Elsevier

Legal Medicine

Volume 17, Issue 2, March 2015, Pages 71-78
Legal Medicine

Age estimation based on bone length using 12 regression models of left hand X-ray images for Asian children below 19 years old

https://doi.org/10.1016/j.legalmed.2014.09.006Get rights and content

Highlights

  • Comparison between 12 regression models for age estimation.

  • This is the first study on age estimation based on bone length using X-ray images.

  • S-curve is the best regression model for age estimation using single bone method.

  • Multiple linear regression (MLR) is more applicable for age estimation.

  • MSE value produced by MLR is good for estimation of age on male.

Abstract

Age estimation was used in forensic anthropology to help in the identification of individual remains and living person. However, the estimation methods tend to be unique and applicable only to a certain population. This paper analyzed age estimation using twelve regression models carried out on X-ray images of the left hand taken from an Asian data set for subjects under the age of 19. All the nineteen bones of the left hand were measured using free image software and the statistical analysis were performed using SPSS. There are two methods to determine age in this study which are single bone method and all bones method. For single bone method, S-curve regression model was found to have the highest R-square value using second metacarpal for males, and third proximal phalanx for females. For age estimation using single bone, fifth metacarpal from males and fifth proximal phalanx from females can be used due to the lowest mean square error (MSE) value. To conclude, multiple linear regressions is the best techniques for age estimation in cases where all bones are available, but if not, S-curve regression can be used using single bone method.

Introduction

Forensic anthropology is the application of methods and knowledge of physical anthropology to solve medical problems with legal significance. The objective is usually to help in the identification of individual to predict what has happened, especially with regards to the evidence of foul play [1]. There have been some attempts to identify individuals using hand measurements [2], vertebral column length [3], leg length [4] and step length [5] in living persons. In addition, there have been several studies conducted on stature estimation from foot length [6] and cephalo-facial dimensions [7] in school age children.

Age estimation in living individuals often presents a clinical forensic medicine challenge with significant social and important legal concerns. Forensic identification requires increasingly sensitive and specific age estimation methods. For children, age estimation can be performed very accurately using morphological methods, because a great number of age-dependent morphological features (especially of the dental and skeletal system) can be evaluated. At the end of skeletal growth and development, only a few age-dependent features (e.g. the development of the third molars and bones of the wrist and hand) remain to be used for age estimation by morphological methods, resulting in a gradual decrease in accuracy with the increase of age. During adulthood, the accuracy of most morphological methods is poor; in this age group, a biochemical method (based on aspartic acid racemization in dentin) offers the most accurate results. Age estimation can be used if the identity of a victim of a murder case is unclear or if legal questions concerning children’s imputativeness have to be clarified. In such cases, age estimation in childhood may play a central role in the clarification of questions which has a major legal and/or social impact on the individual as well as on the community; this is the peculiarity of age estimation in forensic practice [8].

Various human anatomical sites have been analyzed in the estimation of age including the knee [9], teeth [10], [11], and hand-wrist bones [11], [12], [13]. For the latter, several methods have been used such as Greulich and Pyle [12], [13], [14], linear regression [9], [14], multiple regression [10], single quadratic regression (SQR) [11] and support vector regression (SVR) [11].

Literature shows that there are several inherent limitations in the estimation of age affected by different criteria, such as gender, ethnicity, socio-economic citation, nutrition and geographical location [15]. It has been concluded that results obtained from one population are not necessarily applicable in others. As such, specific studies have to be performed as the results would be unique to a particular population. All the reported methods of determining age are also unique to a particular study and may not be applicable for different available samples or data sets. Several studies also proved that, after the age of 30, age estimation methods are unreliable, with an average error of 12 years [16], [17].

The present study was undertaken to determine age using Asian left hand data sets between the age of newborn and 18 years old. From our literature review, our study is the first study using left hand bone measurement from X-ray images to estimate age by looking at the relationship between bone length and age. In contrast to the techniques used in previous works, 12 regression models, applied to the length of each bone in the left hand and its particular age, namely; Linear, Logarithmic, Inverse, Quadratic, Cubic, Compound, Power, S-Curve, Growth, Exponential and Logistic were used for the single bone method, and multiple linear regressions were used for the all bones method.

Section snippets

Materials and methods

A total of 333 X-ray images of Asian left hand bones from 166 males and 167 females were included in this study. To overcome ethical issues, these radiographs were collected from Children’s Hospital Los Angeles, along with patients’ demographic data and radiologists’ readings distributed in nineteen groups (newborn, 1–18) for both male and female. Age distribution of the subjects is shown in Table 1. The radiographs were collected by Image Processing and Informatics Lab of the University of

Results

Table 2 shows the intra-observer trials. They clearly indicated that there was no significant statistical difference between the three repeated measures for each bone in every image (p-value > 0.05). Table 3 shows the inter-observer trials. They indicated that there was no significant statistical difference between the two observers (p-value > 0.05). Table 4 shows an example of the single bone method using eleven regression models used in the analysis, their respective R-square values and

Discussion

Forensic anthropologists are continually attempting to improve methods of estimating the age through skeletal identifications [20]. The bones that are normally analyzed for age estimation are the knee [9], teeth [10], [11], and hand-wrist [12], [13], [14]. All these studies obtained their respective images from X-rays, except the one conducted by the Japanese on the premolar [9], where computed tomography was used, exposing the participants to a higher radiation dosage. The images of the knee

Conclusion

In this study, 333 images of the left hand from the Asian data set were used for age estimation analysis. For the single bone method, eleven regression models were applied on the length of each bones of the left hand while for all bones method, one regression model was applied to all bones of the left hand. The S-curve regression was the best regression model based on the highest R-Square value of 0.960 for male and 0.900 for female while multiple linear regressions applied to all bones is the

Acknowledgement

The authors are grateful to the Ministry of Higher Education, Malaysia, and the Research Management Centre, Universiti Teknologi Malaysia, for the Fundamental Research Grant Scheme. The authors also grateful for the images obtained online from the Image Processing and Informatics Lab, University of Southern California.

Reference (32)

Cited by (0)

View full text