This study has shown that CT is a robust method in the estimation of different dimensions of the maxillary sinuses, frontal sinuses and the adjacent structures as the interobserver agreement ranges from substantial to almost perfect dependent on the measurement in question (Table 1). Despite the moderate interobserver agreement with regard to the measurements of the canine fossa and the orbital floor (ICC ranging between 0.50 and 0.60), the random error was only 0.3-0.4 mm. This depends partly on the fact that these structures are very thin and partly on the limitation of measurements of tiny structures in PACS. The results in the present study are of importance when setting the adjustments of a clinical applicable Doppler ultrasound equipment for the diagnose of rhinosinusitis.
Previous studies have shown that dimensions of maxillary sinuses from measurements on human skulls were similar to those obtained by CT scans  and the consistency of measurements of the paranasal sinuses using CT images have been evaluated in the last decade [2, 5, 10]. Some authors have measured the volume by directly injecting different materials into the paranasal sinuses [11, 12]. However, this procedure cannot be used in living subjects. Furthermore, using such methods in the estimation of the sinus volume usually result in underestimation of the volume in the presence of mucosal thickening and other sinus pathologies [5, 11, 12]. Our analysis was performed on head CT in patients subjected to trauma, and patients with headache, neurological deficit and stroke, epilepsy, and vertigo. Thus, our material represents individuals with no history of sinus pathology and can in practice be considered as "normal population".
The results of the maxillary sinuses measurements were consistent with previous reports [2, 5]. The mean values of the maxillary sinus volume have been reported to range from 11.1 ± 4.5 cm3 to 23.0 ± 6.7 cm3 in previous studies .
We found that there was a significant difference of the maxillary sinus volume between males and females, mainly due to the fact that male exhibit higher and wider maxillary sinuses than females. Similarly, the anteroposterior diameter of the frontal sinus was larger in men. Some authors have reported difference of the volume of the maxillary sinuses between males and females [2, 4, 13] whereas others have showed no such difference . Ariji et al have described the correlation between the craniocaudal diameter of the maxillary sinus and body height, body weight and age . As men are generally larger than women, this could explain our observed difference in gender for maxillary sinus volume.
In our work we only included adults (age 18-65 years) and we found neither significant age difference nor significant difference between the left and right maxillary sinus volume. Previous reports suggested that the maxillary sinus volume increase with both age  and loss of teeth . On the other hand Ariji found no significant difference between dentate and edentulous patients .
To our knowledge this is the first report on the thickness of the canine fossa. The bone thickness was 1.1 mm (mean value for study cohort), which correlates well with our surgical experience or when inspecting dried skulls. The thickness of the soft tissue in front of the bone of the canine fossa varied from 5 to 20 mm. These results are of special importance in our future work with the evaluation of Doppler ultrasound as a diagnostic tool for staging rhinosinusitis, as bone attenuates ultrasound waves considerably, and soft tissue does not. The volume and anteroposterior diameter of the sinuses are also relevant for the development of this new Doppler application, as we previously showed that the radius of the ultrasound transducer should correspond to half the radius of the sinus cavity . This novel application of the Doppler ultrasound technique makes it possible to determine the properties of paranasal sinus fluids safely and non-invasively. It has previously been proved that the Doppler ultrasound technique can be used to identify mucopurulent rhinosinusitis . This method could improve the diagnosis of rhinosinusitis, reduce the suffering of patients with rhinosinusitis and potentially decrease the prescription of antibiotics. This in turn would lead to a decrease in antibiotic resistance and a significant cost reduction for the health care services as a whole.
In our study we measured the anteroposterior diameter and the anterior wall thickness of the frontal sinuses at the level of the orbital roof (Figure 1F). We chose this reference point since one upon performing an ultrasound examination of the frontal sinuses usually hold the ultrasound probe against this area and it is consequently the dimensions of this area that affect the ultrasound wave of the prospective Doppler equipment. The thickness of the bone of the anterior wall of the frontal sinuses is approximately twice as thick as the anterior bony wall of the maxillary sinus in our material, which implies that the attenuation of the ultrasound waves would be much higher when examining the frontal sinuses. Subsequently, it would be hard to induce acoustic streaming in secretions in the frontal sinuses. The anteroposterior diameter of the frontal sinus at this reference point may not be the deepest of the frontal sinuses and consequently our data are difficult to compare to results of other authors.
This study showed a good concordance between the manual and automatically calculated volume of the maxillary sinus with ICC ranging between 0.90 and 0.93. The results from the automatically computed data were 14-17% higher than the manually calculated volumes, which enable a rough estimation of the maxillary sinus volume by measuring the sinus diameter in three planes. Although such estimation is not suitable for research purposes, we believe that this tool might be beneficial in clinical practice for approximate estimation of the maxillary sinus volume, where volume measurement applications are not available.
In our study, there were incidental findings of the paranasal sinuses in 35% of the patients which correlates well to previous reports, where mucosal changes in the paranasal sinuses have been detected in 17-42.5% of CT scans for non-rhinological disease [16–18]. Non-specific mucosal swelling was the commonest finding (27% of the patients) in our material, whereas the incidence of maxillary mucosal cysts was less frequent than previously reported (12.4 to 22%) [19, 20].
The measurements of this study were done by two radiologists. One of the drawbacks of this study was that some selection bias might have occurred by the subjective selection of the slice by each reader. However, the reader's choice of the slice should have been almost identical to give such a good agreement in most of the measurements that is shown in Table 1. Other drawbacks are the retrospective nature of the study and inclusion of patients rather than healthy individuals. However, the radiation doses of head CT amounts to 2-2.5 mSv, which make the exposure of healthy individuals to such high radiation doses ethically unacceptable.