Introduction

To avoid unfairly advantaging or disadvantaging people with no valid identity documents, courts and official bodies require reliable information on whether a person has crossed a range of different age thresholds. The search for an appropriate, evidence-based methodology for forensic age estimation has resulted in the establishment and development of different investigatory techniques for examining age-related developmental characteristics.

According to the current recommendations from the Study Group on Forensic Age Diagnostics (AGFAD), investigation of the maturity of the medial clavicular epiphyses is of key importance in determining whether someone has attained the age of 18, an age threshold of frequent legal significance [12]. A large number of X-ray and computed tomography studies in this area have now been published [1, 2, 7–9, 17, 25, 26]. Based on the information available at present, X-ray imaging should only be used in exceptional cases, since superimposition of other structures limits the usefulness of sagittal projections and there is insufficient reference data for oblique projections [25]. Of greater practical relevance, however, is computed tomographic evaluation of the clavicle, the evidence base for which is currently unequalled [13].

Where there is a legal basis for doing so, the use of X-rays for the purpose of forensic age estimation may be legitimate for a number of reasons. To minimise the effects of ionising radiation administered during investigations and improve the reliability of age estimation where there is no legal basis for performing X-ray examinations, in recent years, research efforts in the forensic age estimation field have focused strongly on non-X-ray methods for assessing ossification of the medial clavicular epiphysis.

Although results from ultrasound evaluation of clavicular development have been very promising [18], a high dependency on the experience of the investigator and a lack of documentation admissible in court proceedings have impeded the wider use of ultrasound techniques in forensic practice [19]. Not least for this reason, many authors consider the analysis of time-dependent changes to the appearance of the medial clavicular epiphysis in magnetic resonance imaging (MRI) to offer particular promise.

The first systematic attempts to use clavicular MRI for the purpose of forensic age estimation were made as long ago as 2007 [15]. In the interim, the impediments to its widespread use for practical age estimation discussed at the time appear less significant. MRI scanners are no longer confined to larger medical centres; investigations are no longer so expensive to perform and can be carried out much more quickly. Nevertheless, only a small number of MRI studies on clavicular forensic skeletal age estimation in both sexes and using exclusively the main stages of epiphyseal ossification described by Schmeling et al. [11] have been performed [3–5, 15, 20, 21, 23]. In comparison with modern computed tomography, the ability of this technique to provide meaningful information has therefore remained limited. In addition, a lack of scientifically validated concordant reference values means that a proposal to apply results from CT studies to staging using magnetic resonance imaging [20] can be discounted. There is significant evidence that assessment of the determined ossification stage of the medial clavicular epiphysis is dependent on the imaging technique used [22].

In addition to the main stages, this paper aims for the first time to apply the sub-stages of ossification of the medial clavicular epiphysis described by Kellinghaus et al. [6] to magnetic resonance imaging and to do so for both sexes. Use of these sub-stages is established practice in CT-based forensic age estimation.

Materials and methods

Three hundred ninety-five clavicle specimens retained following legally mandated autopsies at the Institutes for Legal Medicine in Berlin, Essen, Frankfurt, Hamburg, and Münster between 2004 and 2011 were examined. All of the institutes involved in the study have obtained the approval of the ethics committee of the medical faculty at the university of which they are a part for this research.

The study material comes from 125 female and 270 male cadavers aged between 10 and 30.

Table 1 shows an overview of the number of subjects in each age group for each sex. The ages used in compiling this table are calculated from the dates of birth and death given in the public prosecutor’s files. Because the different specimens were removed in the course of routine forensic work, it can be assumed that the ethnic origin and socioeconomic status of the study population roughly mirrors that of the German population as a whole. The study population showed no evidence of diseases likely to affect the speed of skeletal maturation.

Table 1 Age distribution (n = 395)
Full size table

After removal, the specimens were vacuum-sealed and frozen at −20 °C. Prior to the MRI scans, the specimens were stored at room temperature (20 °C) for 24 h.

Diagnostic imaging was carried out at the Translational Research Imaging Center (TRIC) operated by the Institute of Clinical Radiology at University Hospital Münster. A 3-T magnetic resonance scanner (Gyroscan Intera 3 T), manufactured by Philips Medical Systems, Best, the Netherlands, was used for image acquisition using a flexible two-channel coil (Flex-M coil).

