Bone age is a crucial variable in personal identification [1-3]. The pelvis, particularly the hip joint comprising the acetabulum and femoral head (Fig. 1), is a reliable skeletal element for age estimation [4-6]. The acetabulum is frequently utilized in bone age assessments [5, 7-11], along with the femoral head [12-15]. Factors like bone porosity and morphological variations are commonly employed to estimate age [9-15]. Currently, bone age assessments are conducted and reported by specialists, and their conclusions can vary. The logistical challenge of delivering physical bones to specialists also complicates the process. Bone imaging medical techniques [16-18] offer a faster and more convenient method than handling actual bones. This research used image processing to analyze the morphology and porosity of the acetabulum and femoral head. The area ratios and percent porosity were calculated and the relationships between these parameters and age were examined. The application of a standardized image processing protocol improved the accuracy and comparability of the results.

The bone samples were obtained from body donations made between 2011 and 2019 through the Forensic Osteology Research Center at Chiang Mai University, Thailand. The present study was conducted with the approval of the Research Ethics Committee, Faculty of Medicine, Chiang Mai University, Thailand (NONE-2565-08937). The dataset included images from 167 skeletons, comprising 59 females aged 30 to 88 and 108 males aged 28 to 97. Younger bone donors were scarce, with individuals aged 28 years and older in this study. Most donors succumbed to illness, resulting in reduced bone strength in many instances. A notable challenge is ensuring the cleanliness of the bones, as this directly influences image pixel values, potentially leading to inaccuracies in the analysis.

The laboratory photography setup involve a camera mounted on a tripod, with a shutter release cable to minimize any camera shake. A digital single-lens reflex camera with a 105 mm lens was employed for the imaging. The camera’s sensitivity to light was consistently set to 100, in accordance with established standards [19]. When using an electronic flash, the camera’s shutter speed was adjusted to 1/60 of a second, fast enough to prevent motion blur, with the aperture set to f/16 to ensure sharpness across the entire image. The bones were positioned at over 50 cm from the camera, ensuring that the acetabulum was fully captured within the frame. The camera was aligned parallel to the plane of the acetabulum. Images (Fig. 1) were taken of the acetabulum and femoral head on both the left and right sides, yielding four photographs per skeleton.

The image processing procedure first converted the color image (Fig. 1) into a grayscale image (Fig. 2; left). This involved transforming an red, green, blue image into a grayscale format [20]. The pixel values in the grayscale image ranged from 0 to 255, with 0 corresponding to black and 255 to white (Fig. 2; right). Values between 0 and 255 represented different shades of gray, transitioning from darker to lighter tones.

A histogram of the acetabular fossa was generated (Fig. 3), with the x-axis representing pixel intensity values ranging from 0 to 255 and the y-axis indicating the number of pixels at each intensity value [17, 21, 22]. A high peak in the histogram suggested a large concentration of pixels at a specific intensity. The histogram of the bone did not follow a normal distribution and lacked the characteristic symmetric bell curve. Statistical metrics such as the mean, standard deviation, maximum intensity, minimum intensity, the number of pixels within the acetabular fossa, and the total pixel count were calculated.

After extracting the statistical values from the histogram, the two key variables for analysis were the area ratio and percent porosity. Bone porosity and morphological variations are commonly employed to estimate age [9-15]. Bone strength or porosity varies according to a person’s age. Numerous studies have investigated shape variability or morphological changes associated with age. The area ratio was defined as the proportion of the acetabular fossa area relative to the total acetabulum area. This was calculated by dividing the number of pixels representing the acetabular fossa by the total number of pixels (Fig. 3). Percent porosity quantifies the porosity within the acetabular fossa area. This was calculated by determining the ratio of pixels identified with porosity to the total number of pixels within the acetabular fossa. Porosity pixels were defined based on the histogram’s mean and standard deviation, with a threshold set at the mean minus the standard deviation; pixels below this threshold were classified as porosity pixels [18, 21, 22]. The percent porosity was then computed by dividing the number of porosity pixels by the total number of pixels in the acetabular fossa [23, 24]. The percent porosity and area ratio, as the two primary variables, were then further refined to analyze and investigate their interrelationships with bone age.

