Evaluating dental student competency in forensic rugoscopy for human identification Authors: Panjit Chunhabundit, Tawepong Arayapisit, Natchalee Srimaneekarn Published: 2025 ABSTRACT Rugoscopy is a recognized method for human identification. Its reliability, however, may be compromised when general dentists with limited forensic training perform identification. Evaluating the role of future practitioners—dental students—is essential to address this competency gap. This study assessed the ability of dental students in recognizing and interpreting palatal rugae patterns for forensic identification. Fifty-four final-year dental students were evaluators. Forty maxillary casts (10 pre-orthodontic, 30 post-orthodontic) were used. All casts were duplicated and modified to retain only the palatal rugae area, with all teeth removed to prevent dental feature-based recognition. Participants matched each Set 1 cast to one in Set 2. The procedure was repeated after one week to determine test-retest reliability, assessed using prevalence-adjusted bias-adjusted kappa (PABAK). The mean accuracy for rugoscopy-based identification was 83.06% (± 15.49%), with an average time of 1.37 (± 0.34) minutes per matching. Reliability ranged from substantial to almost perfect, showing agreement rates between 74.07% and 96.30%. This simulated exercise demonstrated that trained dental students could achieve good accuracy and reliability in identifying individuals based on palatal rugae patterns. Consequently, trained dental students may play a valuable supportive role in forensic identification tasks when specialized forensic odontologists are unavailable. Further research is recommended to establish standardized criteria to improve both accuracy and reliability. Introduction In forensic science, the accurate establishment of personal identity is essential, particularly in cases involving mass fatalities, natural disasters, or severely decomposed remains. While standardized techniques such as DNA profiling, fingerprint analysis, and dental record comparison have proven effective, they are not without limitations. DNA analysis is sensitive to degradation from environmental exposure, requires specialized infrastructure, and incurs significant costs1–3. Likewise, fingerprint identification may be compromised by thermal damage, water immersion, or tissue decomposition4–6. Dental identification, although more resistant to postmortem changes, depends on the availability and quality of antemortem records, which may be incomplete or outdated, especially in underserved populations5,7. Given these challenges, alternative anatomical markers have gained attention, with palatal rugae emerging as a promising adjunct in forensic identification. Palatal rugae are irregular, asymmetric mucosal ridges located on the anterior portion of the hard palate, just behind the incisive papilla. Their configuration, including length, number, shape, and orientation, is considered unique to each individual, and remains relatively stable throughout life. This anatomical consistency forms the foundation of rugoscopy, the forensic study and application of palatal rugae patterns for personal identification. Unlike teeth or fingerprints, which may be damaged, missing, or deliberately altered, palatal rugae are well-protected within the oral cavity and less susceptible to postmortem changes, rendering them particularly valuable in challenging identification scenarios. The forensic value of rugoscopy lies in the resilience of palatal rugae under extreme conditions. Research has shown that these structures often remain intact even after exposure to fire, decomposition, or prolonged submersion5,6,8,9. The anatomy of the oral cavity, including the maxillary arch, surrounding musculature, lips, cheeks, tongue, and even residual dentition, serves as a natural guard, shielding the rugae from external forces. Moreover, palatal rugae are not significantly altered by dental extraction, orthodontic treatment, or aging, allowing for their use in identification even among edentulous individuals or in cases lacking complete dental records. Although modern approaches to rugoscopy, such as palatal scanning with intraoral scanners and the application of machine learning, have yielded promising results10, manual rugoscopy remains indispensable in low-resource settings. It provides a cost-effective and practical identification tool that requires minimal instrumentation and can be employed in field conditions without reliance on advanced technology. Despite the scientific merit and practical utility of palatal rugae in forensic identification, their routine application remains limited. This is primarily due to a lack of widespread training and awareness among general dental practitioners. In Thailand, for example, the number of certified forensic odontologists is extremely limited and centralized within only a few major provinces. Consequently, the responsibility of forensic consultation in rural or underserved regions often falls to general dentists, most of whom receive little to no formal