检索结果-内蒙古大学图书馆

3rd International Workshop on Applications of medical Artificial Intelligence, AMAI 2024 held in conjunction with the 27th International Conference on medical Image computing and Computer Assisted Intervention, MICCAI 2024

作者： Chen, Zhennong Yoon, Siyeop Strotzer, Quirin Khalid, Rehab Naeem Tivnan, Matthew Li, Quanzheng Gupta, Rajiv Wu, Dufan Center for Advanced Medical Computing and Analysis Massachusetts General Hospital and Harvard Medical School BostonMA02114 United States Department of Radiology Massachusetts General Hospital and Harvard Medical School BostonMA02114 United States

ISBN: (纸本)9783031820069

Head motion is a major source of image artifacts in head computed tomography (CT), degrading the image quality and impacting diagnosis. Image-domain-based motion correction is practical for routine use since it doesn’t rely on hard-to-obtain CT projection data. However, existing convolutional neural network (CNN)-based methods tend to over-smooth images, particularly in cases of moderate to severe 3D motion artifacts. Motivated by the improved image quality and more stable training of diffusion-based generative models, we propose a novel 3D head CT motion correction approach based on conditional diffusion, named HeadMotion-EDM (HM-EDM). This approach has three features. Firstly, we utilize motion-corrupted images as the conditional input. Secondly, we leverage the advanced Elucidated Diffusion Model (EDM) framework, which integrates several pivotal engineering improvements in the diffusion model and significantly expedites the sampling process. Thirdly, we design an efficient 3D-patch-wise training method for 3D CT data. Comparative studies demonstrate that our approach surpasses CNN-based techniques as well as the denoising diffusion probabilistic model (DDPM) in both simulation and phantom studies. Furthermore, radiologists reviewed the results of applying HM-EDM to real-world portable head CT scans, showing its effectiveness in eliminating motion artifacts and improving diagnostic value. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

MAST-Pro: Dynamic Mixture-of-Experts for Adaptive Segmentation of Pan-Tumors with Knowledge-Driven Prompts

arXiv

引用

arXiv 2025年

作者： Meng, Runqi Song, Sifan Jin, