Quantification of age-related changes in midsagittal facial profile using Fourier analysis: A longitudinal study on Japanese adult males

Abstract

Objectives

The aim of the present study was to use Fourier analysis to quantify and study age-related changes in midsagittal facial profile.

Materials and methods

Midsagittal facial profiles were extracted as lists of x and y coordinates from 125 pairs of 3D facial scans captured at an average of 10.5 years apart for adult Japanese males aged 23–52 years. These were categorized into three 10-year-long age groups. Files of x and y coordinates underwent Fourier analysis at 30 harmonic levels. Paired t-tests were used to determine statistical significance of differences across corresponding harmonic coefficients. Mean harmonic coefficients were used to construct mean pre and post ageing profiles for each age group for qualitative comparisons.

Results

Full detail of facial profile was described by the first 20 harmonics. With increasing age, there was a trend of longitudinal changes involving more midsagittal shape features with increased magnitudes. However, all changes were lower than 1 mm.

Conclusions

Fourier analysis is a useful morphometric approach to quantify age-related midsagittal facial changes. The small variations in the study groups prompt for testing Fourier analysis on the elderly and on other parasagittal and transverse facial features.

Introduction

The quantitative information held in the human face is a valuable indicator of identity in forensic science. It is particularly useful in the identification of the living such as criminals and missing persons or the identification of the deceased such as victims of mass disasters. The quantitative analysis of age-related changes in the face can aid with person identification.

Pre-existing photographic records of the face are necessary sources for the identification process. However, recent photographic reference records captured at the time of investigation are not always available. In such situations, the investigator is challenged by predicting how the face of a person under investigation would look like after some period of time. With the recent development of 3D imaging technology, there have been successful attempts to quantify and predict changes in facial topography induced by growth [1] and ageing [2]. Although facial features in adulthood are generally more stabilised when compared to the changes taking place during craniofacial growth, it has been reported that facial topography may be subject to some skeletal modifications brought about by ageing [[3], [4], [5], [6]] and obesity [7].

Age-related changes in facial morphology are attributed to changes in the facial soft tissues and residual remodelling in the underlying facial skeleton. It is often supposed that body growth is not noticeable beyond the critical age of 18–19 years [8]. However, studies using 3D CT scans of people from young adulthood to old ages revealed age-related angular changes in the midfacial skeleton [[3], [4], [5], [6]]. Although it was reported that some residual facial growth especially in the lower third of the face still occurs in young adult males and that facial growth is almost completed in females after the age of 15 years [9], age-related changes later than 25 years of age were found more intense in females [3,5]. Generally, age-related soft tissue changes are due to loss of volume and skin atrophy [10,11]. Age-related changes can also follow, to a lesser extent, underlying skeletal changes [12,13]. Changes in facial topography can also be attributed to non-ageing factors such as obesity [7].

The process of facial identification is often challenged further by the lack of direct frontal views, e.g. where surveillance footage only shows a lateral view of the face. Moreover, the only facial identity records available at many dental practices are lateral cephalograms which only show the soft-tissue profile of the face. Therefore, analysis of facial profile can be also important to the identification process.

Various methods exist for quantifying the facial profile. While direct anthropometric parameters such as angles and measurements were used in some studies [[14], [15], [16]], others used more advanced mathematical methods such as Procrustes superimposition, principal component analysis [17] and Fourier analysis [13,[18], [19], [20], [21], [22], [23]].

Fourier shape analysis decomposes contours or curvatures into harmonic functions defined by Fourier descriptors [24]. Fourier shape analysis is a mathematical method based on converting a curved outline such as the midsagittal profile of the face into an infinite successive sine and cosine functions amenable to quantitative statistical description, comparison and analyses [25]. Fourier analysis has been used to quantify human facial profiles [[19], [20], [21], [22], [23],26,27]. Most of the articles that employed Fourier analysis in studying longitudinal changes in facial profile focused on changes occurring during growth and development [19,26,28]. Studies using Fourier analysis to quantify age-related changed in adulthood have been very limited [13,20]. Ferrario et al. [20] used Fourier harmonic analysis to quantify and longitudinally evaluate the facial profiles of 14 adult subjects after 10 years. They computed the areas enclosed in each facial outlines and measured the difference in the area between pre and post ageing profiles. Ferrario et al. [20] reported significant age-related changes in facial profile in men and women. On the other hand, Kapur et al. [13] used lateral cephalograms of 77 subjects to extract the facial profiles. They then applied Fourier analysis to measure age-related longitudinal changes over a nine-year interval. They reported the greatest change between the upper and lower lips and the least change at the tip of the nose.

The limited literature quantifying age-related changes in facial profile through Fourier analysis has been limited further to small numbers of subjects uncategorised into age groups, the aim of the present study was to employ Fourier analysis to quantify age-related changes in facial profiles across three 10-year-long age groups.