AGE ESTIMATION ON 2D IMAGES: APPLICATION OF THE METRICAL FACIAL APPROACH FOR THE ASCERTAINMENT OF AGE IN CASES OF SUSPECTED JUVENILE PORNOGRAPHY

Age estimation is one of the most difficult and sensitive procedure in forensic pathology, and is characterized by a long tradition and a number of fields of application: the most common cases of age estimation in the forensic scenario deal with dead people, and in detail with personal identification; however, in the last years, age estimation has begun to include also other forensic cases, especially within the so-called Clinical Forensic Medicine, which is the branch of forensic pathology dealing with the living, and consequent evaluation of clinical data for judicial purposes. The procedure of age estimation therefore is now applied to the living for the ascertainment of imputability before trial, or for the correct assignation of scholar classes of adopted children. In recent times, also the age estimation of the living adult, which in the most challenging and difficult field of application, has begun to be performed in order to verify the age of retirement, usually in old immigrants. One of the most interesting phenomena recorded during this long evolution is the constant and progressive modification of methods of age estimation according to the specific field of application: some methods cannot be applied in specific cases, and this lack of tools has led to the evolution of new procedures, or the amelioration of the existing ones. However, the more specific the field of application, the fewer are the methods which can be applied. The extreme evolution of this phenomenon concerns the most recent field of application of age estimation in the living which deals with 2D images: the ascertainment of age in case of photos may be judicially important, since most Countries state the crime of juvenile pronography, which concerns all the images reproducing minors in a pornographic context. In these cases the forensic anthropologist may be requested to provide an age estimation of the persons in the photo: this is the final evolution of any procedure of age estimation, which has passed in the last year from the dead subject (where any investigative method is substantially allowed), to the living person (where only few methods can be applied) up to the image of the living person itself. One can clearly consider the pitfalls included in such as operation, since the forensic anthropologist or pathologist is usually requested to give an indication concerning the biological profile of a real person on his photos. If in case of the living age estimation has tried to change its methods, now with photos the forensic anthropologist faces a new challenge which consists not only in the development of new tools, but also in verifying if such analyses can be performed also in photos. The tasks are therefore two, and both of them difficult to achieve: the first one consists in finding biological information related with age, and the second in ascertaining if such information is also verifiable in photo. As one can imagine, the procedure of age estimation in cases of images faces new challenges, with new questions which are still waiting for an answer. This study aims at exposing the results of different investigations performed during the PhD course, aimed at verifying the relation between facial measurements and age, in vivo and in photo: the line of research followed a project aiming first at ascertaining the reliability of linear measurements in photos, both for subjects aged under and above 18 year threshold, and then at verifying the chance of extrapolating new biological parameters from the face useful for age estimation. The first steps will consist in finding a standardization of facial metrical assessment by an analysis of reliability of collocation of facial landmarks. After this phase, the study will attempt at verifying the reliability of linear measurements for age estimation in photos in subjects aged under 18 years. A similar experiment will be applied as well to young adults aged over 18 years, after a preliminary study of in vivo measurements in order to verify the correlation of such parameters with age. In conclusion, the study will attempt at finding new biological geometrical parameters (in detail, facial surfaces) for age estimation, in vivo and in photo. This path led to interesting results which may be sumed up in the following points: 1) the study performed on facial landmarks showed that not all the points are reliable for a standardized procedure of facial assessment: in detail, the anatomical landmarks are the most trustable, probably because are defined by anatomical structures which can be easily detected, and are consequently less subjective. More interestingly, such landmarks show the same reliability in photos taken from subjects with different ages, and this means that anatomical points are crucial also for the issue of age estimation from 2D images; 2) the second study pointed out that ratios between linear measurements show a correlation with age also in photos, and therefore may be used for age estimation, although with a high error range, and only taking into consideration the main age thresholds; from this point of view, such parameters may provide a method for verifying if a subject is close to one of the chosen age limits, but data are too limited to provide adequate regression formulae useful to put in relation age and measurements. However, these results represent the first step for the development of an age estimation method useful for reconstructing age with higher precision; 3) the third study confirmed that some linear measurements show a correlation with age which may be promising for the development of regression formulae useful for age estimation in photos; however, the error range is estimated in 3-4 years, which means that age may be reconstructed in an interval of several years: however, this error range matches with difficulties to the actual forensic purposes, where an age estimation with a more limited error is often requested; the need for the search for new facial parameters is therefore even more evident; 4) the fourth study attempted at verifying the modifications of facial parameters in juvenile adults and subjects in transition phase, and pointed out that ear and mouth show a constant increase also after 18 years; from a general point of view, the parameters which are related with age in vivo should show the same relation also in photos, and therefore may be useful for age estimation; 5) the fifth study focused on verifying the applicability of ear characteristics to age estimation in photos: results show that neither the ratios between linear measurements nor the areas show a correlation with age; these results therefore confirm that photograph is an independent manner of acquisition of reality, with its own rules and relations between facial parameters and age, which not necessarily correspond to those observed in vivo. These results mean that the relation of facial parameters should be verified directly in photos, and underline the need for determining new facial measurements useful for forensic purposes; 6) the last study aimed at verifying the relation of a novel facial parameter (in detail, facial surfaces) with age; results show that the correlation are lower than those showed by linear measurements: the analysis performed in photos confirmed the independence of measurements taken in photos and in vivo (actually, on the digital 3D models of the face), since some parameters are related with age in vivo, not in photos, and vice versa. Anyway, facial surfaces are too variable and are influenced by too many environmental and individual factors to be usable for the development of an age estimation method. In conclusion, facial metrical assessment actually cannot be considered a scientifically valid method for forensic purposes: in detail, linear measurements have been widely explored by past literature which showed their constant modification with age. However, this information finds a limited importance in forensic practice, and is affected by a relevant limit in 2D images, which deals with the distorsion of measurements in two dimensions. The introduction of the new 3D acquisition systems now allows the operators to perform a more detailed analysis of face by the measurements of surfaces and volumes; however, literature concerning this topic is still at the beginning, and mainly deals with geometrical measurements which are more standardized, but also more influenced by other variability factors than age, such as the weight, etc. From this point of view, the analysis of surfaces and volumes adds precious information to the morphology of face, and it will be one of the main field of research of the modern anatomy. From the anatomical point of view, the experimental project pointed out relevant suggestions for different issues: in detail, the chromatic analysis of facial modifications may provide a new useful tool for standardizing and quantifiying variables such as increasing or decreasing of weight, the influence of facial expressions, the resemblance of children with their parents. The application of modern 3D image acquisition system may therefore radically improve the anatomical study of faces, adding new information potentially useful also for clinical purposes.