The rapid advancements in natural language processing, particularly the development of large language models (LLMs), have opened new avenues for managing complex clinical text data. However, the inherent complexity and specificity of medical texts present significant challenges for the practical application of prompt engineering in diagnostic tasks. This paper explores LLMs with new prompt engineering technology to enhance model interpretability and improve the prediction performance of pulmonary disease based on a traditional deep learning model. A retrospective dataset including 2965 chest CT radiology reports was constructed. The reports were from 4 cohorts, namely, healthy individuals and patients with pulmonary tuberculosis, lung cancer, and pneumonia. Then, a novel prompt engineering strategy that integrates feature summarization (F-Sum), chain of thought (CoT) reasoning, and a hybrid retrieval-augmented generation (RAG) framework was proposed. A feature summarization approach, leveraging term frequency-inverse document frequency (TF-IDF) and K-means clustering, was used to extract and distill key radiological findings related to 3 diseases. Simultaneously, the hybrid RAG framework combined dense and sparse vector representations to enhance LLMs' comprehension of disease-related text. In total, 3 state-of-the-art LLMs, GLM-4-Plus, GLM-4-air (Zhipu AI), and GPT-4o (OpenAI), were integrated with the prompt strategy to evaluate the efficiency in recognizing pneumonia, tuberculosis, and lung cancer. The traditional deep learning model, BERT (Bidirectional Encoder Representations from Transformers), was also compared to assess the superiority of LLMs. Finally, the proposed method was tested on an external validation dataset consisted of 343 chest computed tomography (CT) report from another hospital. Compared with BERT-based prediction model and various other prompt engineering techniques, our method with GLM-4-Plus achieved the best performance on test dataset, attaining an F1-score of 0.89 and accuracy of 0.89. On the external validation dataset, F1-score (0.86) and accuracy (0.92) of the proposed method with GPT-4o were the highest. Compared to the popular strategy with manually selected typical samples (few-shot) and CoT designed by doctors (F1-score=0.83 and accuracy=0.83), the proposed method that summarized disease characteristics (F-Sum) based on LLM and automatically generated CoT performed better (F1-score=0.89 and accuracy=0.90). Although the BERT-based model got similar results on the test dataset (F1-score=0.85 and accuracy=0.88), its predictive performance significantly decreased on the external validation set (F1-score=0.48 and accuracy=0.78). These findings highlight the potential of LLMs to revolutionize pulmonary disease prediction, particularly in resource-constrained settings, by surpassing traditional models in both accuracy and flexibility. The proposed prompt engineering strategy not only improves predictive performance but also enhances the adaptability of LLMs in complex medical contexts, offering a promising tool for advancing disease diagnosis and clinical decision-making.

Robotic central pancreatectomy is increasingly used for pre- or low-grade malignant tumors in the pancreatic body balancing preservation of pancreatic function while removing the target lesion.^1-3 Today, there is no established reconstruction method and high rates of postpancreatectomy fistulas (POPF) remain a significant concern. ^4,5 We developed novel technique involving transgastric pancreaticogastrostomy with an omental pedicle advancement flap to reduce the risk of POPF. Additionally, preoperative deep-learning 3D organ modeling plays a crucial role in enhancing spatial understanding to enhance procedural safety.^6,7 METHODS: A 76-year-old female patient with a 33-mm, biopsy-confirmed high-risk IPMN underwent robotic-assisted central pancreatectomy. Preoperative CT was processed with a deep-learning system to create a patient-specific 3D model, enabling virtual simulation of port configurations. The optimal setup was selected based on the spatial relationship between port site, tumor location, and anatomy A transgastric pancreaticogastrostomy with omental flap reinforcement was performed to reduce POPF leading to a simpler reconstruction compared to pancreaticojejunostomy. The procedure lasted 218 min with minimal blood loss (50 ml). No complications occurred, and the patient was discharged on postoperative Day 3 after drain removal. Final pathology showed low-grade dysplasia. This approach, facilitated by robotic assistance, effectively preserves pancreatic function while treating a low-grade malignant tumor. Preoperative 3D organ modeling enhances the spatial understanding with the goal to increase procedural safety. Finally, the omental pedicle advancement flap technique shows promise in possibly reducing the incidence or at least the impact of POPF.

