ORIGINAL ARTICLE

Morphometric measurements and sexual dimorphismof the piriform aperture in adults

Eric Moreddu • Laurent Puymerail • Justin Michel •

Michael Achache • Patrick Dessi • Pascal Adalian

Received: 4 November 2012 / Accepted: 29 March 2013

� Springer-Verlag France 2013

Abstract

Purpose The aim of this study was to evaluate the

dimensions (maximal width and length), the size and the

shape of the PA and their sexual dimorphism.

Methods Using 3D-CT scan reconstructions and land-

marks positioning around the piriform aperture and on the

face, a collective of 170 non-pathologic subjects (79

female, 91 male) from Marseille (France) was examined in

classical and geometric morphometrics methods.

Results The mean width of the piriform aperture was

24.00 mm in females and 25.32 mm in males, the mean

length was 32.54 mm in females and 36.35 mm in males.

The difference between males and females was significant,

and our data correlates well with the previously data

acquired from humans skulls. Facial measurements also

showed a statistically significant dimorphism. In morpho-

metric geometrics, the correlation between the centroıd size

and PC1 in the shape space was weak, while this correlation

was strong in the size and shape space. Visualization of

shape differences was achieved on 2D wireframes.

Conclusion Shape and size analysis of the piriform

aperture showed the existence of a significant sexual

dimorphism. These results encourage us to go further with

functional and imaging correlations.

Keywords Piriform aperture � Dimensions � Sexual

dimorphism � Nose � Geometric morphometrics � 3D-CT

reconstruction

Introduction

The piriform aperture (PA) is the skeletal aperture located

in the middle part of the face and limited by the frontal

processes of the maxillary bones, the nasal bones, and the

anterior nasal spine [8]. It corresponds to the anterior limit

of the skeletal nose, and a major component of the size of

the nose.

It is one of the upper respiratory passages, involved in

warming, filtration, moistening and guiding the incoming

air. A good knowledge of this element is fundamental in

understanding these nasal functions. However, the articles

published in the literature mainly correlate the functional

problems of the nose to the nasal mucosa or cartilages, but

few of them study the role of the PA, except for the well-

described pediatric stenosis of the PA [3].

Some data regarding the morphology and the dimen-

sions of the PA have already been published, but none have

used the techniques of geometric morphometrics. The

maximum width and height and sexual dimorphism of the

piriform aperture has yet been described on dry skulls [4,

12, 19]. The interalar width has been correlated with the

E. Moreddu (&) � L. Puymerail � P. Adalian

Unite d’Anthropologie Bio-Culturelle, Droit, Ethique et Sante

(ADES), UMR 7268, Universite d’Aix-Marseille-EFS-CNRS,

Boulevard Pierre-Dramard, 13344 Marseille, France

e-mail: eric.moreddu@etu.univ-amu.fr

L. Puymerail

e-mail: laurent.puymerail@univ-amu.fr

P. Adalian

e-mail: pascal.adalian@univ-amu.fr

E. Moreddu � J. Michel � M. Achache � P. Dessi

Service ORL et Chirurgie Cervicofaciale, AP-HM, CHU

La Timone, 264 rue St-Pierre, 13385 Marseille cedex, France

e-mail: justin.michel@ap-hm.fr

M. Achache

e-mail: michael.achache@ap-hm.fr

P. Dessi

e-mail: patrick.dessi@ap-hm.fr

123

Surg Radiol Anat

DOI 10.1007/s00276-013-1116-2

PA and its variations in function of the skin color in adult

male skulls [10]. The upper and lower width of the PA was

measured on 3D-CT reconstructions [11]. The shape of the

PA has been compared in Gorilla gorilla, Pan troglodytes,

and modern Homo sapiens by 2D Fourier analysis [17].

The nasal airflow has been correlated to external nasal

dimensions, but neither to the PA dimensions nor shape

[9].

From a clinical and surgical point of view, it is inter-

esting to perform in vivo PA measurements and shape

evaluation, with their sexual dimorphism. It is known that

surgical and traumatic modifications of this area lead to air

passage modifications, and we think that its morphological

variations may lead to functional variations.

