The development and testing of Thai facial soft tissue thickness data in three-dimensional computerized forensic facial reconstruction

Abstract

Forensic facial reconstruction is a useful tool to assist the public in recognizing human remains, leading to positive forensic investigation outcomes. To reproduce a virtual face, facial soft tissue thickness is one of the major guidelines to reach the accuracy and reliability for three-dimensional computerized facial reconstruction, a method that is making a significant contribution to improving forensic investigation and identification. This study aimed to develop a facial soft tissue thickness dataset for a Thai population, and test its reliability in the context of facial reconstruction. Three-dimensional facial reconstruction was conducted on four skulls (2 males and 2 females, with ages ranging between 51 to 60 years). Two main tools of three-dimensional computer animation and modeling software—Blender and Autodesk Maya—were used to rebuild the three-dimensional virtual face. The three-dimensional coordinate (x, y, z) cutaneous landmarks on the mesh templates were aligned homologous to the facial soft tissue thickness markers on the three-dimensional skull model. The final three-dimensional virtual face was compared to the target frontal photograph using face pool comparison. Four three-dimensional virtual faces were matched at low to moderate levels, ranging from 30% to 70% accuracy. These results demonstrate that the facial soft tissue thickness database of a Thai population applied in this study could be useful for three-dimensional computerized facial reconstruction purposes.

Keywords

Forensic science facial reconstruction three-dimensional imaging computer-assisted Thai

Introduction

Forensic facial reconstruction is a useful method to use with human remains in complex circumstances, such as where there is advanced decomposition, where ante-mortem information is unclear, and when there is a lack of proper evidence regarding forensic circumstances.^1–3 It can be used as an alternative forensic method, and also holds value within forensic investigations. Its purpose is to recreate a realistic ante-mortem face from skull remains, to allow recognition by the family or general public, thereby assisting investigative processes.^4,5

Traditionally, three-dimensional (3-D) manual facial reconstruction, a scientific art method, was first introduced in the late 1800s and classified into three main techniques—the Russian, American, and Manchester techniques.^1,6 However, these techniques are highly time-consuming, requiring anthropological and anatomical training. Recently, however, 3-D computerized facial reconstruction has been introduced, and it overcomes shortfalls in traditional manual methods.^7–10 Technology enables more flexibility in how we visualize the different components of a reconstruction (e.g. we can change the opacity of the facial soft tissue thickness—FSTT—pegs, muscles or skin, without altering the form of the reconstruction model). We can establish databanks of pegs, muscles and facial features that can be superimposed into a reconstruction and modified accordingly, rather than model these details from the beginning every time.¹¹ 3-D computerized facial reconstruction is normally based on a semi-automated approach to reconstruction (e.g. 3-D modeling software such as 3DStudio Max,^4,5 Freeform,^11–13 Blender,¹⁴ or Zbrush¹⁵⁾ that still depends on traditional manual methods and requires good knowledge of anatomy and anthropology, or an automated approach to reconstruction that requires sophisticated computer-based techniques and craniofacial information to generates facial reconstruction (using software specifically programmed to run reconstruction algorithms and to generate a face from a skull model).

One of the most commonly used piece of cranio-facial information is the FSTT database.⁸ FSTT values indicate the mean depth of tissues across various craniofacial landmarks, and support facial reconstruction by describing how, on average, the soft tissue face fits over the skull. The application of FSTT data is a major contributor to the accuracy and reliability of a facial reconstruction.¹² Demographic variances have previously been identified and have resulted in the development of FSTT datasets that take into account different populations,^16–24 sex,^25–27 age²⁸ and even body mass index (BMI) variables.^25–28 Multiple studies testing the validity of 3-D facial reconstruction techniques have been conducted.^4,5,12–14 In a study by Fernandes et al.,⁴ it was found that the recognition rate of 3-D facial reconstructions, using population-specific FSTT data, out-performed those that used an international average. With this in mind, we acknowledge that no existing FSTT data or computerized facial reconstruction studies exist using Thai data. Therefore, in this study, we aimed to develop a Thai-specific FSTT dataset and test its reliability when applied to 3-D facial reconstruction.

Methods

Ethical considerations

This study received ethical approval from the Research Ethics Committee of the Faculty of Medicine at Chiang Mai University (clearance no. EXEMPTION-6566/2019).

