Background Pelvic insufficiency fractures are increasingly common in older adults and frequently involve the posterior pelvic ring. However, their population-based three-dimensional distribution has not yet been systematically characterized across the complete Fragility Fractures of the Pelvis (FFP) classification. This study aimed to generate a standardized CT-based three-dimensional fracture map, identify reproducible anatomical hotspots, and visualize fracture distribution across FFP types I–IV. Methods CT datasets from 67 geriatric patients with pelvic insufficiency fractures (mean age 82.78 ± 6.84 years; 77.6% women) were retrospectively analyzed. Fractures were classified according to the FFP classification of Rommens and Hofmann, segmented in 3D Slicer, registered to a standardized pelvic template, and visualized as frequency-based three-dimensional heatmaps in Blender. Results A total of 184 fracture lines were mapped. Fracture clustering was most pronounced in the sacral alae, accounting for 54% of all mapped fracture locations, followed by the parasymphyseal region (22%) and the remaining anterior pelvic ring (15%). Iliac wing and acetabular involvement was less frequent, accounting for 6% and 3%, respectively. FFP-specific heatmaps demonstrated a transition from isolated anterior fractures in FFP type I to progressively broader posterior and bilateral involvement in FFP types II–IV. Conclusion Standardized CT-based three-dimensional fracture mapping identified the sacral alae as the principal hotspot of pelvic insufficiency fractures and visualized characteristic spatial patterns across the FFP classification. This approach provides an anatomical framework for improved CT interpretation and future biomechanical, diagnostic, and surgical-planning studies. MINI ABSTRACT Pelvic insufficiency fractures are an increasing consequence of osteoporosis, but their three-dimensional distribution remains poorly understood. CT-based heatmapping identified the sacral alae as the principal fracture hotspot and visualized progressive posterior extension across the FFP classification. These findings improve the anatomical understanding of osteoporotic pelvic fragility fractures.
Contrast-enhanced computed tomography (CT) is useful for detecting active bleeding in patients with pelvic trauma. However, early detection of active bleeding from CT images is challenging due to the extensive image volume, which can delay intervention. Although the development of accurate and rapid detection support systems for pelvic active bleeding is highly anticipated, a significant data imbalance between bleeding and non-bleeding images in real-world hospital settings poses a significant challenge to clinically applicable models. We reverse-engineered the physician's diagnostic process and implemented a curriculum-based deep learning algorithm using indirect findings to prevent overfitting caused by the imbalance. In this multicenter study, we developed and implemented a multistage deep learning algorithm for the detection of active bleeding using enhanced CT of pelvic trauma patients. From April 18, 2008 to December 31, 2023, we collected 258,057 slices of whole-body CT images of 2,178 patients for blunt trauma with pelvic injuries from five emergency centers. Two types of classification-based deep learning models were developed and evaluated: anatomical structure extraction (ASE) and active bleeding detection (ABD) models. An algorithm using these models was implemented, and its accuracy in detecting active bleeding in the pelvis and the inference speed were evaluated. For validation data, the ASE and ABD models recognized the pelvis and active bleeding with areas under the receiver operating characteristic curve (AUCs) (accuracies) of 0.999 (99.4%) and 0.920 (83.0%), respectively. For test data, the pipeline achieved an AUC of 0.901 (accuracy: 77.5%) and an average inference time of 1.93s. In the observer performance study, the algorithm outperformed residents and demonstrated no difference with that of board-certified specialists. Our curriculum-based deep learning algorithm achieves a sufficiently high accuracy and inference speed for practical use. Our training approach using indirect findings offers a viable solution for developing imaging AI models even with an imbalanced dataset.
BACKGROUND:Occult Neck of Femur (NOF) fractures are radiographically elusive injuries. Undiagnosed, they lead to delayed treatment, displacement, and increased morbidity. While MRI is the gold standard for detection, CT is frequently used when MRI is unavailable. Physical examination manoeuvres, specifically the hip log-roll (HR) and straight leg raise (SLR), are often employed to guide decisions regarding cross-sectional imaging. This study explores the utility of these physical examinations in guiding the workup for occult NOF fractures. METHODS:A prospective service evaluation was performed at a 400-bed UK district general hospital (Jan 2023-Apr 2025). An electronic data-capture form integrated into CT/MRI requests identified 133 imaging requests ordered for suspected occult NOF fractures (120 CT and 14 MRI). Fifteen requests were cancelled (12 CT and 1 MRI out of a total 146 requests in 143 patients), and 14 CT Investigations related to suspected periprosthetic fracture were excluded from the primary native-hip analysis. Clinician-recorded results for HR, SLR, axial loading, and heel-percussion tests were compared against CT and MRI outcomes. Sensitivity, specificity, predictive values, and area under the ROC curve (AUC) were calculated for each test. A 4-point risk score (age > 70, female gender, positive HR, positive SLR) was evaluated for its discriminatory power. RESULTS:CT confirmed 38 occult NOF fractures, with 4 further cases confirmed on MRI (including one initially missed on CT) and 3 on initial radiographic re-assessment. ). HR and SLR were performed in more than 95% of patients, demonstrating high sensitivities of 0.96 and 0.91, respectively, but low specificities (<0.35). Axial loading and heel percussion showed lower sensitivities (0.68 and 0.64). Combined HR + SLR yielded a fair AUC (0.68), which improved to 0.72 when age and gender were incorporated into the risk score. CONCLUSIONS:In this setting, CT detected 97% of occult NOF fractures, missing only one micro-trabecular injury subsequently detected on MRI. The hip log roll and straight leg raise tests demonstrated high sensitivity but limited specificity, with improved performance when combined with demographic factors. These findings suggest that clinical examination manoeuvres are most useful for ruling out, rather than confirming, occult NOF fractures. Importantly, they reinforce the role of careful bedside physical examination in guiding appropriate imaging, particularly in an era of increasing reliance on radiological investigations.