instruction in forensic identification methods during their education. As a result, rugoscopy is rarely incorporated into standard forensic procedures, and its credibility in legal proceedings may be undermined due to inconsistent application or lack of expertise. This gap highlights an urgent need to integrate rugoscopy into dental curricula and forensic training programs, ensuring that future dental professionals are equipped with the knowledge and skills necessary to apply this procedure with confidence and accuracy. The successful application of palatal rugae in forensic practice depends not only on the inherent anatomical properties of the structure but also on the skill and knowledge of the professionals interpreting them. Without proper training and standardization, misinterpretation or inconsistent analysis could undermine the credibility of forensic evidence in court and affect the outcome of legal proceedings. To bridge this gap, the present study aims to evaluate the competency of dental students in recognizing and interpreting palatal rugae patterns for forensic identification. By assessing their skills, this research contributes to enhancing the preparedness of future dental professionals and supports the broader goal of strengthening forensic identification capabilities particularly in contexts where conventional methods are impractical or inaccessible. Methodology Ethical approval This study was approved by the Human Research Ethics Committees of the Faculty of Dentistry / Faculty of Pharmacy, Mahidol University, Institutional Review Board (Certificate of Approval No. MU-DT/PY-IRB 2025/049.3007). All methods were carried out in accordance with relevant guidelines and regulations. The informed consent was obtained from all subjects. Preparation of study models A total of 40 maxillary orthodontic study casts were obtained from patients who had previously undergone orthodontic treatment. To ensure the quality and consistency of the samples, inclusion criteria required clearly defined palatal rugae, absence of air bubbles, and comparable maxillary arch dimensions in terms of size, shape, and contour. Casts displaying prominent anatomical variations or anomalies, such as torus palatinus, were excluded to eliminate potential confounding identifiers. The selected casts were systematically categorized into two main sets: Set 1 comprised 10 pre-orthodontic treatment casts, further divided into three subgroups: Subgroup 1 (n = 6): Casts with corresponding post-treatment casts in Set 2, and no observable changes in maxillary arch dimensions.Subgroup 2 (n = 2): Casts with corresponding casts in Set 2, showing noticeable changes in arch dimensions following treatment.Subgroup 3 (n = 2): Casts with no corresponding post-treatment pairs in Set 2. Set 2 included 30 post-orthodontic treatment casts, also divided into three subgroups: Subgroup 1 (n = 6): Casts that matched those in Set 1 Subgroup 1, with unchanged arch dimensions.Subgroup 2 (n = 2): Casts that corresponded to Set 1 Subgroup 2, reflecting changes in arch dimensions.Subgroup 3 (n = 22): Casts with no corresponding pre-treatment pairs in Set 1. Set 1 comprised 10 pre-orthodontic treatment casts, further divided into three subgroups: Subgroup 1 (n = 6): Casts with corresponding post-treatment casts in Set 2, and no observable changes in maxillary arch dimensions.Subgroup 2 (n = 2): Casts with corresponding casts in Set 2, showing noticeable changes in arch dimensions following treatment.Subgroup 3 (n = 2): Casts with no corresponding post-treatment pairs in Set 2. Set 1 comprised 10 pre-orthodontic treatment casts, further divided into three subgroups: Subgroup 1 (n = 6): Casts with corresponding post-treatment casts in Set 2, and no observable changes in maxillary arch dimensions.Subgroup 2 (n = 2): Casts with corresponding casts in Set 2, showing noticeable changes in arch dimensions following treatment.Subgroup 3 (n = 2): Casts with no corresponding post-treatment pairs in Set 2. Subgroup 1 (n = 6): Casts with corresponding post-treatment casts in Set 2, and no observable changes in maxillary arch dimensions. Subgroup 1 (n = 6): Casts with corresponding post-treatment casts in Set 2, and no observable changes in maxillary arch dimensions. Subgroup 2 (n = 2): Casts with corresponding casts in Set 2, showing noticeable changes in arch dimensions following treatment. Subgroup 2 (n = 2): Casts with corresponding casts in Set 2, showing noticeable changes in arch dimensions following treatment. Subgroup 3 (n = 2): Casts with no corresponding post-treatment pairs in Set 2. Subgroup 3 (n = 2): Casts with no corresponding post-treatment pairs in Set 2. Set 2 included 30 post-orthodontic treatment casts, also divided into three subgroups: Subgroup 1 (n = 6): Casts that matched those in Set 1 Subgroup 1, with unchanged arch dimensions.Subgroup 2 (n = 2): Casts that corresponded to Set 1 Subgroup 2, reflecting changes in arch dimensions.Subgroup 3 (n = 22): Casts with no corresponding pre-treatment pairs in Set 1. Set 2 included 30 post-orthodontic treatment