Facial soft tissue thickness database

The average of FSTTs (10 midlines and 17 lateral landmarks) were collected from 100 fresh Thai cadavers (50 males and 50 females), using a needle puncture technique. The sample used in this study ranged in age between 51–95 years, with a mean age of 72.2 years (SD = 12.2) for males and 69.5 years (SD = 11.9) for females. The sample was preserved at a low temperature (1–2 °C) and the data were collected within 24–72 h. following donation, without embalming. All samples were positioned in a supine position and set the cadaveric head with Frankfort horizontal plane position. Samples exhibiting asymmetry of the face, facial trauma, fracture of the head, craniofacial surgery, and extreme obesity were excluded from the study. In this study, the FSTT was measured at 27 cutaneous landmarks (Table 1 and Figure 1), most of which were selected following earlier publications.^29–31

Figure 1.

A diagrammatic reference of cutaneous landmarks corresponding to skeletal landmarks considered in the study.

Table 1.

The definition of the twenty-seven cutaneous landmarks corresponding to skeletal landmarks considered in FSTT measurement and placement of the FSTT markers on the 3-D skull model.

No.	Cutaneous landmarks		Skeletal landmarks
No.	Landmarks	Definition	Landmarks	Definition
	Midline
1	Glabella	The most anterior midline point on the forehead between the superciliary arches. Define the landmark by carefully palpation.	Glabella	The most anterior midline point or the most anterior projecting point of the frontal bone, on the brow ridge, in between the superciliary arches.
2	Sellion	The midline point between eyes where deepest midline point of nasofrontal angle.	Nasion	The midline point of the nasofrontal angle between eyes where intersection of the nasofrontal sutures.
3	Rhinion	The most anterior midline point which most prominent connection point between nasal bone and nasal cartilage. Define the landmark by carefully palpation.	Rhinion	The most inferior midline point on the internasal suture.
4	Subnasale	The most inferior midline point of the nose and not certainly located at the tip of the nasal.	Subspinale	The most inferior midline point of the nasal aperture bellow base of the nasal spine.
5	Mid-philtrum	The midline and midway point between Subnasale and Labiale superius.	Mid-philtrum	The midline and midway point between Subspinale and Prosthion.
6	Labiale superius	The most anterior midline point of vermillion of the upper lip.	Prosthion	The most anterior midline point of the central upper incisors.
7	Labiale inferius	The most anterior midline point of vermillion of the lower lip.	Infradentale	The most anterior midline point of the central lower incisors.
8	Supramentale	The deepest midline point of the chin fold below the Labiale inferius.	Supramentale	The deepest midline point of the mentolabial sulcus and superior to mental eminence.
9	Pogonion	The most anterior middle point of the chin where the prominent of the mandible. Define the landmark by carefully palpation.	Pogonion	The most anterior prominence middle point of the mandible.
10	Menton	The most inferior midline point of the chin. Define the landmark by carefully palpation.	Menton	The most inferior midline point of the mandible.
	Lateral
11	Mid-supraorbital	The most anterior middle point of the supraorbital margin, at a line that vertically bisects of orbit. Define the landmark by using a ruler to measure the half-distance. between Endocanthion and Ecthocanthion and create a perpendicular line up form that midway point to the supraorbital margin.	Mid-supraorbital	The most anterior middle point of the supraorbital margin, at a line that vertically bisects of orbit.
12	Mid-infraorbital	The most anterior middle point of the infraorbital margin, at a line that vertically bisects of orbit. Define the landmark by using a ruler to measure the half-distance between Endocanthion and Ecthocanthion and create a perpendicular line down to the infraorbital margin.	Mid-infraorbital	The most anterior middle point of the infraorbital margin, at a line that vertically bisects of orbit.
13	Endocanthion	The medial point located at the inner commissure of inner eye fissure.	Maxillofrontale	The point on fronto-maxillary suture intersect with the anterior lacrimal crest
14	Ecthocanthion	The lateral point located at the outer commissure of outer eye fissure.	Ectoconchion	Lateral point of the orbital margin opposed to the Maxillofrontale.
15	Inferior malar	The most prominent point of the maxilla bone, at a line that vertically to Infraorbital. Define the landmarks by carefully palpation.	Inferior malar	The most prominent point of the maxilla bone, at a line that vertically to Infraorbital just under the zygomatic process
16	Zygion	The most curvature point of the zygomatic arch. Define the landmarks by carefully palpation.	Zygion	The most curvature point of the zygomatic arch
17	Lateral orbit	The anterolateral point vertically to the lateral orbital margin on the zygomatic bone.	Lateral orbit	The anterolateral point vertically to the lateral orbital margin on the center of the zygomatic bone
18	Alar base	The base of the nostril where the midway point between Subnasale and Alar curvature.	Alar base	The base of the nostril where the midway point between Subspinale and Alar curvature.
19	Alar curvature	The most posterolateral point of the alar of nose.	Alar curvature	The most curvature point of alar aperture, approximately 5 mm lateral to the most lateral point on the nasal aperture.
20	Supra canine	The point on superior alveolar ridge of the upper canine.	Supra canine	The point on the superior alveolar ridge of the upper canine.
21	Infra canine	The point on inferior alveolar ridge of the lower canine.	Infra canine	The point on the inferior alveolar ridge of the lower canine.
22	Supra M2	The point on superior alveolar ridge of the second upper molar.	Ectomolare	The point on the superior alveolar ridge of the second upper molar
23	Infra M2	The point on inferior alveolar ridge of the second lower molar.	Inframolare	The point on the inferior alveolar ridge of the second lower molar.
24	Gonion	The most prominent point of the angle of the mandible. Define the landmark by carefully palpation.	Gonion	The most prominent point of the angle of the mandible.
25	Occlusal line	The point of the anterior portion of the mandibular ramus in a place of dental occlusion or in a horizontal level of the lateral most point at the angle of the lips. Define the landmarks by carefully palpation.	Occlusal line	The point of the anterior portion of the mandibular ramus in a place of dental occlusion.
26	Mid-mandibular border	The point on the border of the mandible where the midway point between Gonion and Pogonion.	Mid-mandibular border	The point on the border of the mandible where the midway point between Gonion and Pogonion.
27	Mid-ramus	The point at the center of the mandibular ramus where the midway point between Occlusal line and outer border of mandible.	Mid-ramus	The point at the center of the mandibular ramus where the midway point between the Occlusal line and outer border of the mandible.

