Introduction: A negative computed tomography (CT) scan does not preclude an insulinoma, as small lesions frequently remain undetected on conventional imaging. This case highlights the indispensable role of endoscopic ultrasound (EUS)— not just for localizing occult tumors, but as a definitive, minimally invasive therapeutic alternative to high-risk surgical resection. Case: A 39-year-old female with underlying hypertension presented with a 5-month history of predominantly fasting hypoglycemia (glucose <3.0 mmol/L) and neuroglycopenic symptoms, fulfilling Whipple’s triad. A supervised 72-hour fast confirmed endogenous hyperinsulinemic hypoglycemia at 31 hours, with a nadir glucose of 1.4 mmol/L, insulin 116 pmol/L, and C-peptide 821 pmol/L. Notably, contrastenhanced CT of the pancreas was reported as normal. To overcome this, EUS was performed, successfully identifying a hidden 19 × 18 mm lesion in the head of the pancreas, intimately abutting the main pancreatic duct. Despite medical therapy with diazoxide and strict dietary modifications, her hypoglycemia remained refractory. Given the tumor’s proximity to the main pancreatic duct, surgical enucleation carried a prohibitively high risk of complications. Consequently, she underwent EUS-guided radiofrequency ablation (RFA). Immediate post-procedure outcomes demonstrated near-complete resolution of the hypoglycemic episodes. Diazoxide was subsequently stopped. Outpatient continuous glucose monitoring confirmed sustained normoglycemia and marked symptom resolution, with no procedure-related complications. Conclusion The absence of a pancreatic lesion on CT demands persistent clinical suspicion in cases of biochemically proven hypoglycemia. EUS remains paramount for detecting occult lesions missed by standard imaging. Importantly, EUSRFA serves as a highly effective, tissue-sparing alternative to surgical resection for insulinomas, especially when conventional surgery poses prohibitive anatomical risks.
Insulinoma ; Tomography, X-Ray Computed

Introduction: Insulinoma is a rare functioning pancreatic neuroendocrine tumor and the most common cause of endogenous hyperinsulinemic hypoglycemia. Although biochemical confirmation is usually straightforward, tumor localization may be difficult when lesions are small and not detected on conventional cross-sectional imaging. In such cases, endoscopic ultrasound (EUS) plays an important role in identifying occult lesions and facilitating definitive treatment. Case: A 52-year-old female was admitted in April 2025 with recurrent seizures secondary to hypoglycemia for 3 years, with increasing frequency over time. She fulfilled Whipple’s triad, with documented capillary glucose of 1.8 mmol/L during an episode and symptom resolution following glucose administration. A supervised prolonged fasting test confirmed endogenous hyperinsulinemic hypoglycemia, with plasma glucose 1.4 mmol/L, insulin 122 pmol/L, and C-peptide 1,010 pmol/L. Short Synacthen test demonstrated adequate adrenal reserve. Due to persistent hypoglycemia, she required high-dose diazoxide. Contrast-enhanced computed tomography abdomen did not reveal a pancreatic lesion but incidentally detected a right ovarian teratoma. She underwent total abdominal hysterectomy and bilateral salpingo-oophorectomy in June 2025, with histopathology confirming a mature cystic teratoma without malignancy. However, hypoglycemic episodes persisted. Further evaluation with EUS in July 2025 identified a highly vascular isoechoic 8 × 8 mm lesion in the pancreatic body. Fine-needle biopsy confirmed a well-differentiated neuroendocrine tumor (WHO grade 1) with Ki-67 index of 2%. She subsequently underwent EUSguided radiofrequency ablation in August 2025. Follow-up EUS in December 2025 showed post-ablation change, and her hypoglycemic episodes resolved completely, allowing diazoxide to be discontinued. Conclusion This case highlights the diagnostic challenge of occult insulinoma in the presence of negative conventional imaging. EUS was pivotal for tumor localization and tissue diagnosis, while EUS-guided radiofrequency ablation provided effective minimally invasive treatment in a carefully selected patient.
Insulinoma ; Tomography, X-Ray Computed