The same procedure was applied to the femoral head. The three primary variables—area ratio, percent porosity, and the age of the femoral head—were similarly adjusted. The area ratio was defined as the proportion of the fovea capitis area to the overall femoral head area. Percent porosity was determined by calculating the ratio of porosity pixels within the fovea capitis to the total number of pixels in the fovea capitis.

Digital image processing techniques were applied to all images. The three variables—percent porosity, area ratio, and bone age—were adjusted for the acetabulum and femoral head, and then analyzed using statistical methods to examine their interrelationships.

Bone age was assumed to have a linear relationship with the other two variables. Regression analysis, F-tests, and analysis of variance (ANOVA) tables were used to determine these relationships, with a significance level of α=0.10, corresponding to a 90% confidence level (0.90). The medical parameter exhibited high variability, and the sample group did not adequately cover very young bones. To reduce the risk of type II errors, a significance level of α=0.10 was chosen [25, 26]. F-tests and ANOVA tables were used to assess whether the variables x and y were related, where y represented bone age and x represented either percent porosity or area ratio. The P-value reported the relationship between the variables x and y and compared this to the significance level. If the P-value was less than the significance level (α=0.10), the null hypothesis was rejected, indicating a statistically significant relationship between the variables x and y.

Each analysis was performed separately for 12 distinct groups, with results classified according to sex (female or male), represented as F or M; anatomical region (acetabulum or femoral head), indicated as Ac or Fh; and side (left, right, or both), denoted as L, R, or B. The term ‘both sides’ considered data from both the left and right sides. The two primary variables examined were percent porosity and area ratio. Comparative graphs were generated for each group to assess the relationship between these variables and bone age. Fig. 4 illustrates the correlation between percent porosity and age for the left acetabulum in males while Fig. 5 demonstrates the relationship between area ratio and age for the right femoral head in females.

The relationship between percent porosity and bone age (Table 1) was analyzed, with the number of samples indicated in each group. A significance level (α) of 0.10, corresponding to a 90% confidence level, was used. A P-value less than 0.10 signified a significant relationship between percent porosity and estimated age. A statistically significant correlation was observed in the male group, with the relationship evident across both the acetabulum and femoral head regions, as well as on both the left and right sides. The sample sizes in the male group were 83 for the acetabulum and 108 for the femoral head. In the female group, the sample sizes were 40 and 59 for the acetabulum and femoral head, respectively.

The relationship between the area ratio and age is presented in Table 2. At the same significance level, a significant relationship was only observed in the femoral head region, with no significant correlation found in the acetabulum. This relationship was evident in both males and females, and also on both the left and right sides.

As shown in Table 1, a relationship existed between percent porosity and estimated bone age in males, with no such relationship identified in females. Khomkham et al. [9] reported statistically significant correlations between age at death and the left side of the acetabular groove in females (r=0.61), acetabular rim porosity (r=0.59), and apex activity score in the left side of the male acetabulum (r=0.62). By contrast, Khomkham et al. [9] found no significant correlation between the porosity of the acetabular fossa and age at death. Our results concurred with Khomkham et al.’s [9] findings.

Considering the relationship between area ratio and age (Table 2), no significant correlation was found within the acetabulum, regardless of sex or laterality. The P-values for all comparisons within the acetabulum exceeded 0.10. San-Millán et al. [10] noted that sex and age significantly influenced acetabular shape variation, with age-related changes occurring in the acetabular structure in both sexes. These changes were associated with bone formation related to aging and affected the entire border of the lunate surface [10]. Our findings differed from San-Millán et al. [10]; however, there were methodological differences between the two studies. This research measured the area ratio of the acetabular fossa relative to the acetabulum, whereas San-Millán et al. [10] focused on the shape of the acetabular fossa.

The current method is based on statistical analysis at a 90% confidence level. Rissech et al. [5] reported an accuracy of 89% within 10-year intervals. Forensic bone analysis typically presents age as a range such as 40–50±10 years. Our study assessed images from 167 skeletons, with the minimum age 28 years. Ages below 28 years may not align with the established relationships. An essential consideration is the cleanliness of the bone, as this has a direct impact on pixel values. Analysis using image processing provides results more conveniently and quickly than direct bone observation by experts while sending bone images electronically to specialists is also significantly easier and faster than sending actual bone samples.