Twenty samples (10 males and 10 females) were randomly selected and re-measured two hours following the first assessment by the same investigator. According to the suggestion obtained in Stephan et al.³² and Meikle and Stephan,³³ the intra-observer error rates were examined by using the technical error measurement (TEM), relative technical error measurement (rTEM), and coefficient reliability (R), respectively.

Sample collection and preparation for facial reconstruction

The sample of this study included 4 skulls of 2 males and 2 females, with ages ranging between 51 to 60 years (Subject 1: Male; 51 years, Subject 2: Male; 54 years, Subject 3: Female; 51 years, and Subject 4: Female; 54 years). Four skulls with their ante-mortem frontal photographs were randomly collected from the Forensic Osteology Research Center (FORC), Faculty of Medicine, Chiang Mai University. All samples were dry skulls from bodies that had been donated for educational and research purposes.

The skull rearticulated with mandible was set up on an adjustable stand in an upright position. The 3-D skull image was obtained by using a DentiiScan cone-beam computed tomography (CBCT) machine^34,35 (NSTDA, Pathum Thani, Thailand) with 90 kV, 6 mA, a voxel size of 0.4 mm, and field of view of 130 mm × 160 mm. The 3-D skull image was saved as digital imaging and communications in medicine file of CBCT data and converted to stereolithographic file by using 3-D slicer version 4.10.2.³⁶

3-D Computerized facial reconstruction

Our study used two 3-D computer animation and modeling software as follows: Blender version 4.80³⁷ and Autodesk Maya³⁸ to reconstruct the 3-D virtual face. Additionally, the 3-D facial reconstruction was performed as a blind study—i.e., the researcher could not see the original frontal standardized photographs before reconstructing the 3-D virtual face.

The overview of the 3-D computerized facial reconstruction is presented in Figure 2 by following these steps. The 3-D skull was orientated in Frankfort horizontal position by setting the posterior nasal spine to the center of rotation. After the skull position was corrected, the 3-D skull model defined 27 skeletal landmarks (10 midlines and 17 lateral landmarks) onto the skull surface (see Figure 1 and Table 1 for detailed description of skeletal landmarks). The FSTT markers, cylindrical dowels, were placed onto the skull model that corresponded to those of anatomical landmarks. Next, the facial features, including eyeballs position, the tip of nose location, mouth height and width, were determined based on previous 3-D computerized facial reconstruction studies.^12,14 All of these steps were performed using Blender software³⁷ (Figure 2(A) and (B)).

Figure 2.

3-D Computerized facial reconstruction processes in lateral and anterior view (left column and right column, respectively). (A and B) showed 27 FSTT markers, cylindrical dowels, speared on the anatomical landmark of the 3-D skull model; (C and D) showed the face mesh template aligned onto the 3-D skull model at the level of FSTTs; (E and F) showed final 3-D facial reconstruction.