OBJECTIVE: To explore the relationship between the effect of acupuncture and layered localization of computed tomography (CT) in treatment of lumbar disc herniation. METHODS: Based on the CT layered localization, the herniated lumbar discs were positioned in 5 layers, A, B, C, D and E among 300 patients with lumbar disc herniation. Combined with the horizontal and the frontal planes, the three-dimensional location was formed. Acupuncture was delivered at acupoints including bilateral Shenshu (BL23), Dachangshu (BL25), and Huantiao (GB30), Weizhong (BL40) on the affected side. One intervention of acupuncture was 30 min, once daily; 1 course of treatment was composed of 10 interventions and 2 courses were required. Before and after treatment, Japanese orthopaedic association (JOA) score was recorded, and the effect was evaluated. The curative effect was classified and compared with the CT layered localization. RESULTS: Of 300 patients, 226 cases were effective and the effective rate was 75.33%. The JOA scores of all patients, and in the effective group and the non-effective group were higher compared with the scores before treatment (P<0.05). With the layered localization considered, acupuncture was more effective on the cases positioned in C layer. Regarding the horizontal plane, the effect was better on the cases with zone 1 and zone 1-2 involved. In terms of the grade of frontal plane, acupuncture was more effective on the cases graded Ⅰ and Ⅱ. CONCLUSION The clinical effect of acupuncture on lumbar disc herniation is related with the layer and the horizontal zone of herniated disc positioned, as well as to the grade of the frontal plane.
Humans ; Acupuncture Therapy ; Intervertebral Disc Displacement/diagnostic imaging* ; Male ; Female ; Middle Aged ; Adult ; Tomography, X-Ray Computed ; Lumbar Vertebrae/diagnostic imaging* ; Acupuncture Points ; Aged ; Young Adult ; Treatment Outcome

BACKGROUND: There is little literature describing the artificial intelligence (AI)-aided diagnosis of severe pneumonia (SP) subphenotypes and the association of the subphenotypes with the ventilatory treatment efficacy. The aim of our study is to illustrate whether clinical and biological heterogeneity, such as ventilation and gas-exchange, exists among patients with SP using chest computed tomography (CT)-based AI-aided latent class analysis (LCA). METHODS: This retrospective study included 413 patients hospitalized at Xinhua Hospital diagnosed with SP from June 1, 2015 to May 30, 2020. AI quantification results of chest CT and their combination with additional clinical variables were used to develop LCA models in an SP population. The optimal subphenotypes were determined though evaluating statistical indicators of all the LCA models, and clinical implications of them such as guiding ventilation strategies were further explored by statistical methods. RESULTS: The two-class LCA model based on AI quantification results of chest CT can describe the biological characteristics of the SP population well and hence yielded the two clinical subphenotypes. Patients with subphenotype-1 had milder infections ( P <0.001) than patients with subphenotype-2 and had lower 30-day ( P <0.001) and 90-day ( P <0.001) mortality, and lower in-hospital ( P = 0.001) and 2-year ( P <0.001) mortality. Patients with subphenotype-1 showed a better match between the percentage of non-infected lung volume (used to quantify ventilation) and oxygen saturation (used to reflect gas exchange), compared with patients with subphenotype-2. There were significant differences in the matching degree of lung ventilation and gas exchange between the two subphenotypes ( P <0.001). Compared with patients with subphenotype-2, those with subphenotype-1 showed a relatively better match between CT-based AI metrics of the non-infected region and oxygenation, and their clinical outcomes were effectively improved after receiving invasive ventilation treatment. CONCLUSIONS A two-class LCA model based on AI quantification results of chest CT in the SP population particularly revealed clinical heterogeneity of lung function. Identifying the degree of match between ventilation and gas-exchange may help guide decisions about assisted ventilation.
Humans ; Tomography, X-Ray Computed/methods* ; Male ; Female ; Retrospective Studies ; Middle Aged ; Artificial Intelligence ; Aged ; Pneumonia/diagnosis* ; Latent Class Analysis ; Adult

