Objective The aim of this study was to develop a predicted body weight-based drowning index (PBW-DI) and compare its diagnostic performance with that of the conventional drowning index (DI). Methods This retrospective study included 46 drowning and 26 acute cardiac death cases. The PBW-DI was calculated by dividing the combined weight of the lungs and pleural fluid by predicted body weight (PBW) derived from height and sex. Diagnostic performance was evaluated using receiver operating characteristic analysis. A secondary analysis was performed to assess the effects of advanced decomposition. Results PBW-DI values were significantly higher in drowning than in acute cardiac death cases (median, 27.7 [interquartile range, 25.0–31.4] vs. 18.4 [16.0–20.0] g/kg; p < 0.001). The area under the curve was 0.948 (95% confidence interval [CI], 0.888–1.000) for the PBW-DI and 0.893 (95% CI, 0.802–0.984) for the DI, with no significant difference (p = 0.280). In exploratory analyses, PBW-DI also showed higher AUCs than indices normalized to actual body weight (0.901; p = 0.0335) or height (0.906; p = 0.0371). The optimal PBW-DI cutoff was 23.6 g/kg, with 0.935 sensitivity and 1.000 specificity. PBW-DI values were significantly lower with advanced than with mild-to-moderate decomposition (22.9 [17.5–27.4] vs. 27.7 [25.0–31.4] g/kg; p = 0.002). Conclusions The PBW-DI showed good diagnostic performance without requiring spleen weight, which may be affected by pathological conditions. However, advanced decomposition should be considered when interpreting results. The PBW-DI may serve as a novel objective adjunct to the postmortem diagnosis of drowning.
Background Knife documentation is essential to compare weapon dimensions and shape with the wounds. Computed tomography (CT) is commonly used; however, it is affected by metal artifacts and optical three-dimensional methods struggle with reflective metals. Three-dimensional Gaussian splatting (3DGS), a photograph-based method, may overcome these limitations; however, its accuracy in weapon documentation is unknown. This preliminary study evaluated whether 3DGS can document the shape and surface appearance of knives with sufficient dimensional accuracy. Methods Eight knives were reconstructed using 3DGS from handheld photographs and scanned using CT. Seven dimensions were measured twice by two observers using a caliper on 3DGS models and CT images. Reliability was assessed using intraclass correlation coefficients (ICCs), and agreement with the physical measurement was assessed using the Bland–Altman analysis. Character legibility was scored to assess surface texture. Results Reliability for the blade length and width was excellent for all methods (ICC ≥ 0.991). The 3DGS measurement agreed closely with the physical measurement (mean difference, −0.29 to +0.12 mm) for blade width, whereas CT underestimated the blade width by 1.23–1.82 mm. The thin spine parts were overestimated by both methods and more markedly by CT. 3DGS 89–90% of characters were legible, including those ≤1.5 mm high. Conclusion In this preliminary study, 3DGS documented the shape and surface texture of knives using only a handheld camera, and the blade length and width agreed closely with physical measurement. However, measurement of thin structures requires caution, and comparison with standardized photography and validation in actual casework are needed.