casts, also divided into three subgroups: Subgroup 1 (n = 6): Casts that matched those in Set 1 Subgroup 1, with unchanged arch dimensions.Subgroup 2 (n = 2): Casts that corresponded to Set 1 Subgroup 2, reflecting changes in arch dimensions.Subgroup 3 (n = 22): Casts with no corresponding pre-treatment pairs in Set 1. Subgroup 1 (n = 6): Casts that matched those in Set 1 Subgroup 1, with unchanged arch dimensions. Subgroup 1 (n = 6): Casts that matched those in Set 1 Subgroup 1, with unchanged arch dimensions. Subgroup 2 (n = 2): Casts that corresponded to Set 1 Subgroup 2, reflecting changes in arch dimensions. Subgroup 2 (n = 2): Casts that corresponded to Set 1 Subgroup 2, reflecting changes in arch dimensions. Subgroup 3 (n = 22): Casts with no corresponding pre-treatment pairs in Set 1. Subgroup 3 (n = 22): Casts with no corresponding pre-treatment pairs in Set 1. Fig. 1Cast preparation for palatal rugae analysis. The original orthodontic cast was trimmed to isolate the palatal rugae region for focused examination. (a) Original maxillary cast. (b) Trimmed cast showing only the palatal rugae. Cast preparation for palatal rugae analysis. The original orthodontic cast was trimmed to isolate the palatal rugae region for focused examination. (a) Original maxillary cast. (b) Trimmed cast showing only the palatal rugae. All selected casts were duplicated using alginate impression material (Alginate, Take 1®, KerrHawe S.A., Switzerland) and poured with plaster of Paris (Siam Gypsum Plaster L.P., Thailand). To standardize the specimens for palatal rugae analysis, the duplicated casts were trimmed by removing all teeth and alveolar ridge structures, leaving only the palatal rugae region intact (Fig. 1). Each cast was subsequently assigned a unique identification code for tracking and analysis purposes. Assessment of dental student matching performance Fifty-four sixth-year dental students from the Faculty of Dentistry at Mahidol University, all of whom had completed their graduation requirements, including coursework in basic forensic odontology, albeit without specific instruction in rugoscopy, were invited to participate in this study. Each participant was individually presented with casts from Set 1, while the corresponding Set 2 casts were randomly arranged on a table for comparison. Participants were instructed to match each Set 1 cast to a corresponding cast in Set 2 using only the palatal rugae pattern including number, shape, length and orientation of rugae as the basis for comparison without other materials. Once a match was made, the matched pair was collected, and both the identification number and the time taken for matching were recorded. The examiner timed each matching attempt using a stopwatch. Then, the same matching procedure was repeated with the same participants one week later to evaluate the consistency of performance. Assessment of accuracy and reliability The accuracy and mean matching time per cast were calculated and summarized using descriptive statistics. To assess the test–retest reliability between the two matching sessions, Cohen’s Kappa test, bias index, prevalence index and prevalence-adjusted bias-adjusted kappa (PABAK) were employed. All statistical analyses were conducted using R Statistical Software (v4.1.2; R Core Team 2021). Results A total of 54 sixth-year dental students from Mahidol University participated in this study, including 15 males and 39 females aged between 23 and 25 years. In the first matching session, using post-orthodontic casts with no observable changes in maxillary arch dimensions, participants achieved an average accuracy of 85.80% (± 18.43). The mean time required to match each cast in this group was 1.29 min (± 0.35). For the group in which post-treatment casts exhibited alterations in maxillary arch size, the matching accuracy dropped to an average of 61.11% (± 35.91), with an average matching time of 2.00 min (± 0.59). In the group consisting of unmatched casts with no corresponding pre-treatment pair, the mean accuracy was 69.44% (± 35.57), and the average time per match was 2.30 min (± 1.18). In the second matching session conducted one week later, a general improvement in both accuracy and efficiency was observed. For the group with no dimensional changes, the average matching accuracy increased to 93.83% (± 9.87), while the mean matching time decreased to 1.11 min (± 0.26). Among the casts with arch size changes, the accuracy improved to 79.63% (± 28.35), and the mean time spent per match was reduced to 1.35 min (± 0.45). In the unmatched group, the average accuracy increased to 81.48% (± 32.64), with a slightly reduced mean matching time of 2.22 min (± 0.44). These results indicated that casts with stable maxillary arch dimensions allowed for more accurate and efficient matching, and that repeated exposure improved performance across all groups. A detailed summary of these findings was provided in Table 1. Table 1Matching accuracy (%) and time spent per match (min) of both tests.Matching sessionGroupMatching accuracy (%)Time spent per match (min)Mean ± SD95% CIMean ± SD95% CIFirst1Matched casts unchanged arch dimensions85.80 ± 18.4380.77–90.831.29 ± 0.351.19–1.382Matched casts with observable changes in arch dimensions61.11 ± 35.9151.31–70.912.00 ± 0.591.44–2.163Unmatched casts69.44 ± 35.5759.73–79.162.30 ± 1.182.09–2.52Total77.59 ± 18.6372.51–82.681.44 ± 0.421.33–1.56Second1Matched casts unchanged arch dimensions93.83 ± 9.8791.13–96.521.11 ± 0.261.03–1.172Matched casts with observable changes in arch dimensions.79.63 ± 28.3571.89–87.371.35 ± 0.451.23–1.483Unmatched casts81.48 ± 32.6472.57–90.392.22 ± 0.442.09–2.34Total88.52 ± 12.3585.15–91.891.29 ± 0.271.22–1.37 Table 1Matching accuracy (%) and time spent per match (min) of both tests.Matching sessionGroupMatching accuracy (%)Time spent per match (min)Mean ± SD95% CIMean ± SD95% CIFirst1Matched casts unchanged arch dimensions85.80 ± 18.4380.77–90.831.29 ± 0.351.19–1.382Matched casts with observable changes in arch dimensions61.11 ± 35.9151.31–70.912.00 ± 0.591.44–2.163Unmatched casts69.44 ± 35.5759.73–79.162.30 ± 1.182.09–2.52Total77.59 ± 18.6372.51–82.681.44 ± 0.421.33–1.56Second1Matched casts unchanged arch dimensions93.83 ± 9.8791.13–96.521.11 ± 0.261.03–1.172Matched casts with observable changes in arch dimensions.79.63 ± 28.3571.89–87.371.35 ± 0.451.23–1.483Unmatched casts81.48 ± 32.6472.57–90.392.22 ± 0.442.09–2.34Total88.52 ± 12.3585.15–91.891.29 ± 0.271.22–1.37 Matching accuracy (%) and time spent per match (min) of both tests. The reliability of the matching performance over time was evaluated using Cohen’s Kappa test, bias index, prevalence index and PABAK comparing results from the first and second matching sessions conducted one week apart. In Group 1, which included post-orthodontic casts without changes in maxillary arch dimensions, the agreement between the two sessions was statistically significant, with percentage agreement ranging from 83.33% to 96.30%. For Group 2, comprising casts with changes in maxillary arch dimensions, statistically significant agreement was also observed, with agreement rates ranging from 75.93% to 79.63%. In Group 3, which included unmatched casts, the consistency between sessions remained statistically significant, with agreement ranging from 74.07% to 75.93%. These results demonstrated a high degree of intra-rater reliability, particularly when cast morphology remained stable (Table 2). Table 2The percentage of agreement on the kappa index, bias index, prevalence index and PABAK between the first and second tests.GroupCast No.Percentage of agreementKappa IndexBias IndexPrevalence IndexPABAK1196.300.7310.0370.8520.926294.440.3820.0560.9070.889394.440.6990.0560.7960.889492.590.6300.0740.7780.852588.890.5210.1110.7410.778683.330.4030.1670.6850.6672779.630.4280.2040.5740.593875.930.5040.1670.2410.5193974.070.3640.1110.4490.4821075.930.2990.1300.5740.519 Table 2The percentage of agreement on the kappa index, bias index, prevalence index and PABAK between the first and second tests.GroupCast No.Percentage of agreementKappa IndexBias IndexPrevalence IndexPABAK1196.300.7310.0370.8520.926294.440.3820.0560.9070.889394.440.6990.0560.7960.889492.590.6300.0740.7780.852588.890.5210.1110.7410.778683.330.4030.1670.6850.6672779.630.4280.2040.5740.593875.930.5040.1670.2410.5193974.070.3640.1110.4490.4821075.930.2990.1300.5740.519 The percentage of agreement on the kappa index, bias index, prevalence index and PABAK between the first and second tests. Based on the interpretation criteria established by Landis and Koch11, the reliability assessment using the PABAK revealed varying levels of agreement across the different cast groups. In Group 1, which included post-orthodontic casts with no changes in maxillary arch dimensions, the level of agreement ranged from substantial to almost perfect agreement, with a percentage of agreement between 83.33% and 96.30%. In Group 2, where the post-treatment casts exhibited changes in maxillary dimensions, the level of agreement was classified as moderate, with percentage agreement ranging from 75.93% to 79.63%. For Group 3, consisting of unmatched casts, the level of agreement was considered moderate, with a percentage of agreement ranging from 74.07% to 75.93%. Discussion This study evaluated the ability of senior dental students to perform human identification through palatal rugae matching (rugoscopy), a method increasingly recognized for its potential application in forensic odontology. To simulate real-world forensic conditions, both pre- and post-orthodontic treatment dental casts were used. In this context, pre-treatment models were considered analogs to ante-mortem records, while post-treatment models represented post-mortem remains. This design reflects common forensic scenarios in which identification depends on comparing dental features over time—particularly relevant when no other biometric identifiers are available. Previous research has consistently shown high accuracy levels in palatal rugae matching when conducted by experts or using advanced technological tools. English et al. (1988) achieved 100% matching accuracy when