An exploratory radiologic index for evaluating diaphragmatic injury risk in left thoracoabdominal penetrating stab wounds: the diaphragmatic penetration index.

objectivesPenetrating injuries to the left thoracoabdominal (LTA) region pose significant diagnostic challenges due to the diaphragm's anatomical location. Early diaphragmatic injuries may be clinically silent and radiologically occult, whereas delayed diagnosis can result in visceral herniation and life-threatening complications. No studies have investigated the role of individualized anatomical parameters, specifically the distance from the entry wound to the diaphragm and diaphragm area, in predicting injury risk. Recognizing this gap, we aimed to introduce and assess a novel parameter, the Diaphragmatic Penetration Index (DPI), that incorporates both factors.

methodsThis retrospective study included 67 adult patients with single-entry LTA penetrating trauma who underwent laparoscopic or open surgical exploration between 2010 and 2025. Patients with multiple wounds, blunt trauma, cardiopulmonary resuscitation, missing CT scans, or loss to follow-up were excluded. DPI was defined as the ratio of diaphragm area (DA) (mm²) to entry wound-to-diaphragm distance (EHD) (mm), both measured from preoperative CT using 3D Slicer and Medseg.ai. Receiver operating characteristic (ROC) curve analysis was applied to evaluate DPI's diagnostic performance.

resultsThe mean patient age was 29.37 ± 11.80 years; 94.0% were male. Diaphragmatic injury was intraoperatively confirmed in 37.3% (n = 25) of patients. Median EHD was 82.48 mm, median DA was 32,450 mm², and median DPI was 394.9. ROC analysis demonstrated that EHD alone did not reach statistical significance (AUC 0.567, 95% CI: 0.415-0.716; p = 0.184), while DA (AUC 0.664, 95% CI: 0.530-0.791; p = 0.013) and DPI (AUC 0.654, 95% CI: 0.516-0.787; p = 0.018) achieved statistical significance but only modest discriminatory ability. Pairwise comparison revealed no significant difference between DA and DPI (p = 0.900), indicating that DA is the principal contributor to DPI performance. At the optimal DPI cut-off of 266.96 (Youden's index), sensitivity was 48.0% and specificity was 83.3%; retrospective simulation showed that applying this threshold prospectively would have spared 35 patients (83.3% of non-injured patients) unnecessary surgical exploration, while 13 diaphragmatic injuries (52.0%) would have remained undetected.

conclusionNeither DA, EHD, nor DPI demonstrated sufficient individual discriminatory performance to reliably predict diaphragmatic injury in penetrating left thoracoabdominal trauma. EHD was not independently associated with injury likelihood, and DPI offered no significant discriminatory advantage over DA alone. These findings suggest that anatomical parameters derived from CT-based measurements are inadequate as standalone triage tools, and that the high false negative rate at any clinically applicable threshold precludes their use as rule-out criteria. The quantification of this limitation is itself a clinically meaningful contribution, establishing a benchmark for future research. Prospective studies incorporating multiparametric models that combine anatomical, clinical, and radiological variables are required before any of these parameters can be considered for clinical adoption.