To reconstruct the virtual face (Figure 2(C)–(F)), the 3-D Asian face template was selected from open source, MakeHuman1 1.1.1.³⁹ First, The face mesh template was modified to fit the skull and aligned with ten cutaneous landmarks as follows: glabella (g’)–glabella (g), sellion (se’)–nasion (n), labiale superius (ls’)–prosthion (pr), labiale inferius (li’)–infradentale (id), pogonion (pg’)–pogonion (pg), menton (me’)–menton (me), mid-supraorbital (mso’)– mid-supraorbital (mso), mid-infraorbital (mio’)–mid-infraorbital (mio), alare curvature (ac’)–alare curvature (ac), and zygion (zy’)–zygion (zy). Next, the soft selection components tool in Autodesk Maya⁴⁰ was used to remodel the coordinates vertex of the imported face model to fit the FSTT markers on the skull. When selecting primary vertices for modification using this tool, a surrounding falloff area is selected and influenced accordingly, making the process of remodeling and sculpting more organic. After this, the area of the mouth, cheeks, and jaw was adjusted to the FSTT marker levels, respectively. The final 3-D facial reconstruction was imported as an object (OBJ) file format.

Comparison of the final 3-D facial reconstruction to the target photograph

The 3-D facial reconstructions underwent recognition testing via face pool comparison. Each reconstruction was therefore matched with a possible ante-mortem face, taken from a standardized frontal profile face pool array comprising one target photograph and 4 foil photographs. The percentage of correct reconstruction to target matches, consequently, describes the accuracy of the reconstruction. Twenty assessors that are inexperienced in face comparison techniques were involved in the face pool testing, including graduate students and staff from the Department of Anatomy, Faculty of Medicine, Chiang Mai University.

A descriptive analysis of the quantitative data was calculated using Microsoft Office Excel 2016.

Results

The average FSTT measurements collected using a Thai population sample are summarized in Table 2. Overall, the TEM values for intra-observer error are low. The rTEM values range from 2.12% to 8.10%. All measurement variables have R-value greater than 0.90 (Table 3).

Table 2.

Averages of FSTT in millimeters for a Thai population.

Landmark	Male		Female
Landmark	Mean	SD	Mean	SD
Midline
Glabella	3.37	0.63	3.46	0.66
Sellion	2.93	0.70	2.88	0.72
Rhinion	2.29	0.47	2.21	0.60
Subnasale	4.90	0.94	4.70	0.95
Mid-philtrum	4.38	0.91	4.10	0.89
Labiale superius	4.23	0.81	3.90	0.93
Labiale inferius	4.58	0.78	4.53	1.15
Supramentale	5.14	1.04	5.28	1.19
Pogonion	5.10	1.42	5.37	1.54
Menton	3.49	0.83	3.66	1.23
Lateral
Mid-supraorbital	3.32	0.66	3.72	0.75
Mid-infraorbital	2.76	0.88	3.00	1.05
Endocanthion	3.03	0.64	2.97	0.68
Ecthocanthion	2.30	0.57	2.62	0.85
Inferior malar	4.97	1.12	5.19	1.37
Zygion	3.98	1.46	4.45	1.36
Lateral orbit	5.55	1.83	6.57	2.08
Alar base	5.07	0.88	4.85	0.84
Alar curvature	4.75	0.87	4.64	0.86
Supra canine	3.65	0.67	3.59	0.85
Infra canine	3.81	0.91	3.64	0.79
Supra M2	5.27	1.05	5.01	1.23
Infra M2	5.04	1.12	5.11	1.04
Gonion	5.74	2.52	6.39	2.26
Occlusal line	5.63	2.09	6.94	2.37
Mid-mandibular border	3.44	1.22	3.76	1.11
Mid-ramus	10.89	4.02	11.42	3.42

Table 3.

Statistical analysis of TEM, rTEM, and R values for intra-observer error.

Landmark	TEM (mm)	rTEM (%)	R
Midline
Glabella	0.14	3.84	0.96
Sellion	0.17	5.36	0.94
Rhinion	0.13	5.48	0.96
Subnasale	0.24	5.17	0.91
Mid-philtrum	0.22	4.95	0.95
Labiale superius	0.22	5.00	0.93
Labiale inferius	0.20	3.94	0.95
Supramentale	0.12	2.12	0.99
Pogonion	0.20	3.51	0.98
Menton	0.18	4.82	0.94
Lateral
Mid-supraorbital	0.19	5.04	0.93
Mid-infraorbital	0.23	6.63	0.94
Endocanthion	0.17	5.33	0.91
Ecthocanthion	0.14	5.48	0.95
Inferior malar	0.24	4.33	0.97
Zygion	0.34	8.10	0.93
Lateral orbit	0.24	3.51	0.99
Alar base	0.16	3.17	0.97
Alar curvature	0.18	3.58	0.95
Supra canine	0.21	5.42	0.91
Infra canine	0.22	5.77	0.91
Supra M2	0.14	2.73	0.91
Infra M2	0.13	2.41	0.97
Gonion	0.25	3.75	0.98
Occlusal line	0.21	3.22	0.99
Mid-mandibular border	0.26	7.19	0.93
Mid-ramus	0.35	2.84	0.99