In the following study, the question is to evaluate the

shape and the dimensions of the PA and their sexual

dimorphism on 3D-reconstructions of facial CT scans

performed on non-pathologic subjects using geometric

morphometrics [1, 14].

Materials and methods

Subjects

We collected 170 CT scan imaging of human adults aged

from 15 to 97 years (mean = 53 years), from La Timone

hospital, Marseille (France). These patients were selected

from non-traumatic patients, without any surgical treatment

on the nasal area, and had enhanced CT scans for non-

rhinologic problems: pulmonary embolism, stroke or

exploration of laryngeal carcinomas. 79 patients were

female, and 91 were male (Fig. 1). The examinations have

the following properties: pixel height and width between

312 and 529 lm, voxel depth between 300 and 1.25 mm.

The scans were anonymous when received by the authors,

with only sex and age data retained. Specific information

about ethnicity of the subjects was unavailable, but the

sample was taken as being representative of a typical

contemporary Mediterranean population.

3D reconstruction

As a basis for this imaging, computer tomography of head

and neck was used and was carried out in axial alignment.

Using the data of these examinations, a 3D reconstruction

of the face and skull could be generated with Avizo v.6.2

software package. To virtually isolate its cortical shell, and

reconstruct its entire volume, a semi-automatic threshold-

based segmentation with manual corrections was carried

out following the half-maximum height method [18]

(HMH) and the region of interest thresholding protocol [6]

and by taking repeated measurements on 20 different zones

of the virtual stack [5] by the Avizo v.6.2. (Visualization

Sciences Group Inc.) and ImageJ v.1.45 s packages.

Landmarks

We used eight homologous landmarks around the PA on

170 subjects (Fig. 2). Four were type-1 landmarks (anterior

nasal spine, rhinion, left and right naso-maxillary junc-

tions), and four were type-2 (left and right lateral piriform

points, left and right piriform bases). To position correctly

the lateral piriform points, we displayed the Frankfort plan,

translated it until the maximal width of the PA, and put the

landmarks at its intersection with the anterior part of the

surface.

49 10 13 13 10

15

51

12 78

24

1715

7

0

5

10

15

20

25

30

35

15-20 20-30 30-40 40-50 50-60 60-70 70-80 80+

Nu

mb

er

Age

Male

Female

5

21

17

21

37

2730

12

Fig. 1 Number of individuals for PA study by age group. Females

are represented in green and males in blue. Total number: 170

subjects

Fig. 2 Illustration of the eight PA points and four facial points

measured on 3D reconstruction of computed CT-scan. Numbers 1, 4,

5, 6 are anatomical PA landmarks; numbers 2, 3, 7, 8 are surface PA

constructed landmarks and numbers 9, 10, 11, 12 are facial

anatomical landmarks. 1 Nasal spine, 2 Right piriform base, 3 Right

lateral piriform point, 4 Right naso-maxillary junction, 5 Rhinion, 6Left naso-maxillary junction, 7 Left lateral piriform point, 8 Left

piriform base, 9 Right zygomaxillar point, 10 Left zygomaxillar point,

11 Prosthion, 12 Nasion

Surg Radiol Anat

123

We added four type-1 landmarks on the face: nasion,

prosthion, left and right zygomaxillary points. These

landmarks could be added on 156 subjects (70 female, 86

male) of the sample. The 14 missing subjects had a cut

facial CT scan.

Analyses

Geometric linear distances were calculated to determine

the PA horizontal width (PAW) between the two lateral

piriform points (landmarks 3 and 7) and its vertical length

(PAL) between the rhinion and the anterior nasal spine

(landmarks 1 and 5). The same method was used for the

width of the face between the zygomaxillary points

(landmarks 9 and 10) and its height between nasion and

prosthion (landmarks 11 and 12). We did not consider the

study of asymmetry of the PA, which was not the objective

of this study.