The following examination sequences were used:

  • FFE-3D-T1 gradient echo sequence with selective water excitation (TR = 10 ms; TE = 4.6 ms; flip angle = 10°; number of excitations (NSA) = 12; matrix M × P = 120 × 174, reconstructed using zero filling to 480; slice thickness = 0.7 mm; FOV (FH × RL × AP) = 60 × 90 × 29.4 mm; acquisition time = 7.19 min)

  • 2D-T2 turbo spin echo sequence (TR = 2000 ms; TE = 80 ms; TSE factor, 12; number of excitations (NSA) = 8; matrix M × P = 200 × 296, reconstructed using zero filling to 512; slice thickness = 1.26 mm; FOV (FH × RL × AP) = 100 × 150 × 39.1 mm; acquisition time = 15.36 min; slices = 28).

The T2-weighted images primarily served, where necessary, to demonstrate the presence of residual cartilage from the former epiphyseal plate.

The MR images were evaluated on dedicated screens approved for use in reporting radiological images. The T1- and T2-weighted images in the coronal plane were examined and the extent of ossification of the medial clavicular epiphysis classified using the stages described by Schmeling et al. [11] (Table 2) and Kellinghaus et al. [6] (Table 3). Staging was carried out by two investigators, one of whom possessed very extensive experience in evaluation of the clavicle. The investigators were at no point provided with any information which would allow them to make inferences concerning the sex or chronological age of the person from whom the specimen originated.

Table 2 X-ray-based staging of the ossification of the medial clavicular epiphysis after Schmeling et al. [11]
Full size table
Table 3 Computed tomographic evaluation of sub-stages of ossification stages II and III of the medial clavicular epiphysis after Kellinghaus et al. [6]
Full size table

The two investigators initially produced a consensual evaluation of the clavicles from 315 specimens selected at random. To determine inter-observer agreement, each of the two investigators then independently evaluated the remaining 80 specimens. To determine intra-observer agreement, these 80 specimens were then re-evaluated by the experienced investigator after an interval of approximately 3 months. Subsequently, these 80 specimens were also subjected to consensual evaluation.

The two clavicles from the same specimen were initially staged independently. Where different ossification stages were assigned to the two clavicles, the higher of the two stages was taken as definitive.

The SPSS software package (IBM SPSS Statistics; version 22.0.0.0) was used for statistical analysis and to describe the data. Initially a range of statistics (mean, standard deviation, lower quartile, median, upper quartile, minimum, maximum) were calculated for the specific distributions of chronological age within the consensually determined ossification stage. To identify possible sex differences in the rate of clavicular ossification, the data was subjected to statistical analysis using the Mann-Whitney U test (significance level α = 0.05). Finally, the weighted kappa coefficients for measuring inter-observer and intra-observer agreement were calculated using the results from the 80 separately evaluated clavicles.

Results

Consensual evaluation by both investigators rated 32 of the 395 specimens (8.1 %) as unclassifiable. The main reason was the presence of unclassifiable anatomical variants (e.g. bowl-like shape) at the medial end of the clavicle. Three hundred sixty-three pairs of clavicles from 246 male and 117 female cadavers were therefore available for statistical analysis.

Figures 1, 2, 3, 4 and 5 show sample T1-weighted MR images illustrating the morphology of ossification stages 1 to 5.

Fig. 1
figure 1

Example of bilateral ossification stage 1 (age, 14; sex, male)

Full size image
Fig. 2
figure 2

Example of bilateral ossification stage 2c (age, 19; sex, male)

Full size image
Fig. 3
figure 3

Example of bilateral ossification stage 3c (age, 19; sex, female)

Full size image
Fig. 4
figure 4

Example of bilateral ossification stage 4 (age, 23; sex, female)

Full size image
Fig. 5
figure 5

Example of bilateral ossification stage 5 (age, 29; sex, female)

Full size image

Tables 4 and 5 provide an overview of statistics relating to the stage-specific distribution of chronological age calculated from the consensual evaluations.