Abdominal aortic aneurysms (AAAs) are progressive focal dilatations of the abdominal aorta. AAAs may rupture, with a survival rate of only 20\%. Current clinical guidelines recommend elective surgical repair when the maximum AAA diameter exceeds 55 mm in men or 50 mm in women. Patients that do not meet these criteria are periodically monitored, with surveillance intervals based on the maximum AAA diameter. However, this diameter does not take into account the complex relation between the 3D AAA shape and its growth, making standardized intervals potentially unfit. Personalized AAA growth predictions could improve monitoring strategies. We propose to use an SE(3)-symmetric transformer model to predict AAA growth directly on the vascular model surface enriched with local, multi-physical features. In contrast to other works which have parameterized the AAA shape, this representation preserves the vascular surface's anatomical structure and geometric fidelity. We train our model using a longitudinal dataset of 113 computed tomography angiography (CTA) scans of 24 AAA patients at irregularly sampled intervals. After training, our model predicts AAA growth to the next scan moment with a median diameter error of 1.18 mm. We further demonstrate our model's utility to identify whether a patient will become eligible for elective repair within two years (acc = 0.93). Finally, we evaluate our model's generalization on an external validation set consisting of 25 CTAs from 7 AAA patients from a different hospital. Our results show that local directional AAA growth prediction from the vascular surface is feasible and may contribute to personalized surveillance strategies.

This study aimed to develop a novel deep learning model, U₂-Attention-Net (U₂A-Net), for precise segmentation of parotid glands on radiotherapy localization CT images. CT images from 79 patients with head and neck cancer were selected, on which the label maps were delineated by relevant practitioners to construct a dataset. The dataset was divided into the training set (n = 60), validation set (n = 6), and test set (n = 13), with the training set augmented. U₂A-Net, divided into U₂A-Net V₁ (sSE) and U₂A-Net V₂ (cSE) based on different attention mechanisms, was evaluated for parotid gland segmentation based on the DL loss function with U-Net, Attention U-Net, DeepLabV3+, and TransUNet as comparision models. Segmentation was also performed using GDL and GD-BCEL loss functions. Model performance was evaluated using DSC, JSC, PPV, SE, HD, RVD, and VOE metrics. The quantitative results revealed that U₂A-Net based on DL outperformed the comparative models. While U₂A-Net V₁ had the highest PPV, U₂A-Net V₂ demonstrated the best quantitative results in other metrics. Qualitative results showed that U₂A-Net's segmentation closely matched expert delineations, reducing oversegmentation and undersegmentation, with U₂A-Net V₂ being more effective. In comparing loss functions, U₂A-Net V₁ using GD-BCEL and U₂A-Net V₂ using DL performed best. The U₂A-Net model significantly improved parotid gland segmentation on radiotherapy localization CT images. The cSE attention mechanism showed advantages with DL, while sSE performed better with GD-BCEL.

Shortened Abstract Cone-beam computed tomography (CBCT) has become a widely adopted modality for image-guided radiotherapy (IGRT). However, CBCT suffers from increased noise, limited soft-tissue contrast, and artifacts, resulting in unreliable Hounsfield unit values and hindering direct dose calculation. Synthetic CT (sCT) generation from CBCT addresses these issues, especially using deep learning (DL) methods. Existing approaches are limited by institutional heterogeneity, scanner-dependent variations, and data privacy regulations that prevent multi-center data sharing. To overcome these challenges, we propose a cross-silo horizontal federated learning (FL) approach for CBCT-to-sCT synthesis in the head and neck region, extending our FedSynthCT framework. A conditional generative adversarial network was collaboratively trained on data from three European medical centers in the public SynthRAD2025 challenge dataset. The federated model demonstrated effective generalization across centers, with mean absolute error (MAE) ranging from $64.38\pm13.63$ to $85.90\pm7.10$ HU, structural similarity index (SSIM) from $0.882\pm0.022$ to $0.922\pm0.039$, and peak signal-to-noise ratio (PSNR) from $32.86\pm0.94$ to $34.91\pm1.04$ dB. Notably, on an external validation dataset of 60 patients, comparable performance was achieved (MAE: $75.22\pm11.81$ HU, SSIM: $0.904\pm0.034$, PSNR: $33.52\pm2.06$ dB) without additional training, confirming robust generalization despite protocol, scanner differences and registration errors. These findings demonstrate the technical feasibility of FL for CBCT-to-sCT synthesis while preserving data privacy and offer a collaborative solution for developing generalizable models across institutions without centralized data sharing or site-specific fine-tuning.