All the geometric morphometrics procedures and sta-

tistical analysis were carried out with MORPHOJ v.1.05a

and R v.2.15.0 software packages. The homologous land-

marks were converted to shape coordinates by generalized

least squares Procrustes superimposition method [7, 16].

Principal component analysis (PCA) of shape [1, 2, 13, 15]

was computed separately from the covariance matrix of the

Procrustes coordinates of facial and PA landmarks.

Analyses were also computed separately from the Pro-

crustes coordinates of facial and PA shape for identifying

similarities and differences by gender.

The analysis of sexual dimorphism was carried out with

R v.2.15.0 package. Multivariate analysis of variance

(MANOVA) by a Wilks test allowed us to determine the

significance of the shape and size differences between the

genders, and Student t test was performed on the linear

measurements. The shape space and the size and shape

space were depicted from the first principal component

(PC1) and second principal component (PC2), which

explain most of the part of the original variance.

A wireframe representation of landmarks configuration

according to PC1 was carried out by MORPHOJ v.1.05a.

Intra-operator and inter-operator reproducibility of land-

marks positioning were performed on 50 CT scans ran-

domly selected among the list of subjects, using the

random function in Microsoft EXCEL v.12.0 software.

Shape and centroıd size (CS) of the data sets were com-

pared by a Wilks test MANOVA carried out by MORPHOJ

v.1.05a.

Repeatability was performed ten times on a randomly

selected CT scan on ten different days, to avoid memori-

zation bias. The maximum, minimal, and mean distance

between the positions of each landmark were calculated.

Repeatability of the whole procedure (HMH, oblique slice

and landmark positioning) was analyzed by a Wilks test

ANOVA of shape and CS on ten randomly selected

subjects.

Results

Reproducibility and repeatability

The intra-operator and inter-operator reproducibility tests

on 50 individuals showed no statistically significant

Table 1 Mean dimensions of PA (in mm)

Female Male p value

(Student t test)

PAW

(mean ± SD)

24.00 ± 1.77 25.32 ± 1.86 \0.001

PAW range [18.91–29.25] [21.27–30.53]

PAL

(mean ± SD)

32.54 ± 2.70 36.35 ± 3.07 \0.001

PAL range [26.29–38.83] [24.99–43.25]

female male

2022

2426

2830

a

female male

2530

3540

b

Fig. 3 Boxplots of median PA dimensions in females and males: PAW on the left (a) and PAL on the right (b)

Surg Radiol Anat

123

difference. The repeatability test on an individual showed a

good repeatability with a maximal difference of 1.51 mm

for a landmark and a maximal mean difference of

0.59 mm. Repeatability of the whole procedure showed no

statistically significant difference.

Piriform aperture measurements

Mean PAW measure was 24.00 mm (±1.77) in females

and 25.32 mm (±1.86) in males. Mean PAL measure was

32.54 mm (±2.70) in females and 36.35 mm (±3.07) in

males (Table 1). Student t test performed between the

values calculated in both genders was statistically signifi-

cant for PAW and PAL.

Figure 3 represents the boxplots of median dimensions

of the PA: PAW and PAL were superior in males, with few

outliers.

Table 2 shows Pearson’s product-moment correlation

between PAW and PAL, which was not statistically sig-

nificant for both females and males (p [ 0.05). Figure 4 is

a representation of PAW and PAL correlation with the

linear regression lines.

Facial measurements

Mean measurements of the facial width and height were

superior in males as showed in the Table 3. Student t test

performed between the values calculated in both genders

was statistically significant for facial width and height.

Figure 5 shows the boxplots of these facial dimensions.

Table 4 shows Pearson’s product-moment correlation

between the measurements of face and PA. The heights are

correlated in the two genders, while the widths are corre-

lated only in males.

Piriform aperture shape analyses

The PCA of Procrustes analysis for the eight landmarks

surrounding the piriform aperture yielded 17 PCs.

The shape space depicted from the first two PCs, which

jointly explained 46.32 % of the original variance, is

showed on Fig. 6. Wilk’s test MANOVA showed a sig-

nificant difference between males and females (Table 5),

despite the fact the ellipses merge.

