Drafting radiology reports is a complex task requiring flexibility, where radiologists tail content to available information and particular clinical demands. However, most current radiology report generation (RRG) models are constrained to a fixed task paradigm, such as predicting the full ''finding'' section from a single image, inherently involving a mismatch between inputs and outputs. The trained models lack the flexibility for diverse inputs and could generate harmful, input-agnostic hallucinations. To bridge the gap between current RRG models and the clinical demands in practice, we first develop a data generation pipeline to create a new MIMIC-RG4 dataset, which considers four common radiology report drafting scenarios and has perfectly corresponded input and output. Secondly, we propose a novel large language model (LLM) based RRG framework, namely LLM-RG4, which utilizes LLM's flexible instruction-following capabilities and extensive general knowledge. We further develop an adaptive token fusion module that offers flexibility to handle diverse scenarios with different input combinations, while minimizing the additional computational burden associated with increased input volumes. Besides, we propose a token-level loss weighting strategy to direct the model's attention towards positive and uncertain descriptions. Experimental results demonstrate that LLM-RG4 achieves state-of-the-art performance in both clinical efficiency and natural language generation on the MIMIC-RG4 and MIMIC-CXR datasets. We quantitatively demonstrate that our model has minimal input-agnostic hallucinations, whereas current open-source models commonly suffer from this problem.

Choledochoscopy, a natural orifice transluminal endoscopic surgery, represents an alternative to cholecystolithotomy in comparison to cholecystectomy and open gallbladder-preserving surgery. To reduce surgeon fatigue and increase surgical autonomy, this paper presents an autonomous local navigation (ALN) system for cholecystolithotomy, comprising a novel spring-based choledochoscope (CDS) with a monocular camera, a dataset-independent dynamic adaptive threshold segmentation (DATS) algorithm, and an ALN strategy. During navigation, the DATS detects bile duct bifurcations from the CDS view, selecting the target bifurcation based on bile duct anatomy with a wall-following mechanism. The CDS control method including bending and rotation control is then proposed based on the relationship between the position and direction of the bile duct center and the CDS characteristics. To ensure precise navigation, a closed-loop ALN strategy utilizing three interchangeable proportional controllers is developed, with task-specific proportionality factors. The efficacy of the DATS and CDS is validated through series experiments. Finally, both rigid and soft phantom experiments demonstrate the potential of the ALN system for cholecystolithotomy.

The finite-time tracking control issue is researched for a class of stochastic nonlinear systems with unknown measurement sensitivity and full-state tracking error constraints. Primarily, an improved error transformation technique based on the prescribed performance function (PPF) is introduced to avoid the “singularity” issue in the backstepping design process. Subsequently, the “explosion of complexity” matter is circumvented dexterously by the improved first-order filter. Then, a novel feedback control algorithm is developed to deal with the unknown measurement sensitivity. Furthermore, the symmetric barrier Lyapunov function (BLF) is employed to cope with the requirement of full-state tracking error performance constraints. Moreover, the finite-time stability theorem proves that the developed control project guarantees the boundedness of all system variables, and the tracking errors within the predesigned range in the finite time. Definitively, the validity of the designed scheme is demonstrated by simulations.

Proximal femoral fracture segmentation in computed tomography (CT) is essential in the preoperative planning of orthopedic surgeons. Recently, numerous deep learning-based approaches have been proposed for segmenting various structures within CT scans. Nevertheless, distinguishing various attributes between fracture fragments and soft tissue regions in CT scans frequently poses challenges, which have received comparatively limited research attention. Besides, the cornerstone of contemporary deep learning methodologies is the availability of annotated data, while detailed CT annotations remain scarce. To address the challenge, we propose a novel weakly-supervised framework, namely Rough Turbo Net (RT-Net), for the segmentation of proximal femoral fractures. We emphasize the utilization of human resources to produce rough annotations on a substantial scale, as opposed to relying on limited fine-grained annotations that demand a substantial time to create. In RT-Net, rough annotations pose fractured-region constraints, which have demonstrated significant efficacy in enhancing the accuracy of the network. Conversely, the fine annotations can provide more details for recognizing edges and soft tissues. Besides, we design a spatial adaptive attention module (SAAM) that adapts to the spatial distribution of the fracture regions and align feature in each decoder. Moreover, we propose a fine-edge loss which is applied through an edge discrimination network to penalize the absence or imprecision edge features. Extensive quantitative and qualitative experiments demonstrate the superiority of RT-Net to state-of-the-art approaches. Furthermore, additional experiments show that RT-Net has the capability to produce pseudo labels for raw CT images that can further improve fracture segmentation performance and has the potential to improve segmentation performance on public datasets. The code is available at: https://github.com/zyairelu/RT-Net.