.133). In line with the prespecified plan, distances are therefore summarized as mean ± standard deviation and compared between sexes with the t test, in its Welch form because the variances were larger in women; medians [interquartile range] with the Mann-Whitney U test are reported as a sensitivity analysis. Qualitative variables were compared with the chi-square test, Fisher’s exact test or the Fisher-Freeman-Halton exact test. Age was analyzed as a continuous variable against each distance with Spearman’s rank correlation (ρ), which avoids arbitrary age grouping. Effect estimates accompany the p values. For ICA protrusion, they comprise the proportion affected within each pneumatization type (Wilson 95% confidence interval [CI]), the risk difference between the broad sellar and the presellar/conchal types (Newcombe hybrid score CI), the odds ratio between postsellar and sellar types (conditional exact CI) and Cramér’s V. For distances, they comprise the mean difference between men and women with its Welch 95% CI and Hedges’ g; for correlations, ρ with a 95% CI obtained by Fisher’s z transformation. These estimates were computed from the tabulated data with Python (SciPy 1.17, statsmodels 0.15). The significance threshold was set at p < 0.05. Because about twenty tests were performed without a prespecified primary hypothesis, a Bonferroni-adjusted threshold is reported as a sensitivity analysis (0.05/15 = 0.0033 for the associations with the type of pneumatization; 0.05/4 = 0.0125 for the sex comparisons of distances), and all analytical findings are interpreted as exploratory.

The study used anonymized archived CT images acquired between March and June 2026 as part of routine clinical care. The protocol subsequently received a favorable opinion from the institutional ethics committee of Bouaké University Hospital (opinion no. CEI/CHU-BKE/230/2026; written notification received on 18 August 2026). Study-specific extraction, review and analysis of the archived images were undertaken after this approval. Given the retrospective design and the exclusive use of anonymized archived images, the committee waived the requirement for individual informed consent.

Demographiccharacteristics. Of 150 patients, 97 were male (64.7%) and 53 female (35.3%), a sex ratio of 1.83. Mean age was 39.3 ± 13.9 years (median 35.5 years; range 20 - 79 years). The [20 - 35[ year age group was the most represented (45.3%), followed by the [35 - 50[ (32.7%), [50 - 65[ (16.7%) and [65 - 80] (5.3%) groups (Table 1).

Table 1. Distribution of patients by sex and age group.

Maintypeofpneumatization. The sellar type was the most frequent (60 patients; 40.0%), followed by the postsellar (55; 36.7%), presellar (32; 21.3%) and conchal (3; 2.0%) types (Table 2, Figure 4). Combining sellar and postsellar, broad sellar pneumatization concerned 76.7% of patients.

Table 2. Distribution of patients by main type of pneumatization.

Figure 4. Distribution of sphenoid sinus pneumatization types (n = 150).

Variants,septation,neurovascularvariantsandsellaturcica. Associated pneumatization variants were found in 28 patients (18.7%), exclusively within the sellar and postsellar types, dominated by lateral pneumatization (16 cases; 10.7%), foremost among which the “greater wing of the sphenoid” variant (11 cases). A septation was present in 144 patients (96.0%), single in 88.7% of cases and multiple in 7.3%; among single septations, the complete vertical form predominated (122 cases; 91.7%) and median positioning was the most frequent (66 cases; 49.6%). Variants of the ICA and/or optic nerve were identified in 44 patients (29.3%): ICA protrusion in 15 (10.0%), optic nerve dehiscence in 14 (9.3%), ICA dehiscence in 10 (6.7%), and optic nerve protrusion in 5 (3.3%). Among these 44 patients, 20 (45.5%) had bilateral involvement and 24 (54.5%) had unilateral involvement. The sella turcica was normal in 85 patients (56.7%), with posterior wall irregularity in 57 (38.0%) and another unclassifiable morphology in 8 (5.3%) (Table 3).

Table 3. Distribution of septation, ICA/optic nerve variants and sella turcica morphology.

Sphenoidmeasurements. The nasal orifice-sphenoid sinus and anterior edge-posterior edge distances were on average greater in men, whereas the sinus superior edge-ostium and floor-ostium distances were on average greater in women. Table 4 gives the four distances as mean ± SD with the Shapiro-Wilk normality test for each sex; effect estimates and between-sex comparisons are reported below with the analytical results.

