Discussion

A study has been conducted on the applicability of the Iscan method for assessing age at death via morphological analysis of the sternal end of the fourth right rib.

CT imaging is a crucial method for visualizing bone structure and its internal and external architecture in a non-invasive, non-destructive, and extremely rapid manner.

To our knowledge, this is the first study in literature applying the Iscan method to CT scans performed directly on isolated and treated rib specimens. Previous studies have demonstrated the applicability of the Iscan method using CT scans; however, they analyzed images obtained from post-mortem whole-body CT scans [13], living subjects [10], or clinical databases [14], indirectly reconstructing images of the costochondral joint. Our approach differs, as we physically sampled the rib specimens during autopsy—an approach previously described only by Dedouit et al. [8], who did not further process the specimens—then cleaned and prepared them before CT scanning. Although more complex, this methodological choice allowed us to optimize the visualization of morphological features, minimize potential positioning artifacts, and work under controlled and standardized conditions.

Initially, in our study, a descriptive analysis was carried out on the sample consisting of 112 fragments of the fourth rib, of which 74 came from male subjects and 38 came from female subjects. Then, the 112 ribs were scanned by CT, and the images obtained were analyzed and classified according to the Iscan method by two observers. Each observer made two observations of the 112 fragments of the fourth rib.

An agreement and reliability study was subsequently conducted between the two observers.

It was observed that the intra-observer agreement between the first and second scoring sessions is almost perfect: for the first observer, the value of the K-weighted is 0.980 (p-value < 0.01); for the second observer, the value of the K-weighted is 0.923 (p-value < 0.01).

The agreement between the evaluators’ judgements in the first and second scoring sessions is substantial: for the first scoring session, the concordance between the observers’ judgements has a K-weighted value of 0.749 (p-value < 0.01); for the second scoring session, the concordance between the observers’ judgements has a K-weighted value of 0.759 (p-value < 0.01).

Although the statistical measures are not directly comparable, similar results were observed by Dedouit et al. [8], who applied the Iscan method to PMCT for the assessment of age at death and obtained a substantial intra-operator reliability confirmed by a Krippendorff’s alpha coefficient of 0.79 as well as excellent inter-operator reliability confirmed by a Krippendorff’s alpha coefficient between 0.78 and 0.86.

Blaszkowska et al. [14] obtained for intra-operator concordance a substantial K-weighted value of 0.76 and an excellent intra-observer concordance with a K-weighted value of 0.825.

Recently, Beltran-Aroca et al. [15] have obtained results similar to those of our study, observing an almost perfect intra-operator agreement with a K-weighted value of 0.954 for CT images and a substantial inter-operator agreement with a K-weighted value of 0.727, again for CT scans.

When analyzing success rates by age groups in the present study, the highest accuracy was observed in the 76 and over age group for Observer 1 and the 46–60 age group for Observer 2, where phase assignments were more consistent with actual age. Conversely, the lowest success rates were found in the 31–45 age group for both observers, where misclassifications were more frequent. These findings suggest that Iscan’s method is more reliable for middle-aged and older individuals but shows limitations in both younger and elderly individuals.

It should be noted that phase agreement and classification success represent two different aspects of the method’s performance. In the present study, agreement refers to strict concordance in phase assignment between observers, whereas success reflects whether the assigned phase is compatible with the individual’s chronological age according to the original Iscan age ranges. Because these age ranges are intentionally broad and overlapping, different phase assignments may still correspond to the same age interval. Consequently, lower phase agreement may coexist with relatively high age-compatible success rates, reflecting an intrinsic characteristic of the Iscan phase system rather than a methodological inconsistency.

In terms of Iscan phases, phase 7 exhibited the highest success rate for Observer 1, while phase 6 had the highest accuracy for Observer 2, aligning well with expected age ranges. These results are partly in agreement with those obtained by Beltran-Aroca et al. [15], whose estimates were more accurate in stages 6 and 8. On the other hand, phase 3 had the lowest success rate for Observer 2, often leading to misclassifications. Finally, the low success rate for phase 1/Observer 1 is not meaningful for evaluating the method’s accuracy, since phase 1 did not include any cases with a corresponding actual ag.

This confirms that the method struggles to classify younger individuals accurately, while it performs more reliably in later phases.

The observed differences between Prior and Posterior KDE Density Distributions highlight systematic variations in the classification of rib phases using the Iscan method applied to CT imaging. The overrepresentation of higher phases in the Prior probability suggests that the Iscan model inherently expects a broader distribution of individuals in later phases. However, the rightward shift in Posterior distributions for phases 6 and above indicates a tendency for observers to assign higher phases more frequently than expected, potentially leading to overestimation of age at death in older individuals. The alignment between Prior and Posterior distributions in phases 3 to 5 suggests that these phases are more stable and reliably assigned.

These results contrast with those observed by Beltran-Aroca et al. [15] whose study shows a tendency to overestimate age in the early stages (1–3) and to underestimate age in the later stages (4–8), with two exceptions from phase 3, which tends to be underestimated, and phase 7, which tends to be overestimated.

The present study also shows an increased dispersion in phase 8, indicating greater variability in observer classifications. This finding likely reflects the nature of skeletal remodeling in advanced ages. In later stages of rib metamorphosis, morphological changes tend to become less clearly distinguishable, and the progression of degenerative features such as rim irregularity, osteophyte development, and cortical remodeling may occur in a more heterogeneous and individualized manner. As a consequence, specimens belonging to older age groups may present combinations of morphological features that do not perfectly correspond to a single phase description, increasing the subjective component of classification.