Abdominal aortic aneurysms (AAAs) are progressive focal dilatations of the abdominal aorta. AAAs may rupture, with a survival rate of only 20\%. Current clinical guidelines recommend elective surgical repair when the maximum AAA diameter exceeds 55 mm in men or 50 mm in women. Patients that do not meet these criteria are periodically monitored, with surveillance intervals based on the maximum AAA diameter. However, this diameter does not take into account the complex relation between the 3D AAA shape and its growth, making standardized intervals potentially unfit. Personalized AAA growth predictions could improve monitoring strategies. We propose to use an SE(3)-symmetric transformer model to predict AAA growth directly on the vascular model surface enriched with local, multi-physical features. In contrast to other works which have parameterized the AAA shape, this representation preserves the vascular surface's anatomical structure and geometric fidelity. We train our model using a longitudinal dataset of 113 computed tomography angiography (CTA) scans of 24 AAA patients at irregularly sampled intervals. After training, our model predicts AAA growth to the next scan moment with a median diameter error of 1.18 mm. We further demonstrate our model's utility to identify whether a patient will become eligible for elective repair within two years (acc = 0.93). Finally, we evaluate our model's generalization on an external validation set consisting of 25 CTAs from 7 AAA patients from a different hospital. Our results show that local directional AAA growth prediction from the vascular surface is feasible and may contribute to personalized surveillance strategies.

High-resolution sinogram inpainting is essential for computed tomography reconstruction, as missing high-frequency projections can lead to visible artifacts and diagnostic errors. Diffusion models are well-suited for this task due to their robustness and detail-preserving capabilities, but their application to high-resolution inputs is limited by excessive memory and computational demands. To address this limitation, we propose HiSin, a novel diffusion based framework for efficient sinogram inpainting via resolution-guided progressive inference. It progressively extracts global structure at low resolution and defers high-resolution inference to small patches, enabling memory-efficient inpainting. It further incorporates frequency-aware patch skipping and structure-adaptive step allocation to reduce redundant computation. Experimental results show that HiSin reduces peak memory usage by up to 31.25% and inference time by up to 18.15%, and maintains inpainting accuracy across datasets, resolutions, and mask conditions.

Accurate liver-volume measurements from CT scans are essential for treatment planning, particularly in liver resection cases, to avoid postoperative liver failure. However, manual segmentation is time-consuming and prone to variability. Advancements in artificial intelligence (AI), specifically convolutional neural networks, have enhanced liver segmentation accuracy. We aimed to identify optimal CT phases for AI-based liver volume estimation and apply the model to track liver volume changes over time. We also evaluated temporal changes in liver volume in participants without liver disease. In this retrospective, single-center study, we assessed the performance of an open-source AI-based liver segmentation model previously reported, using non-contrast and dynamic CT phases. The accuracy of the model was compared with that of expert radiologists. The Dice similarity coefficient (DSC) was calculated across various CT phases, including arterial, portal venous, and non-contrast, to validate the model. The model was then applied to a longitudinal study involving 39 patients without liver disease (527 CT scans) to examine age-related liver volume changes over 5 to 20 years. The model demonstrated high accuracy across all phases compared to manual segmentation. Among the CT phases, the highest DSC of 0.988 ± 0.010 was in the arterial phase. The intraclass correlation coefficients for liver volume were also high, exceeding 0.9 for contrast-enhanced phases and 0.8 for non-contrast CT. In the longitudinal study, the model indicated an annual decrease of 0.95%. This model provides high accuracy in liver segmentation across various CT phases and offers insights into age-related liver volume reduction. Measuring changes in liver volume may help with the early detection of diseases and the understanding of pathophysiology.