experts compared 25 pre-treatment and 125 post-treatment maxillary casts12. Similarly, Hemanth et al. (2009) reported 98% accuracy using 100 matched pairs13. Further advancements by Limson and Julian (2004) and Hemanth et al. (2010) introduced computer-aided identification systems based on digital image processing and pattern recognition, which yielded accuracy levels ranging from 92% to 99%14,15. More recently, studies employing intraoral scanners and machine-learning techniques for palatal analysis have also produced promising results10. Despite these technological developments, such approaches often require expert-driven landmark annotation and involve substantial financial and technical resources16, limiting their feasibility in resource-constrained settings. Consequently, manual rugoscopy remains a more practical option in many contexts; however, this raises a critical question: must manual analysis be performed exclusively by experts? To explore this question, the present study evaluated the performance of sixth-year dental students who had completed all graduation requirements but lacked prior experience in forensic casework and had received only basic instruction in forensic odontology without any formal training in rugoscopy. This cohort served as a representative sample of newly graduated general dentists, enabling assessment of whether the current undergraduate curriculum provides sufficient foundational knowledge to perform rugoscopy following only a brief orientation on rugae morphology comparison. The results showed that the students achieved an average matching accuracy of 83.06%. Although this level of accuracy is lower than that reported in expert-led or computer-assisted studies, it remains noteworthy. The findings indicate that even with minimal preparation, dental students are capable of performing palatal rugae matching with moderate accuracy. At the same time, the results highlight the need for enhanced training and methodological standardization to improve the reliability of this technique when applied by non-specialists. The average time taken per matching attempt was 1.37 ± 0.34 min. This is significantly shorter than the 8.98 ± 4.87 min reported by English et al. who utilized a 1:5 matching ratio (one ante-mortem model to five post-mortem models)12. In the current study, a 1:3 ratio was used. This comparison highlights a critical consideration: the time required for rugae-based matching does not increase linearly with the number of comparisons. Rather, it tends to increase exponentially as the volume of post-mortem records grows. This time burden becomes especially relevant in large-scale disasters or situations involving multiple unidentified victims. Thus, while palatal rugae analysis is a promising supplementary method, its practicality is constrained by the time and cognitive demand required for matching large datasets. Further, reliability assessment using Cohen’s kappa revealed inconsistencies that highlight well-known statistical limitations of this metric. Kappa is strongly influenced by prevalence (the imbalance between positive and negative classifications) and bias (systematic differences between raters). When the prevalence of one outcome is high, as in cases where most students correctly identify a cast, the expected agreement by chance becomes artificially inflated. This inflation can depress the kappa value, making reliability appear weaker than the actual level of agreement. Likewise, even small rater biases can distort kappa independently of true performance. Because of these effects, kappa may underestimate reliability in situations with high agreement but imbalanced outcomes17,18. This pattern was evident in our study, particularly for cast numbers 2, 5, and 6, where the observed agreement was high yet the corresponding kappa values remained low. To address this paradox, we additionally applied PABAK, which adjusts for both prevalence and bias. The PABAK results aligned more closely with the observed agreement rates, providing a more accurate reflection of true reliability. The study also analyzed performance across three cast groups designed to reflect different forensic scenarios. Group 1 involved unchanged post-treatment casts and demonstrated the highest accuracy and reliability. In contrast, Group 2 included casts that had undergone noticeable changes in dental arch size following orthodontic treatment. These alterations likely affected the relative width and spacing of the palatal rugae, although the overall rugae pattern remained stable19–21. Shukla et al. similarly reported that orthodontic expansion significantly impacts the lateral rugae width while preserving the general configuration22. Further analysis of the mismatches observed in Group 2 revealed a tendency toward lower accuracy, with post-orthodontically treated casts exhibiting greater dimensional changes without corresponding alterations in rugae morphology, consistent with observations reported in a previous study22. This suggests that spatial distortion, rather than true pattern loss, was the