When evaluated through CT imaging, this variability may become even more apparent. Unlike traditional macroscopic examination, CT visualization allows the observer to appreciate both external morphology and internal structural alterations, including cortical density changes and joint surface sclerosis. These additional details may reveal differences between specimens that would otherwise be grouped within the same phase when using purely surface-based criteria.

In this context, the broad morphological spectrum currently included within phase 8 may encompass individuals at different stages of advanced skeletal remodeling. The dispersion observed in our results therefore suggests that this phase may represent a heterogeneous category rather than a single morphological stage. Future CT-based applications of the method could benefit from exploring a possible subdivision or refinement of the criteria defining this phase, in order to better capture the variability observed in older individuals and potentially improve classification consistency in advanced age groups.

Overall, the results emphasize the need to consider potential biases in phase-based age estimation methods, especially for older individuals, where classification variability increases.

Indeed, as previously stated by Russell et al. [16], one of the main limitations of the Iscan and Hartnett methods is that they are not accurate in describing the weight and quality of the bone, which are two decisive parameters in assigning a sample to the correct age stage, especially for elderly individuals. Furthermore, these are two bone characteristics that can only be assessed by touch, so this could be a limit, even in the case of applying the Iscan method to PMCT [15].

In 2018, Merritt [17] proposed a revised method for applying the Iscan and Hartnett methods to CT scans of the fourth rib that also considered weight and bone quality; through this new method an increase in the accuracy of assigning specimens to the correct age phase was achieved, especially for male subjects in the 30 years range and female subjects under 40 years of age.

Further investigation could explore whether integrating additional morphological parameters or revising phase definitions could improve the reliability of the Iscan method in forensic and anthropological contexts. Relative to this, our study shows how, by introducing a new CT-specific parameter (joint fossa sclerosis), additional information about the age of the individual from the cadavers can be obtained, further narrowing the age ranges.

As previously stated, three patterns of joint fossa sclerosis were identified by PMCT: a score of 0 was assigned if joint fossa sclerosis was not present; a score of 1 was assigned if the sclerosis involved < 50% of the joint surface; a score of 2 was assigned if the sclerosis involved > 50% of the joint surface. Some age groups (e.g., groups 1 and 2) fall solely into the sclerosis group 0, while for other age groups, the presence of sclerosis scores of 1 or 2 may lean toward higher ranges for corpse age-prediction. To our knowledge, no previous study on the application of the Iscan method to PMCT for the estimation of age at death has ever shown the presence of sclerosis of the fossa of the sternal end of the fourth rib.

The regression models from male subjects suggest that while the sclerosis parameter enhances overall age estimation accuracy, it introduces a minor increase in uncertainty for single-case assessments. These findings suggest that sclerosis is a structural modifier of phase-based classification rather than a simple additive factor. While it enhances differentiation among individuals within the same phase, it introduces greater variability in some age groups, particularly in older individuals, where sclerosis progresses more irregularly. The redistribution of estimated ages challenges the sequential interpretation of CT Iscan phases, indicating that skeletal aging follows a more individualized trajectory, with sclerosis playing a key role in accelerating or modifying expected age estimations. It has been observed that in male subjects, the presence of sclerosis leads to assigning the individual to a higher age phase. For example, an individual in stage 2 with a sclerosis score of 2 will be more likely to be assigned to a higher stage, compared to an individual in stage 3 with a sclerosis score of 0.

Conversely, findings of the regression models from female subjects suggest that while sclerosis increases individual-level variability, it does not create major inconsistencies in phase-based classification for females. Instead, it follows a more structured and predictable effect, modifying estimated ages within expected limits rather than shifting individuals into the next phase’s range. This pattern reinforces the importance of integrating sclerosis into forensic age estimation models but also highlights potential sex-based differences in how bone remodeling and sclerosis progression interact with phase-based classification.

It is therefore crucial to highlight how this new parameter should not be used alone in assigning samples to the correct age stage but as an additional evaluation to the Iscan method applied to PMCT.

From a practical forensic perspective, CT-based evaluation of the sternal end of the fourth rib may have several applications in identification contexts. In many forensic scenarios, particularly in cases of advanced decomposition, skeletonized remains, or mass disaster investigations, isolated rib fragments may be among the preserved skeletal elements available for analysis. The possibility of applying the Iscan method through CT imaging allows non-destructive examination, digital archiving of morphological features, and repeated evaluation by multiple experts, which may improve transparency and reproducibility in forensic casework.

In addition, the increasing use of post-mortem computed tomography (PMCT) in forensic practice makes CT-based rib analysis potentially applicable even before skeletal preparation, allowing preliminary age estimation during radiological examinations.

Although the Iscan method was originally developed for age-at-death estimation, CT-based evaluation of rib morphology could also represent a complementary indicator in radiological age estimation of living individuals. In such contexts, non-invasive imaging is essential, and internal bone features such as sclerosis may provide additional information not accessible through traditional examination. However, further validation on clinical datasets would be necessary before considering such applications in medico-legal practice.

Further studies are needed to improve this method of applying the Iscan method to CT scans of the sternal end of the fourth rib to determine the age at death from skeletal remains and also to determine its potential application in determining the age of living subjects for legal purposes [18].