The final 3-D facial reconstruction based on the FSTT database of a Thai population obtained in this study is described in Figure 3. By applying the FSTT database for a Thai population, four final 3-D facial reconstructions (subjects 1–4) were correctly matched by assessors based on face pool comparison assessment with 70%, 30%, 35%, and 40% of accuracy in each subject, respectively.

Figure 3.

Comparison between 3-D facial reconstruction and the target frontal photograph; Row A: The target frontal photograph; Row B: The 3-D facial reconstruction; Row C: superimposition between 3-D virtual face and the actual target frontal photograph.

Discussion and conclusion

The accuracy of 3-D computerized facial reconstruction

The result of this study showed low to moderate (30%–70%) level for examining the use of the FSTT database of the Thai populationfor 3-D facial reconstruction. When comparing the results of 3-D computerized facial reconstruction to previous studies,^4,5,12–14 the accuracy rates in this present study are consistent with, for instance, Fernandes et al.,^4,5 who examined 3-D facial reconstruction using a Brazilian sample. When they compared 3-D facial reconstruction to the image of the target individual and other subjects based on face pool comparison, the database proposed for the Brazilian population provided a correct recognition rate ranging from 20–41%. Similarly, Miranda et al.¹⁴ tested 3-D facial reconstructions using four Brazilian adult subjects, based on a specific population dataset. The result showed that the 3-D reconstruction model, using free opensource software programs, offered a good accuracy rate ranged from 63% to 74% when compared to 3-D target models. Lee et al.¹² evaluated the applicability of the 3-D computerized facial reconstruction using skull models from three living Korean adult subjects. The historic average of the FSTT database of the Korean population, facial anatomy, and musculature was used to remodel the virtual face via 3-D modeling software. They reported that 3-D morphometric surface comparison between 3-D facial reconstruction and 3-D target surfaces showed a good level of success (54%–77%). Lee et al.¹³ also carried out 3-D facial reconstruction by updating average FSTT data of living Koreans and reported that the accuracy of the facial reconstruction was much higher (79%–87%) than in the previous study.

Face pool testing remains valuable, with a simple and quick assessment that assesses the frequency with which a target individual can be positively matched or identified to facial reconstruction. However, this test is solely dependent on unfamiliar face recognition. Moreover, the assessors included in the study have not been trained in facial comparison before. This could, in part, account for the weak face matching results obtained in our study, and in previous studies.^41–43

Facial soft tissue thickness dataset

According to the studies mentioned above,^12–14 facial areas, particularly in the lower zones of the face, nose, and cheek, generally displayed relatively low to high error deviations to the target faces. When the 3-D facial reconstruction was aligned to the target in our study (Figure 3; Row C), differences existed between the two images in those areas. This might be possibly explained due to faces and skulls being a variety of complex structures with global shapes and local detail.² Moreover, other studies^35,44,45 revealed that estimation of nose morphology is a challenging part of the reconstruction process because this facial component has been found to be sexually dimorphic, and to modify between ages. Another area which has been found to be highly inconsistent among different individual is the lower face due to different body weight and facial fat changes related to age.^28,46,47 In our study, the measurement was based on the needle puncture technique. This method holds various advantages, for instance, it requires rather simple and inexpensive instruments, and there is no concern over ionizing radiation exposure; however, it is slightly more difficult to palpate non-prominent landmarks, making some R-values of intra-observer errors less than 0.95, and the gravitational effects during measurement in supine positions might raise limitations. These factors may influence the final facial appearance; therefore an inaccuracy that occurred in this study might have resulted from them. An existing challenge with the current FSTT dataset is that it represents a mature population. It suggests that not only would increasing sample size help, but the future sample would ideally represent a broader range of age groups. Given the above highlighted and key limitations in samples and methodology, data collection from living samples is likely to offer more reliable results. The modalities used in data collection for living subjects such as CBCT or ultrasound are highly recommended. The advantages of these medical imaging techniques are that they enable images to be obtained with subjects in an upright position, thus accurately representing the face in a vertical position and reducing measurement distortion due to gravity-induced soft tissue displacement. Moreover, the results suggests that the total weighted FSTT means data or the tallied FSTTs (T-Table) established by Stephan⁴⁸ appear to be an alternative option to reduce the limitations on the impact of FSTT data noise, problems associated with sample sizes, and measuring errors. Therefore, further research should consider the pooled data from T-Table, which would be an ideal overarching solution for FSTT estimation, as well as improving facial reconstruction performance. It would be better to test how effective population-specific data compares with the general total weighted means. For example, this comparison has been performed in a South African FSTT study looking specifically at the mouth—validation testing identified that a population specific approach offered improved results to the total weighted means, when estimating actual soft tissue depth measurements on a hold-out sample.⁴⁹