Table 4 Statistics for female subjects (in years)
Full size table
Table 5 Statistics for male subjects (in years)
Full size table

The results presented here show that an increase in ossification stage is always accompanied by an increase in mean chronological age. With the exception of ossification stages 1 and 2a in female subjects and ossification stages 2c and 3a in male subjects, the median of the age intervals also increases with increasing ossification stage.

For female subjects in our study population, the minimum age at which ossification stage 2a occurred was 14.0, ossification stage 2b 15.5 and ossification stage 2c 15.8. In female subjects, the minimum age at which ossification stages 3a and 3b occurred was 16.4 and ossification stage 3c 19.3. The minimum age at which ossification stage 4 occurred in female subjects was 21.0. The minimum age at which ossification stage 5 was recorded was 26.6.

For male subjects in our study population, the minimum age at which ossification stage 2a was identified was 15.0, ossification stage 2b 16.0 and ossification stage 2c 17.3. For male subjects, the minimum age at which ossification stage 3a occurred was 16.3, ossification stage 3b 16.5 and ossification stage 3c 19.0.

For male subjects, the minimum age at which ossification stage 4 was diagnosed was 21.5. The minimum age at which ossification stage 5 occurred was 25.8.

From the data presented here, it is clear that the age at which the middle ossification stages occurred was lower in female subjects than in male subjects, though this difference was not statistically significant. This advance in maturation becomes statistically significant at ossification stage 4 (p = 0.014). By contrast, male subjects attained ossification stage 5 earlier than female subjects and this difference was statistically significant (p = 0.026).

The weighted kappa coefficient for determining inter-observer agreement was 0.986. The weighted kappa coefficient for intra-observer agreement was 0.992.

Discussion

Development of most areas of the skeleton is completed before the age of 18 is attained. Their value for forensic age estimation and for determining whether a legally significant age has been attained beyond the age of 17 is therefore limited. Current AGFAD recommendations [12] state that this significant gap in age estimation capability can only be filled by evaluating ossification of the medial clavicular epiphysis.

The current state of research suggests that, once maturation of the hand skeleton is complete, computed tomography of the clavicle is the best validated method for forensic age diagnostics [13, 22]. Nonetheless, the imperative to minimise radiation exposure means that there is a need for intensive research into non-ionising imaging techniques, in particular MRI. To date, there have been only a few papers on this subject, with a range of different study designs [3–5, 15, 20, 21, 23]. With 363 pairs of clavicles evaluated, this study represents the largest MRI study of the maturation of the medial clavicular epiphysis to date.

As in the studies by Hillewig et al. [3, 4] and Vieth et al. [23], we used a magnetic resonance scanner with a magnetic field strength of 3 T. By contrast, Schmidt et al. [15] and Hollnberger et al. [5] used 1.5 T scanners and Tangmose et al. [20, 21] a 1-T scanner. Because of their higher signal-to-noise ratios, higher field strengths allow better imaging of musculoskeletal structures [16]. Devices with a high magnetic field strength should therefore be preferred for MRI scans of the clavicle for the purpose of age estimation.

To obtain a more precise picture of the ossification stage of the medial clavicular epiphysis, in contrast with some previous studies, both T1- and T2-weighted images with differing tissue contrasts were acquired. In our experience, this is useful in borderline cases for demonstrating the presence of early-stage ossification bridges and residual cartilage from the former epiphyseal plate in ossification stage 3 and for identifying remnants of the epiphyseal scar in ossification stage 4.

