Fracture injuries often lead to complex bone fragmentations, posing significant challenges for accurate segmentation in surgical planning and trauma assessment. Manual annotation of each fragment is time-consuming and inconsistent, while existing automated methods often fail to separate individual fragments due to the wide variation in fracture types, irregular fracture surface, and close inter-fragment contact. To address these challenges, we introduce FracSegmentator, a deep learning approach for bone fragment instance segmentation. The model takes extracted bone regions in CT as input and isolates individual fragments by identifying fracture surfaces and separating closely contacting structures. Central to our approach is a Trauma-Prior-Guided Contrastive Learning module, which incorporates clinical knowledge through memory-based attention to better distinguish fractured surfaces from healthy regions. We evaluate FracSegmentator on four datasets that cover a range of anatomical sites and fracture patterns. The method achieves state-of-the-art results across all datasets and demonstrates strong generalization capabilities. By delivering accurate and efficient fragment-level segmentation, FracSegmentator supports critical downstream tasks such as automated fracture diagnosis, surgical planning, and preoperative reduction simulation.

The segmentation of pelvic fracture fragments in CT and X-ray images is crucial for trauma diagnosis, surgical planning, and intraoperative guidance. However, accurately and efficiently delineating the bone fragments remains a significant challenge due to complex anatomy and imaging limitations. The PENGWIN challenge, organized as a MICCAI 2024 satellite event, aimed to advance automated fracture segmentation by benchmarking state-of-the-art algorithms on these complex tasks. A diverse dataset of 150 CT scans was collected from multiple clinical centers, and a large set of simulated X-ray images was generated using the DeepDRR method. Final submissions from 16 teams worldwide were evaluated under a rigorous multi-metric testing scheme. The top-performing CT algorithm achieved an average fragment-wise intersection over union (IoU) of 0.930, demonstrating satisfactory accuracy. However, in the X-ray task, the best algorithm achieved an IoU of 0.774, which is promising but not yet sufficient for intra-operative decision-making, reflecting the inherent challenges of fragment overlap in projection imaging. Beyond the quantitative evaluation, the challenge revealed methodological diversity in algorithm design. Variations in instance representation, such as primary-secondary classification versus boundary-core separation, led to differing segmentation strategies. Despite promising results, the challenge also exposed inherent uncertainties in fragment definition, particularly in cases of incomplete fractures. These findings suggest that interactive segmentation approaches, integrating human decision-making with task-relevant information, may be essential for improving model reliability and clinical applicability.

Pelvic fractures are among the most complex challenges in orthopedic trauma, which usually involve hipbone and sacrum fractures, as well as joint dislocations. Traditional preoperative surgical planning relies on the operator's subjective interpretation of CT images, which is both time-consuming and prone to inaccuracies. This study introduces an automated preoperative planning solution for pelvic fracture reduction, addressing the limitations of conventional methods. The proposed solution includes a novel multi-scale distance-weighted neural network for segmenting pelvic fracture fragments from CT scans, and a learning-based approach to restore pelvic structure, combining a morphable model-based method for single-bone fracture reduction and a recursive pose estimation module for joint dislocation reduction. Comprehensive experiments on a clinical dataset of 30 fracture cases demonstrated the efficacy of our methods. Our segmentation network outperformed traditional max-flow segmentation and networks without distance weighting, achieving a Dice similarity coefficient (DSC) of 0.986 ± 0.055 and a local DSC of 0.940 ± 0.056 around the fracture sites. The proposed reduction method surpassed mirroring and mean template techniques, and an optimization-based joint matching method, achieving a target reduction error of (3.265 ± 1.485) mm, rotation errors of (3.476 ± 1.995)°, and translation errors of (2.773 ± 1.390) mm. In the proof-of-concept cadaver studies, our method achieved a DSC of 0.988 in segmentation and 3.731 mm error in reduction planning, which senior experts deemed excellent. In conclusion, our automated approach significantly improves traditional preoperative planning, enhancing both efficiency and accuracy in pelvic fracture reduction.

Accurate orthopedic fracture reduction planning is essential for ensuring successful postoperative recovery and improving patient outcomes. However, current automatic methods are challenged by the complex and irregular fracture geometries and the scarcity of annotated training data. To address these challenges, we propose a novel approach that integrates learning-based shape restoration and fracture simulation. A transformer-based model is developed, which utilizes patch-to-patch shape translation and recursive fragment registration to iteratively refine fracture reduction poses. A deformable fracture generation model (DFGM) combines statistical shape modeling with clinically representative fracture patterns to generate diverse and realistic datasets, reducing the dependence on annotated samples. Tested on extensive clinical data with hipbone, sacrum, and femoral shaft fractures, the proposed method achieved mean errors of 1.85 mm and 3.40°, outperforming both template-based and existing learning-based methods. In addition, models trained solely on DFGM-synthesized data presented strong generalizability to real clinical data. The ablation experiments demonstrate the effectiveness of the fragment-aware network pipeline and the synthesis steps. Finally, a cadaver study with ground truth derived from the pre-injury scan further validated the performance of the method.

Multi-resolution hierarchical strategy is typically used in conventional optimization-based image registration to address large deformation and improve the chance of a good local minimum. A rough concept of the scale is captured in deep networks by the reception field of kernels, and it has been realized to be both desirable and challenging to capture convolutions of different scales simultaneously in registration networks. In this study, we propose an image registration network that is conscious of and self-adaptive to deformation of various scales. Dilated inception modules (DIMs) are proposed to incorporate receptive fields of different sizes in a computationally efficient way. Scale adaptive modules (SAMs) are proposed to guide and adjust shallow features using convolutional kernels with spatially adaptive dilation rate learned from deep features. DIMs and SAMs are integrated into the registration network which takes a U-net structure. The network is trained in an unsupervised setting and completes registration with a single evaluation run. Experiment with cardiac MRIs showed that the adaptive dilation rate in SAM corresponded well to the deformation scale. Evaluated with left ventricle segmentation, our method achieved a dice of (0.93±0.02), significantly better than SimpleElastix and networks without DIM or SAM. Assessment with respect to average surface distance was less than 2 millimeters (1.6 pixels), comparable to the best-performing SimpleElastix without statistical significance. Experiment with synthetic data also demonstrated the effectiveness of DIMs and SAMs, which leaded to a significant reduction in target registration error based on dense deformation field. The average registration time was 4 milliseconds for 2D image with size 256×256.