Figure 7 is a wireframe 2D visualization of the modifi-

cations accounted by PC1 on the consensus individual from

three points of view. Both visualizations show an almost

constant length of the PA, the shape modifications being

brought by its width.

The CS was significantly superior in males (46.57 mm)

than in females (43.11 mm), as expected by the linear

measurements previously calculated (Table 6). The box-

plots depicting the CS are showed on Fig. 8.

The size and shape space depicted from the first two

PCs, which jointly explained 47.98 % of the original var-

iance, is showed on Fig. 9. The ellipses surrounding 95 %

of the variability of each gender are visually and statisti-

cally different (Table 5), corresponding to the size differ-

ence described above.

The correlation coefficients between CS and PC1

(Table 7) were significant, with a high value in size space:

0.89 (p \ 0.001) and a low value in shape space: -0.25

(p \ 0.001).

Table 2 Pearson’s product-moment correlation between PAW and

PAL

Female Male

Correlation -0.064 -0.014

p value 0.57 0.89

25 30 35 40

2022

2426

2830

PAL

PAW

femalemale

Fig. 4 Plot of PAW and PAL correlation with linear regression lines.

Female linear regression line is shown by the solid line, male linear

regression line is shown by the dotted line

Table 3 Mean facial dimensions (in mm)

Female Male p value

(Student t test)

Facial width

(mean ± SD)

117.06 ± 6.27 126.19 ± 4.97 \0.001

Facial width

range

[95.08–133.11] [114.77–137.25]

Facial height

(mean ± SD)

67.07 ± 4.84 71.68 ± 5.17 \0.001

Facial height

range

[57.02–77.68] [52.63–89.23]

Surg Radiol Anat

123

Discussion

Design of the study

Our study was the first to depict the shape and size of the

PA using geometric morphometrics on 3D-CT recon-

structions. This method was reliable and reproducible, with

a good quality of description of the PA with eight land-

marks: adding more landmarks would not have brought

more information on the sexual dimorphism.

The main difficulty and source of variability was to

position the lateral PA landmarks, because of the lack of

bone sutures or structure allowing an undeniable posi-

tioning. To get round this problem, the reconstruction of a

plan parallel to Frankfort plan allowed us to position the

landmarks 3 and 7 with a good reproducibility, their

position corresponding to the inflexion of the lateral side of

the PA.

Studying soft tissues was not the objective of this study,

although their evaluation in the anatomy of the nose could

be interesting for functional purpose. Developing a reliable

methodology for measuring the thickness of nasal soft

tissues using MRI and CT scan is still a challenge.

Linear measurements

Our results about the width of the PA were in the same

range of variation than the previous studies in the literature

[4, 10–12, 19]. This observation validates the HMH 3D

reconstruction method over the PA; our values were close

to those that were acquired with classical morphometric

methods on dry skulls, and did not overexpress the

dimensions of the PA.

In the literature, there is no study showing an ethnical

difference on PA dimensions. The PAW calculated in our

female male

6070

8090 b

female male

100

110

120

130

a

Fig. 5 Boxplots of median facial dimensions in females and males: facial width on the left (a) and height on the right (b)

Table 4 Pearson’s moment-product correlation of facial and PA

dimensions

Width correlation p value Height correlation p value

Male 0.32 0.003 0.35 0.001

Female 0.23 0.059 0.33 0.004

−0.1 0.0 0.1 0.2 0.3

−0.

2−

0.1

0.0

0.1

PC1 (27.3 %)

PC

2 (1

9.02

%)

FemaleMale

Fig. 6 Scatterplot of the placement of individuals on PC1 and PC2 in

the shape space (PC analysis of the Procrustes shape coordinates

using PA landmarks). Ellipses are showing 95 % of variability for

females (solid line) and males (dotted line)

Table 5 MANOVA of the shape space and of the size and shape

space in females and males (PC scores, sex)

Test Pillai Approx F Df Den Df p

Shape space Wilks 0.13138 3.044 8 161 0.003

Size and

shape space

Wilks 0.417 19.431 6 163 \0.001

Surg Radiol Anat

123

study were close to the values obtained in the other studies,

including Brazilian [4], American [10], German [11], and

Corean [12] samples. Concerning the PAL, only Cantin

3 7

4

2

5

8

6

11

2

3

4

5

6

7

8

12

8

7 3

4 6

5

1

2

3

4 6

7

8

5

PC1 -0.15

PC1 +0.15

Frontal view Lateral view Upper view

5

4 6

3 7

82 1

3 7

42 5

1

86

Fig. 7 Wireframe

representation of landmarks

configuration according to PC1:

consensus shape is shown in

gray and modifications on PC1

are shown in black

Table 6 Centroıd sizes in females and males

Female Male p value (Student t test)

Centroıd size 43.11 46.57 \0.001

female male

4045

50

Fig. 8 Boxplot showing female and male centroıd sizes

−0.1 0.0 0.1 0.2 0.3

−0.

2−

0.1

0.0

0.1

0.2

PC1 (29.58 %)

PC

2 (1

8.4

%)

FemaleMale

Fig. 9 Scatterplot of the placement of individuals on PC1 and PC2 in

the size and shape space. Ellipses are showing 95 % of variability for

females (solid line) and males (dotted line)

Table 7 Pearson’s product-moment correlation between centroıd

size and PC1 in the shape space and in the size and shape space

Correlation coef p value

CS and PC1 Shape space -0.25 [-0.39, -0.11] \0.001

CS and PC1 Size and shape space 0.89 [-0.92, -0.85] \0.001

Surg Radiol Anat

123

Lopez and co-workers [4] had a longer PA in their

Brazilian sample (Table 8), but this sample was composed

of mixed ethnicities (30 skulls of individuals with white

skin, 30 with black skin, and 30 with brown skin), and there

was no statistically significant difference between the

ethnicities.

In all the studies, there was a true dimorphism of PA

dimensions. Males had a larger and longer PA, which

means that the size of PA was correlated to the absolute

size of the individuals. The translation of this difference in

our study using morphometric geometrics was a higher CS

for males.

We could have thought that the facial development, and

especially its horizontal growth, explained the variations of

PA dimensions, but the correlations between facial and PA

dimensions were weak (Table 4). Furthermore, the width

and length of PA were not correlated (Table 2). All these

results are leading us to the conclusion of a non-harmoni-

ous growth of the PA and the face.

Size and shape analysis

The difference of PA shape between female and male

subjects was statistically significant, but the visualization

of this difference was not informative. This difference was

amplified by taking into consideration the CS of the indi-

viduals, which was the main factor of dimorphism as

explained by the size and shape space (Fig. 9) and shape

space (Fig. 6) and the correlation coefficient values

(Table 7).

Furthermore, we tested a discriminant analysis, which

was not conclusive for the PA, which cannot be used as a

good gender predictor.

Conclusion

Our results showed the existence of a significant sexual

dimorphism of the PA, not only in terms of dimensions, but

also in terms of shape.

We can make a connection with the physiologic nasal

airflow, which has been evaluated to be significantly lower

in females than in males, with a good correlation and with

nasal external measurements [9].

These results encourage us to investigate the correlation

between PA morphology and airflow: for example, is a

narrower PA correlated with a lower airflow? Furthermore,

some functional nasal problems are induced by the airflow

regime which can be turbulent or laminar, and the link

between the PA shape and the airflow regime has to be

investigated.

A further study would associate functional examination

and imaging to answer these questions.

Conflict of interest The authors declare that they have no conflict

of interest.