primary factor contributing to identification failures. These findings underscore the need for training that emphasizes recognizing rugae morphology independently of absolute dimensional measurements. Therefore, forensic practitioners must be trained to recognize rugae patterns beyond absolute dimensions when interpreting morphologically altered palatal rugae. Group 3 simulated a common but challenging forensic situation in which ante-mortem data is mistakenly submitted for someone who is not among the deceased. Participants’ lower performance in this group reflects the difficulty of identifying non-matches, especially for inexperienced individuals. In real-world cases, such mismatches can lead to severe consequences, including false identification and delayed processing of remains. These findings highlight the critical importance of operator expertise and suggest that even trained professionals must apply caution and critical evaluation when dealing with ambiguous cases. Interestingly, the second matching session, conducted one week after the first, showed improved accuracy across most participants. This improvement may be attributed to a learning or familiarization effect. After engaging with the initial matching task, participants likely developed a better understanding of the rugae patterns and became more meticulous in their evaluations. This insight has important implications for forensic education: even short-term, hands-on training sessions can significantly enhance performance in pattern-recognition tasks like rugoscopy. It is also important to consider the simulated nature of this study. Participants were aware that the task was academic and hypothetical, which may have affected their sense of urgency or accountability. In real forensic contexts involving actual victims and legal consequences, practitioners face pressure, emotional burden, and legal responsibility, factors that may either enhance caution or, conversely, increase the risk of error. This limitation reduces the ecological validity of the present simulation. Therefore, future research should be conducted in more realistic forensic settings, such as using anonymized forensic case materials under supervision or involving students in structured disaster victim identification drills. Moreover, studies with larger sample sizes may be necessary to detect meaningful group differences or to improve the generalizability of the findings. Such environments would provide a more accurate understanding of practitioner performance under real-world forensic demands. Another key consideration in rugae-based identification is time management. As the number of post-mortem records increases, such as in mass-casualty incidents, the cognitive load and time required for comparisons rise disproportionately. This makes manual rugoscopy increasingly impractical on a large scale unless the workload is supported by advanced tools or distributed among multiple trained personnel. In such contexts, computerized or advanced digital palatal analysis, which has demonstrated high levels of accuracy10,14–16, may play an increasingly important role. Therefore, while palatal rugae analysis can be highly effective in controlled or small-scale situations, its broader application necessitates operational adjustments or the integration of technological solutions. Despite the promising findings, a major limitation in the current field is a lack of a standardized protocol for performing rugoscopy. At present, there is no universally accepted method for identifying, documenting, or comparing palatal rugae in forensic practice. The absence of such guidelines hinders consistency, reproducibility, and cross-institutional validation. Establishing a standard protocol could significantly enhance the utility of rugoscopy by enabling non-specialist dental professionals, such as general dentists, dental students, and rural health providers, to contribute to forensic identification with higher reliability. Consequently, future research should focus on developing, validating, and disseminating standardized rugoscopy procedures that balance scientific rigor with practical feasibility. Conclusion Personal identification using palatal rugae can be effectively performed by dental students who have not received specialized training in forensic odontology. The findings demonstrate a good level of accuracy and a considerable degree of reliability, although the time required for manual matching remains a notable limitation. Consequently, this approach is most suitable for resource-limited, expert-limited settings with a relatively small number of post-mortem records. It is also important to acknowledge that digital and automated methods currently provide superior speed and performance where such technologies are available. Moving forward, the development of standardized protocols, both for manual analysis and for approaches guided by geometric modelling and advances in machine learning, will be essential for improving the consistency, accuracy, and practical utility of rugoscopy-based identification.