R.A.D.I.C.I: Refining Automated age-at-Death estimation through Incremental Cementum Imaging

Age-at-death estimation represents a crucial aspect of biological anthropology, fundamental for reconstructing demographic structures of past populations, understanding human life-history traits, and supporting victims’ identification in forensic contexts. Traditional methods employed for this purpose rely on skeletal and dental changes occurring throughout the lifespan, showing high reliability in non-adults but reduced accuracy in adults, where age estimation reflects biological rather than chronological age. In adults, age estimation is based on degenerative changes, making the assessment of chronological age challenging. In this perspective, dental cementum annulation analysis has emerged as a promising alternative for age-at-death estimation, as cementum is continuously deposited throughout life and preserves yearly growth markers. However, the widespread reliance on manual counting limits the robustness and applicability of this approach because of its time-consuming nature and susceptibility to observer bias. This dissertation aims to address these limitations by evaluating sources of inaccuracy in manual annulation counting and developing a semi-automated, reproducible counting protocol. An R-based algorithm, biologically parametrised for annulation detection, was developed to automatically detect and count cementum annulations from high-resolution microscopic images. The method was validated on an osteological sample of 30 individuals with known age-at-death and subsequently applied to an archaeological population to test its applicability in bioarchaeological contexts, comparing results with osteological methods. Results demonstrate that the semi-automated approach achieves accuracy comparable to manual methods, which are highly prone to observer-related variability and dependent on observers’ experience. Further analyses revealed that pathological alterations of the tooth root represent a major source of error in automated age estimation, highlighting the importance of specimen selection prior to analysis. Overall, this work contributes to refining age-at-death estimation through cementum annulation counting by proposing a standardized approach that enhances reliability, efficiency, and applicability, even when only isolated teeth are available, as often occurs in archaeological and paleoanthropological contexts.