References

1. Bookstein FL (1991) Morphometric tools for landmark data :

geometry and biology. Cambridge University Press, Cambridge

England, New York

2. Bookstein FL (1996) Biometrics, biomathematics and the mor-

phometric synthesis. Bull Math Biol 58(2):313–365

3. Burstein FD, Cohen SR (1995) Piriform aperture stenosis: a rare

cause of neonatal airway obstruction. Ann Plast Surg 34(1):56–58

4. Cantin LM, Suazo GIC, Zavando MDA, Smith RL (2009) Sexual

dimorphism determination by piriform aperture morphometric

analysis in Brazilian human skulls. Int J Morphol 27(2):327–331

5. Coleman MN, Colbert MW (2007) Technical note: CT thres-

holding protocols for taking measurements on three-dimensional

models. Am J Phys Anthropol 133(1):723–725

6. Fajardo RJ, Ryan TM, Kappelman J (2002) Assessing the accu-

racy of high-resolution X-ray computed tomography of primate

trabecular bone by comparisons with histological sections. Am J

Phys Anthropol 118(1):1–10

7. Gower JC (1975) Generalized procrustes analysis. Psychometrika

40(1):33–51

8. Gray H, Standring S, Ellis H, Berkovitz BKB (2005) Gray’s

anatomy : the anatomical basis of clinical practice, 39th edn.

Elsevier Churchill Livingstone, Edinburgh

9. Hall RL (2005) Energetics of nose and mouth breathing, body size,

body composition, and nose volume in young adult males and

females. Am J Hum Biol Off J Hum Biol Counc 17(3):321–330

10. Hoffman BE, McConathy DA, Coward M, Saddler L (1991)

Relationship between the piriform aperture and interalar nasal

widths in adult males. J Forensic Sci 36(4):1152–1161

11. Hommerich CP, Riegel A (2002) Measuring of the piriform

aperture in humans with 3D-SSD-CT-Reconstructions. Ann Anat

Anatomischer Anzeiger 184(5):455–459

Table 8 Comparison of

mean ± SD PA dimensions

with previous studies [4, 10–12,

19]

Width Length

Female Male Female Male

Cantin Lopez 25.27 ± 2.61 26.87 ± 4.80 47.53 ± 3.30 50.83 ± 2.83

Hoffman 25.45 ± 2.53

Hommerich 22.6 23.6

Hwang 25.4 ± 1.7 25.7 ± 1.7 28.0 ± 2.8 30.1 ± 2.6

Vitte 24.1 ± 1.9 30.7 ± 3.8

Present study 24.00 ± 1.77 25.32 ± 1.86 32.54 ± 2.70 36.35 ± 3.07

Surg Radiol Anat

123

12. Hwang T-S, Song J, Yoon H, Cho B-P, Kang H-S (2005) Mor-

phometry of the nasal bones and piriform apertures in Koreans.

Ann Anat Anatomischer Anzeiger 187(4):411–414

13. Marcus LF, Bello E, Garcıa-Valdecasas A (1993) Contributions

to morphometrics. Monografıas/Museo Nacional de Ciencias

Naturales, Museo Nacional de Ciencias Naturales, Consejo

Superior de Investigaciones Cientıficas, CSIC, vol 8. Madrid

14. Mitteroecker P, Gunz P (2009) Advances in Geometric Mor-

phometrics. Evol Biol 36(2):235–247

15. Mitteroecker P, Gunz P, Bernhard M, Schaefer K, Bookstein FL

(2004) Comparison of cranial ontogenetic trajectories among

great apes and humans. J Hum Evol 46(6):679–698

16. Rohlf FJ, Slice D (1990) Extensions of the procrustes method for

the optimal superimposition of landmarks. Syst Zool 39(1):40–59

17. Schmittbuhl M, Le Minor JM, Allenbach B, Schaaf A (1998)

Shape of the piriform aperture in Gorilla gorilla, Pan troglodytes,

and modern Homo sapiens: characterization and polymorphism

analysis. Am J Phys Anthropol 106(3):297–310

18. Spoor CF, Zonneveld FW, Macho GA (1993) Linear measurements

of cortical bone and dental enamel by computed tomography:

applications and problems. Am J Phys Anthropol 91(4):469–484

19. Vitte E, Guerin-Surville H, Baulac M, Cabrol C (1983) Mor-

phometric study of the piririform aperture. Bulletin de l’Associ-

ation des anatomistes 67(198):347–354

Surg Radiol Anat

123