Suggestion and path forward

In our experiments, a wide range of accuracy (30–70%) might also be generated by the 3-D computer processes. This study only based on a few distinct landmarks on mesh template and 3-D skull model yields the inherent defect of providing those discontinuous thickness values, for example, lack of thickness markers between the zygion and ramus as well as between other markers at the mandibular and buccal regions. Therefore, areas between these distinct landmarks need to be interpolated by for example, increasing dense soft tissue maps⁵⁰ and rigging the mesh polygon based on the facial muscles model for a more realistic outcome. Moreover, the 3-D virtual face in this study was produced by a single template that did not represent the average Thai face. It is a fact that the quality of reconstruction depends on an appropriate template. Future studies should acquire more information on facial morphology as well as the prediction of each facial feature (nose, eyes, mouth, ears), based on craniofacial information from both the skull and the face extracted from a Thai population. Consequently, the confidence level of accuracy and reliability for the 3-D facial reconstruction could be increased, enhancing correct matches, and eliminating the incorrect matches for the recognition test

The advantages of the 3-D computerized facial reconstruction presented in this study are that the 3-D facial template can be shown in an x-ray mode transparency and rendered in a mesh, making it easier to align mesh template homologous to reference FSTT cylinder markers.^12,13 The 3-D mesh of the face template can then be animated to display a moving face, and the polygon meshes can be adjusted for a symmetrical purpose, all of which are not available in a sculpting method. Moreover, the time consumed was relatively less than the traditional method. Therefore, we anticipated that this method could lead to a more accurate and objective facial reconstruction, especially in the Thai population.

The phenomena described above indicate that FSTT is a part of the more efficient 3-D facial reconstruction procedures. Indeed, 3-D facial reconstruction should be evaluated in ante-mortem or post-mortem individuals and compared to a recently updated face photo in future studies. Also, it is necessary to develop the 3-D computerized facial reconstruction process in different techniques, for example, semi- or automated method, advanced mathematical programming, and machine learning or artificial intelligence. Further work is proposed to acquire additional samples at a range of training groups and test groups. In this regard, findings demonstrated that the FSTT dataset, as well as 3-D computerized facial reconstruction procedures in the study, could be applied forensically and clinically to recreate the 3-D virtual face for an adult Thai population.

Footnotes

Acknowledgments

The authors are gratefully thankful for the support from the Forensic Osteology Research Center (FORC) and Excellence Center in Osteology Research and Training Center (ORTC) with partial support from Chiang Mai University. The authors are grateful to Dr Nop Kongdee and Dr Arus Kunkhet at the College of Arts, Media and Technology, Chiang Mai University for their helpful comments. Most importantly, the authors are sincerely thankful for all of the donor cadavers who donated their bodies and facial photographs for the improvement of education and research included in this study.

Author contribution

PN: conceptualization, methodology, and writing original draft. PP: conceptualization, methodology, and editing drafts. SP: provision of study materials and visualization. SP: data analysis, interpretation, and visualization. AS: conceptualization, methodology, and visualization. PM: conceptualization, editing drafts, supervision, project administration, and funding acquisition. All authors reviewed and commented on a draft of the manuscript and gave final approval for submission.

Guarantor

Pasuk Mahakkanukrauh—CorrespondingAuthor (pasuk034@gmail.com).

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

This work was supported by the Faculty of Medicine, Chiang Mai University (grant no.070/2563) and partially supported by Excellence Center in Osteology Research and Training Center (ORTC), Chiang Mai University.

ORCID iDs

Pagorn Navic

Pasuk Mahakkanukrauh

References

Wilkinson

. Facial reconstruction--anatomical art or artistic anatomy? J Anat 2010; 216: 235–250.