Table 4. Sphenoid distances (mm) by sex: mean ± standard deviation and Shapiro-Wilk normality test.

Associations between the type of pneumatization and the main characteristics were tested (Table 5). No significant difference was observed with sex, age group (overall or by sex), the presence or type of lateral variant, the presence, number, orientation or positioning of the septation, septal deviation, ICA or optic nerve dehiscence, optic nerve protrusion, or sella turcica morphology. Only ICA protrusion differed across pneumatization types (Fisher-Freeman-Halton exact test, p = 0.012; Cramér’s V = 0.27). Protrusion was absent in the conchal (0/3) and presellar (0/32) types and was observed in 6.7% (95% CI 2.6 to 15.9) of sellar and 20.0% (95% CI 11.6 to 32.4) of postsellar sinuses. The risk difference between the broad sellar and the presellar/conchal types was 13.0 percentage points (95% CI 2.0 to 20.4), and the odds of protrusion were higher in postsellar than in sellar sinuses (odds ratio 3.5, exact 95% CI 0.94 to 15.9). Resting on 15 events, with two empty cells and a p value above the Bonferroni threshold of 0.0033, this association is exploratory (Table 6, Figure 5).

Table 5. Associations tested between the type of pneumatization and patient characteristics (chi-square/Fisher test; unadjusted p values, Bonferroni threshold 0.0033).

Table 6. Distribution of the type of pneumatization by internal carotid artery (ICA) protrusion.

^aWilson score 95% CI. Other percentages refer to the whole sample (n = 150). Broad sellar vs presellar/conchal: 15/115 (13.0%) vs 0/35, risk difference 13.0 percentage points (95% CI 2.0 to 20.4; Newcombe). Postsellar vs sellar: odds ratio 3.5 (exact 95% CI 0.94 to 15.9). V: Cramér’s V. The p value exceeds the Bonferroni threshold (0.0033): exploratory association.

Figure 5. Frequency of internal carotid artery protrusion by type of pneumatization (Fisher-Freeman-Halton exact test, p = 0.012; exploratory association, see Table 6 for 95% CIs).

Comparisonofdistancesbysexandage. Three of the four distances differed between men and women with the Welch t test. The nasal orifice-sphenoid sinus distance was longer in men (mean difference 5.62 mm, 95% CI 3.95 to 7.29; Hedges’ g 1.26; p < 0.001), as was the anterior edge-posterior edge distance (1.20 mm, 95% CI 0.07 to 2.33; g 0.40; p = 0.038), whereas the sinus superior edge-ostium distance was longer in women (1.14 mm, 95% CI 0.28 to 2.00; g 0.46; p = 0.010). The floor-ostium distance did not differ (0.50 mm longer in women, 95% CI −0.10 to 1.10; p = 0.099). After Bonferroni correction (threshold 0.0125), the nasal orifice-sphenoid sinus and superior edge-ostium differences remained significant; the anterior edge-posterior edge difference did not, and it was also nonsignificant in the Mann-Whitney sensitivity analysis (p = 0.092), so it should be regarded as uncertain (Table 7, Figure 6). Age, analyzed as a continuous variable, was not correlated with any distance: all |ρ| were below 0.12, and the 95% CIs exclude correlations stronger than 0.28 in absolute value (Table 8).

Table 7. Comparison of sphenoid distances between men and women: effect estimates, Welch t test and Mann-Whitney sensitivity analysis.

^aMen minus women, with Welch 95% CI (primary analysis). ^bMann-Whitney U test (sensitivity analysis). All p values unadjusted; Bonferroni threshold for four comparisons: 0.0125.

Figure 6. Sphenoid distances by sex (median, interquartile range and individual distribution; men = 97, women = 53). The p values shown are those of the Mann-Whitney sensitivity analysis; the primary comparisons (Welch t test) and effect estimates are given in Table 7.

Table 8. Correlation between age and sphenoid distances (Spearman coefficient ρ with 95% CI).

Our population was predominantly young (nearly three quarters under 50) and male (sex ratio 1.83). This predominance was also reported by Yataco-Wilcas et al. [17] in Peru and by Dogan et al. [2] in Turkey, the latter having demonstrated an association between the type of pneumatization and age group, unlike our work, in which no association was found with sex or age.