Because of the large amount of data available and the method’s good ability to discriminate between stages, computed tomography of the clavicle with combined use of the staging systems described by Schmeling et al. [11] and Kellinghaus et al. [6] has become the established method for the evaluation of the maturity of the medial clavicular epiphysis. By contrast, most published MRI studies of forensic clavicular age estimation [5, 15, 20, 21] rely on the older four stage classification devised by Owings Webb and Myers Suchey [10]. Hillewig et al. [3] are the only authors to have used the extended staging model proposed by Schmeling et al. [11] in their initial investigations. Because, however, they were unable to detect any epiphyseal scars and thus assign any clavicles to ossification stage 4 in their study population, they returned in their follow-up study [4] to the Owings Webb and Myers Suchey classification [10]. Their conclusion that differentiation between ossification stages 4 and 5 is of little legal relevance due to the high minimum age at which ossification stage 5 occurs cannot be sustained. Particularly where a significant period of time lies between the date of investigation and the date on which the subject’s age was of legal relevance, it is frequently necessary to calculate the subject’s age on a previous date. Vieth et al. [23] have shown that the epiphyseal scar at ossification stage 4 as described by Schmeling et al. [11] can also be depicted with MRI. They also introduced staging of male subjects using the sub-stages of ossification stages 2 and 3 described by Kellinghaus et al. [6], which offers a more detailed basis for age estimation. This study for the first time presents statistics for both sexes produced from the combined use of the staging systems described by Schmeling et al. [11] and Kellinghaus et al. [6].

Following on from previous studies, this study confirms that MRI is capable of measuring the clavicular development parameters shown by X-ray imaging procedures to be relevant. The time-dependent maturation process of the medial clavicular epiphysis is well described by combining the classification systems used. Irregularities in the correlation between increasing ossification stage and increasing median chronological age were observed only in ossification stages 1 and 2a in female subjects and ossification stages 2c and 3a in male subjects. In addition to the relatively small number of subjects in each staging group, this can also be seen as confirming the hypothesis that ossification stage 3a of the medial clavicular epiphysis is in many cases attained before ossification stage 2c [24].

Because the aim in a forensic context is to determine whether the subject has attained a legally relevant age, the minimum age of stage-specific frequency distributions is of particular interest [14]. In forensic practice, the minimum age of ossification stage 3c according to Kellinghaus et al. [6] is of key significance. Based on the data presented here, in contrast to the minimum age at which the unsubdivided ossification stage 3 as described by Schmeling et al. [11] is attained, it demonstrates with certainty that the age of 18.0 has been attained. The same conclusion has previously been reached for a group of male amateur footballers in an MRI study carried out by Vieth et al. [23]. The minimum age of 19.0 determined in our study correlates closely with the minimum age of 18.7 given in that study. Our study also for the first time suggests that a similar diagnostic tool can be used for female subjects. According to our study, ossification stage 3c is not attained before the age of 19.3.

In addition, according to the results of this study, identification of a separate ossification stage 5 provides an important basis for forensic age estimation in practice. The relatively high minimum age of 26.6 for female and 25.8 for male subjects offers new diagnostic options where the age at a point in time lying several years in the past needs to be determined.

Statistically significant sex differences within stage-specific distributions of chronological age were observed only for ossification stages 4 and 5. In view of the data available, considerable caution should, however, be applied in interpreting this result. Because various X-ray studies, particularly studies involving the middle ossification stages, suggest that the clavicle matures earlier in females, the demonstration of a statistically significant difference in the age at which ossification stage 4 occurs between male and female subjects may be indicative of the higher specificity of MRI for detecting cartilage residues. This finding should be examined in future studies.

Looking at inter-observer and intra-observer agreement, it is clear that, for our study population, the investigation procedure is robust and resistant to the subjective influence of the investigator. Despite these results, we should not lose sight of the fact that MRI staging requires extensive experience.

This study offers a methodological basis for more comprehensive reference MRI studies. It thus represents an important step on the road to establishing investigation techniques which do not involve exposure to ionising radiation.

Conclusions

  1. 1.

    Magnetic resonance imaging-based evaluation of the maturity of the medial clavicular epiphysis using the staging systems described by Schmeling et al. [11] and Kellinghaus et al. [6] is a valid method for forensic age estimation in living subjects.

  2. 2.

    For both sexes, the inclusion of the sub-stages described by Kellinghaus et al. [6] allows a more precise evaluation to be made of whether a subject has reached the legally significant age of 18.

  3. 3.

    The study presented here represents a contribution towards extending the spectrum of forensic age estimation methods in cases where there is no legal basis for the use of ionising radiation. Where there is a legal basis for carrying out X-ray examinations, by minimising exposure to ionising radiation it also serves to realise one of the key aims of radiation protection legislation.