Deng

Zhou

, et al. A regional method for craniofacial reconstruction based on coordinate adjustments and a new fusion strategy. Forensic Sci Int 2016; 259: 19–31.

De Buhan

Nardoni

. A facial reconstruction method based on new mesh deformation techniques. Forensic Sci Res 2018; 3: 256–273.

Fernandes

Serra Mda

Da Silva

, et al. Tests of one Brazilian facial reconstruction method using three soft tissue depth sets and familiar assessors. Forensic Sci Int 2012; 214: 211.e211–217.

Fernandes

Pereira

Da Silva

, et al. Is characterizing the digital forensic facial reconstruction with hair necessary? A familiar assessors’ analysis. Forensic Sci Int 2013; 229: 164.e161–165.

Verzé

. History of facial reconstruction. Acta Biomed 2009; 80: 5–12.

Wilkinson

. Computerized forensic facial reconstruction. Forensic Sci Med Pathol 2005; 1: 173–177.

Claes

Vandermeulen

De Greef

, et al. Computerized craniofacial reconstruction: conceptual framework and review. Forensic Sci Int 2010; 201: 138–145.

Stephan

Caple

Guyomarc’h

, et al. An overview of the latest developments in facial imaging. Forensic Sci Res 2019; 4: 10–28.

10.

Vanezis

. Cranio-Facial reconstruction in forensic identification — historical development and a review of current practice. Med Sci Law 2000; 40: 197–205.

11.

Short

Khambay

Ayoub

, et al. Validation of a computer modelled forensic facial reconstruction technique using CT data from live subjects: a pilot study. Forensic Sci Int 2014; 237: 147.e141–148.

12.

Lee

Wilkinson

Hwang

. An accuracy assessment of forensic computerized facial reconstruction employing cone-beam computed tomography from live subjects. J Forensic Sci 2012; 57: 318–327.

13.

Lee

Wilkinson

Hwang

, et al. Correlation between average tissue depth data and quantitative accuracy of forensic craniofacial reconstructions measured by geometric surface comparison method. J Forensic Sci 2015; 60: 572–580.

14.

Miranda

Wilkinson

Roughley

, et al. Assessment of accuracy and recognition of three-dimensional computerized forensic craniofacial reconstruction. PLoS One 2018; 13: e0196770.

15.

Rajapakse

Madugalla

Amarasinghe

, et al. Facial muscle anatomy based approach for forensic facial reconstruction in Sri Lanka. In: International Conference on Advances in ICT for Emerging Regions (ed Vidanapathirana

), Colombo, Sri Lanka, 12–15 December 2012, pp. 27–35.

16.

Tedeschi-Oliveira

Melani

RFH

De Almeida

, et al. Facial soft tissue thickness of Brazilian adults. Forensic Sci Int 2009; 193: 127.e121–127.

17.

De Almeida

Michel-Crosato

De Paiva

, et al. Facial soft tissue thickness in the Brazilian population: new reference data and anatomical landmarks. Forensic Sci Int 2013; 231: 404.e401–407.

18.

Hwang

Park

Lee

, et al. Facial soft tissue thickness database for craniofacial reconstruction in Korean adults. J Forensic Sci 2012; 57: 1442–1447.

19.

Guyomarc’h

Santos

Dutailly

, et al. Facial soft tissue depths in French adults: variability, specificity and estimation. Forensic Sci Int 2013; 231: 411.e1–10.

20.

Thiemann

Keil

Roy

. In vivo facial soft tissue depths of a modern adult population from Germany. Int J Legal Med 2017; 131: 1455–1488.

21.

Perlaza Ruiz

. Facial soft tissue thickness of Colombian adults. Forensic Sci Int 2013; 229: 160.e161–169.

22.

Domaracki

Stephan

. Facial soft tissue thicknesses in Australian adult cadavers. J Forensic Sci 2006; 51: 5–10.

23.

Stephan

Preisler

. In vivo facial soft tissue thicknesses of adult Australians. Forensic Sci Int 2018; 282: 220.e1–12.

24.

Sandamini

Jayawardena

Batuwitage

, et al. Facial soft tissue thickness trends for selected age groups of Sri Lankan adult population. Forensic Sci Int 2018; 293: 102.e101–111.

25.

Chan

Listi

Manhein

. In vivo facial tissue depth study of Chinese-American adults in New York city. J Forensic Sci 2011; 56: 350–358.

26.

Dong

Huang

, Feng

, et al. Influence of sex and body mass index on facial soft tissue thickness measurements of the northern Chinese adult population. Forensic Sci Int 2012; 222: 396.e391–397.