The broad sellar type was the most frequent (76.7%), with the presellar accounting for 21.3% and the conchal 2.0%. This hierarchy is consistent with the systematic review by Bechev et al. [3] (sellar type ≈ 72%, presellar 10% - 15%, conchal 2% - 5%) and with the Ethiopian study by Degaga et al. [18]. The isolated proportion of strict sellar type (40.0%) remained lower than that of several earlier series (50% [18], 52.9% [10], 59.4% and 69% [19]), suggesting ethno-geographic variants of pneumatization [3].

Lateral pneumatization (10.7%) remained below the frequency of about 20% reported in the literature [3]. Septation was present in 96.0% of patients, single in 88.7%, a proportion close to the 90% reported in Ethiopia [18], multiple septation (7.3%) being rarer than the usual 20% - 35% [3]. No association was found between the type of pneumatization and septal characteristics. ICA protrusion and dehiscence (10.0% and 6.7%) and those of the optic nerve (3.3% and 9.3%) were lower than the Ethiopian values [18][20][21]; differences in the CT definitions of protrusion and dehiscence across series may contribute to these gaps. ICA protrusion was confined to sellar and postsellar sinuses and was most frequent in the postsellar type (20.0%, 95% CI 11.6 to 32.4), consistent with wall thinning as pneumatization extends behind the sella [3][4]. This estimate, however, rests on 15 events and did not withstand correction for multiple testing; it should be regarded as a hypothesis to be confirmed in larger series.

The sella turcica was normal in 56.7% of patients, posterior wall irregularity—which may reflect a thinning exposing to intraoperative cerebrospinal fluid leakage—being the most frequent abnormality (38.0%) [2]. Unlike Dogan et al. [2], no significant association with the type of pneumatization was demonstrated (p = 0.674), sellar pneumatization moreover being predominant in our series.

The nasal orifice-sphenoid sinus distance was longer in men by about 5.6 mm (95% CI 4.0 to 7.3), a large and robust difference also found in Peru [17]. The sinus superior edge-ostium distance was longer in women by about 1.1 mm (95% CI 0.3 to 2.0), unlike the Peruvian series [17], a discrepancy possibly related to the interindividual variability of the sinus [3]. The anteroposterior depth of the sinus was slightly greater in men (1.2 mm, 95% CI 0.1 to 2.3), but this difference did not withstand correction for multiple testing or the rank-based analysis, and the floor-ostium distance did not differ by sex. The width of these intervals, of the order of millimeters, is comparable to the margins that matter in the sellar approach, which is a further argument for individual measurement. No meaningful correlation was found between age and the distances (|ρ| < 0.12), suggesting stable sphenoid dimensions in adulthood [3].

The predominance of sellar and postsellar pneumatization indicates that, in most patients of this series, the sphenoid sinus offers a potentially favorable bony corridor to the sella. This anatomical observation does not, by itself, establish that the transsphenoidal approach is feasible or safe in a given patient: ICA protrusion was found precisely in the most extensively pneumatized sinuses, and septal insertions, dehiscences and distances vary widely from one patient to another [3][22]. Each patient therefore still requires a specific preoperative CT assessment of the corridor, of the septa and of the carotid and optic relationships, with instruments and trajectory adapted to the measured distances [17][23][24], together with careful intraoperative observation [25][26]. Our data can guide the reading of these examinations locally, not replace it.

This study has the limitations of a retrospective single-center work. The sample consisted of patients referred for cranial CT, who may not represent the general population, and no a priori sample-size calculation was performed; classification criteria may also differ across published series [3]. Scans were rated in consensus by two readers without prior independent reading, so inter-observer reproducibility could not be quantified, and the distinction between protrusion and dehiscence remains partly subjective on CT. The absence of association for some variables may reflect a lack of power related to rare variants rather than a true absence of relationship, and several contingency tables contained sparse or empty cells, which makes the effect estimates imprecise. Finally, about twenty tests were performed: only the sex differences in the nasal orifice-sphenoid sinus and superior edge-ostium distances withstand Bonferroni correction, and the association with ICA protrusion must be considered exploratory.