Large language models (LLMs) such as ChatGPT, Claude, Llama, and Qwen are emerging as transformative technologies for the diagnosis and treatment of various diseases. With their exceptional long-context reasoning capabilities, LLMs are proficient in clinically relevant tasks, particularly in medical text analysis and interactive dialogue. They can enhance diagnostic accuracy by processing vast amounts of patient data and medical literature and have demonstrated their utility in diagnosing common diseases and facilitating the identification of rare diseases by recognizing subtle patterns in symptoms and test results. Building on their image-recognition abilities, multimodal LLMs (MLLMs) show promising potential for diagnosis based on radiography, chest computed tomography (CT), electrocardiography (ECG), and common pathological images. These models can also assist in treatment planning by suggesting evidence-based interventions and improving clinical decision support systems through integrated analysis of patient records. Despite these promising developments, significant challenges persist regarding the use of LLMs in medicine, including concerns regarding algorithmic bias, the potential for hallucinations, and the need for rigorous clinical validation. Ethical considerations also underscore the importance of maintaining the function of supervision in clinical practice. This paper highlights the rapid advancements in research on the diagnostic and therapeutic applications of LLMs across different medical disciplines and emphasizes the importance of policymaking, ethical supervision, and multidisciplinary collaboration in promoting more effective and safer clinical applications of LLMs. Future directions include the integration of proprietary clinical knowledge, the investigation of open-source and customized models, and the evaluation of real-time effects in clinical diagnosis and treatment practices.
Humans ; Large Language Models ; Tomography, X-Ray Computed

Artificial intelligence (AI), particularly deep learning, has demonstrated remarkable performance in medical imaging across a variety of modalities, including X-ray, computed tomography (CT), magnetic resonance imaging (MRI), ultrasound, positron emission tomography (PET), and pathological imaging. However, most existing state-of-the-art AI techniques are task-specific and focus on a limited range of imaging modalities. Compared to these task-specific models, emerging foundation models represent a significant milestone in AI development. These models can learn generalized representations of medical images and apply them to downstream tasks through zero-shot or few-shot fine-tuning. Foundation models have the potential to address the comprehensive and multifactorial challenges encountered in clinical practice. This article reviews the clinical applications of both task-specific and foundation models, highlighting their differences, complementarities, and clinical relevance. We also examine their future research directions and potential challenges. Unlike the replacement relationship seen between deep learning and traditional machine learning, task-specific and foundation models are complementary, despite inherent differences. While foundation models primarily focus on segmentation and classification, task-specific models are integrated into nearly all medical image analyses. However, with further advancements, foundation models could be applied to other clinical scenarios. In conclusion, all indications suggest that task-specific and foundation models, especially the latter, have the potential to drive breakthroughs in medical imaging, from image processing to clinical workflows.
Humans ; Artificial Intelligence ; Deep Learning ; Diagnostic Imaging/methods* ; Magnetic Resonance Imaging ; Tomography, X-Ray Computed ; Positron-Emission Tomography

Among numerous medical imaging modalities, diffusion weighted imaging (DWI) is extremely sensitive to acute ischemic stroke lesions, especially small infarcts. However, magnetic resonance imaging is time-consuming and expensive, and it is also prone to interference from metal implants. Therefore, the aim of this study is to design a medical image synthesis method based on generative adversarial network, Stroke-p2pHD, for synthesizing DWI images from computed tomography (CT). Stroke-p2pHD consisted of a generator that effectively fused local image features and global context information (Global_to_Local) and a multi-scale discriminator (M ²Dis). Specifically, in the Global_to_Local generator, a fully convolutional Transformer (FCT) and a local attention module (LAM) were integrated to achieve the synthesis of detailed information such as textures and lesions in DWI images. In the M ²Dis discriminator, a multi-scale convolutional network was adopted to perform the discrimination function of the input images. Meanwhile, an optimization balance with the Global_to_Local generator was ensured and the consistency of features in each layer of the M ²Dis discriminator was constrained. In this study, the public Acute Ischemic Stroke Dataset (AISD) and the acute cerebral infarction dataset from Yantaishan Hospital were used to verify the performance of the Stroke-p2pHD model in synthesizing DWI based on CT. Compared with other methods, the Stroke-p2pHD model showed excellent quantitative results (mean-square error = 0.008, peak signal-to-noise ratio = 23.766, structural similarity = 0.743). At the same time, relevant experimental analyses such as computational efficiency verify that the Stroke-p2pHD model has great potential for clinical applications.
Humans ; Tomography, X-Ray Computed/methods* ; Diffusion Magnetic Resonance Imaging/methods* ; Cerebral Infarction/diagnostic imaging* ; Stroke/diagnostic imaging* ; Neural Networks, Computer ; Image Processing, Computer-Assisted/methods* ; Algorithms