Mesenteric malperfusion (MMP) is an uncommon but devastating complication of acute aortic dissection (AAD) that combines 2 life-threatening conditions-aortic dissection and acute mesenteric ischemia. The complex pathophysiology of MMP poses substantial diagnostic and management challenges. Currently, delayed diagnosis remains a critical contributor to poor outcomes because of the absence of reliable individualized risk assessment tools. This study aims to develop and validate a deep learning-based model that integrates multimodal data to identify patients with AAD at high risk of MMP. This multicenter retrospective study included 525 patients with AAD from 2 hospitals. The training and internal validation cohort consisted of 450 patients from Beijing Anzhen Hospital, whereas the external validation cohort comprised 75 patients from Nanjing Drum Tower Hospital. Three machine learning models were developed: the benchmark model using laboratory parameters, the multiorgan feature-based AAD complicating MMP (MAM) model based on computed tomography angiography images, and the integrated model combining both data modalities. Model performance was assessed using the area under the curve, accuracy, sensitivity, specificity, and Brier score. To improve interpretability, gradient-weighted class activation mapping was used to identify and visualize discriminative imaging features. Univariate and multivariate regression analyses were used to evaluate the prognostic significance of the risk score generated by the optimal model. In the external validation cohort, the integrated model demonstrated superior performance, with an area under the curve of 0.780 (95% CI 0.777-0.785), which was significantly greater than those of the benchmark model (0.586, 95% CI 0.574-0.586) and the MAM model (0.732, 95% CI 0.724-0.734). This highlights the benefits of multimodal integration over single-modality approaches. Additional classification metrics revealed that the integrated model had an accuracy of 0.760 (95% CI 0.758-0.764), a sensitivity of 0.667 (95% CI 0.659-0.675), a specificity of 0.783 (95% CI 0.781-0.788), and a Brier score of 0.143 (95% CI 0.143-0.145). Moreover, gradient-weighted class activation mapping visualizations of the MAM model revealed that during positive predictions, the model focused more on key anatomical areas, particularly the superior mesenteric artery origin and intestinal regions with characteristic gas or fluid accumulation. Univariate and multivariate analyses also revealed that the risk score derived from the integrated model was independently associated with inhospital mortality risk among patients with AAD undergoing endovascular or surgical treatment (odds ratio 1.030, 95% CI 1.004-1.056; P=.02). Our findings demonstrate that compared with unimodal approaches, an integrated deep learning model incorporating both imaging and clinical data has greater diagnostic accuracy for MMP in patients with AAD. This model may serve as a valuable tool for early risk identification, facilitating timely therapeutic decision-making. Further prospective validation is warranted to confirm its clinical utility. Chinese Clinical Registry Center ChiCTR2400086050; http://www.chictr.org.cn/showproj.html?proj=226129.

Brain aneurysm detection models, both in the literature and in industry, continue to lack generalizability during external validation, limiting clinical adoption. This challenge is largely due to extensive exclusion criteria during training data selection. The authors developed the first model to achieve generalizability using novel methodological approaches. Computed tomography angiography (CTA) scans from 2004 to 2023 at the study institution were used for model training, including untreated unruptured intracranial aneurysms without extensive cerebrovascular disease. External validation used digital subtraction angiography-verified CTAs from an international center, while prospective validation occurred at the internal institution over 9 months. A public web platform was created for further model validation. A total of 2194 CTA scans were used for this study. One thousand five hundred eighty-seven patients and 1920 aneurysms with a mean size of 5.3 ± 3.7 mm were included in the training cohort. The mean age of the patients was 69.7 ± 14.9 years, and 1203 (75.8%) were female. The model achieved a training Dice score of 0.88 and a validation Dice score of 0.76. Prospective internal validation on 304 scans yielded a lesion-level (LL) sensitivity of 82.5% (95% CI: 75.5-87.9) and specificity of 89.6 (95% CI: 84.5-93.2). External validation on 303 scans demonstrated an on-par LL sensitivity and specificity of 83.5% (95% CI: 75.1-89.4) and 92.9% (95% CI: 88.8-95.6), respectively. Radiologist LL sensitivity from the external center was 84.5% (95% CI: 76.2-90.2), and 87.5% of the missed aneurysms were detected by the model. The authors developed the first publicly testable artificial intelligence model for aneurysm detection on CTA scans, demonstrating generalizability and state-of-the-art performance in external validation. The model addresses key limitations of previous efforts and enables broader validation through a web-based platform.