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.

Tomography recovers internal volume from projection measurements. Formulated as inverse problems, classic computed tomography generally reconstructs attenuation property in a preset cartesian grid coordinate. While this is intuitive and convenient for digital display, such discretization leads to forward-backward projection inconsistency, and discrepancy between digital and effective resolution. We take a different perspective by considering the image volume as continuous and modelling forward projection as a hybrid continuous-to-discrete mapping from volume to detector elements, which we call "ray bundles". The ray bundle can be regarded as an unconventional heterogenous coordinate. Projections are modeled as line integrations along ray bundles in the continuous volume space and approximated by numerical integration using customized sample points. This modeling approach is conveniently supported with an implicit neural representation approach. By representing the volume as a function mapping spatial coordinates to attenuation properties and leveraging ray bundle projection, this approach reflects transmission physics and eliminates the need for explicit interpolation, intersection calculations, or matrix inversions. A novel sampling strategy is further developed to adaptively distribute points along the ray bundles, emphasizing high gradient regions to allocate computational resources to heterogenous structures and details. We call this system T-ReX to indicate Transmission Ray bundles for X-ray geometry. We validate T-ReX through comprehensive experiments across three scenarios: simulated full-fan projections with primary signal only, half-fan setups with simulated scatter and noise, and an in-house dataset with realistic acquisition conditions. These results highlight the effectiveness of T-ReX in sparse view X-ray tomography.

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.

Medical note de-identification is critical for protecting private information. The task demands the complete removal of all patient names and other sensitive information such as addresses and phone numbers from medical records. Accomplishing this goal is challenging, with many variations in the medical note formats and string representations. Existing deidentification approaches include pattern matching where extensive dictionary lists are constructed a prior; and entity tagging, which trains on a large word-wise annotated corpus. This motivates us to study an alternative to the existing approaches with a reduced annotation burden. In this work, we propose a novel approach that implicitly accounts for the language territory of sensitive information. Specifically, our approach incorporates a contextualized word embedding module and a multilayer perceptron to simultaneously infer the similarity of sensitive and non-sensitive vocabularies to a constructed landmark set, providing an overall sparsely supervised classification. To demonstrate the rationale, we present the principle and work pipeline with the task of name removal, but the proposed method applies to other strings as well. On a large cohort of hybrid clinical reports, including various forms of consulting, on-treatment-visit, and follow-up notes, we achieved >0.99 accuracies in our constructed training, validation, and testing sets. The sensitivity and specificity were 1.0 and 0.9973 respectively for two randomly selected reports, comparing favorably to the benchmark Stanford NER tagger, which achieved 0.8529 and 0.9969.