27.

Jia

, Yang

, et al. Ultrasonic measurement of facial tissue depth in a northern Chinese Han population. Forensic Sci Int 2016; 259: 247.e241–246.

28.

De Greef

Claes

, Vandermeulen

, et al. Large-scale in-vivo caucasian facial soft tissue thickness database for craniofacial reconstruction. Forensic Sci Int 2006; 159: S126–146.

29.

Guyomarc’h

Dutailly

Charton

, et al. Anthropological facial approximation in three dimensions (AFA3D): computer-assisted estimation of the facial morphology using geometric morphometrics. J Forensic Sci 2014; 59: 1502–1516.

30.

Stephan

Preisler

Bulut

, et al.

Turning the tables of sex distinction in craniofacial identification: why females possess thicker facial soft tissues than males, not vice versa

. Am J Phys Anthropol 2016; 161: 283–295.

31.

Caple

Stephan

. A standardized nomenclature for craniofacial and facial anthropometry. Int J Legal Med 2016; 130: 863–879.

32.

Stephan

Meikle

Freudenstein

, et al. Facial soft tissue thicknesses in craniofacial identification: data collection protocols and associated measurement errors. Forensic Sci Int 2019; 304: 109965.

33.

Meikle

Stephan

. B-mode ultrasound measurement of facial soft tissue thickness for craniofacial identification: a standardized approach. J Forensic Sci 2020; 65: 939–947.

34.

Thitiorul

Prapayasatok

Na Lampang

, et al. Nose width and mouth width predictions in a Thai population using cone-beam computed tomography data. Aust J Forensic Sci 2019; 53: 1–10.

35.

Thitiorul

Mahakkanukrauh

Prasitwattanaseree

, et al. Three-Dimensional prediction of the nose for facial approximation in a Thai population. J Forensic Sci 2020; 65: 707–714.

36.

Slicer Community. http://www.slicer.org (2021, accessed 2 Augest 2021).

37.

Blender Foundation. https://www.blender.org (2018, accessed 2 Augest 2021).

38.

Autodesk Maya. http://usa.autodesk.com/maya/ (2019, accessed 2 Augest 2021).

39.

MakeHuman Community. http://www.makehumancommunity.org (2020, accessed 2 Augest 2021).

40.

Autodesk Maya. Soft select components. https://knowledge.autodesk.com/support/maya/learn-explore/caas/CloudHelp/cloudhelp/2022/ENU/Maya-Basics/files/GUID-FF7C8670-97C7-4C13-9A6F-3B0A8F881EC9-htm.html?us_oa=akn-us&us_si=eaef925b-f0b1-4cf9-9afa-7abe91a54495&us_st=Soft%20Select%20components (2020, accessed 2 Augest 2021).

41.

Stephan

Arthur

. Assessing facial approximation accuracy: how do resemblance ratings of disparate faces compare to recognition tests? Forensic Sci Int 2006; 159: S159–S163.

42.

Stephan

Cicolini

. Measuring the accuracy of facial approximations: a comparative study of resemblance rating and face array methods. J Forensic Sci 2008; 53: 58–64.

43.

Lee

Wilkinson

. The unfamiliar face effect on forensic craniofacial reconstruction and recognition. Forensic Sci Int 2016; 269: 21–30.

44.

Sforza

Grandi

De Menezes

, et al. Age- and sex-related changes in the normal human external nose. Forensic Sci Int 2011; 204: 205.e201–209.

45.

Chen

, et al. Age and sex related measurement of craniofacial soft tissue thickness and nasal profile in the Chinese population. Forensic Sci Int 2011; 212: 272.e271–276.

46.

Albert

Ricanek

Jr Patterson

. A review of the literature on the aging adult skull and face: implications for forensic science research and applications. Forensic Sci Int 2007; 172: 1–9.

47.

De Donno

Sablone

Lauretti

, et al. Facial approximation: soft tissue thickness values for caucasian males using cone beam computer tomography. Leg Med (Tokyo) 2019; 37: 49–53.

48.

Stephan

. 2018 Tallied facial soft tissue thicknesses for adults and sub-adults. Forensic Sci Int 2017; 280: 113–123.

49.

Houlton

TMR

Jooste

Steyn

. Testing regression and mean model approaches to facial soft-tissue thickness estimation. Med Sci Law 2021; 61: 170–179.

50.

Gietzen

Brylka

Achenbach

, et al. A method for automatic forensic facial reconstruction based on dense statistics of soft tissue thickness. PLoS One 2019; 14: e0210257.