In this study, we propose an automatic contour outlining method to measure the spatial resolution of homemade automatic tube current modulation (ATCM) phantom by outlining the edge contour of the phantom image, selecting the region of interest (ROI), and measuring the spatial resolution characteristics of computer tomography (CT) phantom image. Specifically, the method obtains a binarized image of the phantom outlined by an automated fast region convolutional neural network (AFRCNN) model, measures the edge spread function (ESF) of the CT phantom with different tube currents and layer thicknesses, and differentiates the ESF to obtain the line spread function (LSF). Finally, the values passing through the zeros are normalized by the Fourier transform to obtain the CT spatial resolution index (RI) for the automatic measurement of the modulation transfer function (MTF). In this study, this algorithm is compared with the algorithm that uses polymethylmethacrylate (PMMA) to measure the MTF of the phantom edges to verify the feasibility of this method, and the results show that the AFRCNN model not only improves the efficiency and accuracy of the phantom contour outlining, but also is able to obtain a more accurate spatial resolution value through automated segmentation. In summary, the algorithm proposed in this study is accurate in spatial resolution measurement of phantom images and has the potential to be widely used in real clinical CT images.
Phantoms, Imaging ; Tomography, X-Ray Computed/instrumentation* ; Algorithms ; Neural Networks, Computer ; Image Processing, Computer-Assisted/methods* ; Humans ; Polymethyl Methacrylate

Alveolar bone reconstruction simulation is an effective means for quantifying orthodontics, but currently, it is not possible to directly obtain human alveolar bone material models for simulation. This study introduces a prediction method for the equivalent shear modulus of three-dimensional random porous materials, integrating the first-order Ogden hyperelastic model to construct a computed tomography (CT) based porous hyperelastic Ogden model (CT-PHO) for human alveolar bone. Model parameters are derived by combining results from micro-CT, nanoindentation experiments, and uniaxial compression tests. Compared to previous predictive models, the CT-PHO model shows a lower root mean square error (RMSE) under all bone density conditions. Simulation results using the CT-PHO model parameters in uniaxial compression experiments demonstrate more accurate prediction of the mechanical behavior of alveolar bone under compression. Further prediction and validation with different individual human alveolar bone samples yield accurate results, confirming the generality of the CT-PHO model. The study suggests that the CT-PHO model proposed in this paper can estimate the material properties of human alveolar bone and may eventually be used for bone reconstruction simulations to guide clinical treatment.
Humans ; Tomography, X-Ray Computed/methods* ; Porosity ; Alveolar Process/physiology* ; Bone Density ; Computer Simulation ; Elasticity ; X-Ray Microtomography ; Stress, Mechanical ; Finite Element Analysis ; Models, Biological

To address the issues of difficulty in preserving anatomical structures, low realism of generated images, and loss of high-frequency image information in medical image cross-modal translation, this paper proposes a medical image cross-modal translation method based on diffusion generative adversarial networks. First, an unsupervised translation module is used to convert magnetic resonance imaging (MRI) into pseudo-computed tomography (CT) images. Subsequently, a nonlinear frequency decomposition module is used to extract high-frequency CT images. Finally, the pseudo-CT image is input into the forward process, while the high-frequency CT image as a conditional input is used to guide the reverse process to generate the final CT image. The proposed model is evaluated on the SynthRAD2023 dataset, which is used for CT image generation for radiotherapy planning. The generated brain CT images achieve a Fréchet Inception Distance (FID) score of 33.159 7, a structure similarity index measure (SSIM) of 89.84%, a peak signal-to-noise ratio (PSNR) of 35.596 5 dB, and a mean squared error (MSE) of 17.873 9. The generated pelvic CT images yield an FID score of 33.951 6, a structural similarity index of 91.30%, a PSNR of 34.870 7 dB, and an MSE of 17.465 8. Experimental results show that the proposed model generates highly realistic CT images while preserving anatomical accuracy as much as possible. The transformed CT images can be effectively used in radiotherapy planning, further enhancing diagnostic efficiency.
Humans ; Tomography, X-Ray Computed/methods* ; Magnetic Resonance Imaging/methods* ; Image Processing, Computer-Assisted/methods* ; Neural Networks, Computer ; Brain/diagnostic imaging* ; Algorithms ; Radiotherapy Planning, Computer-Assisted ; Generative Adversarial Networks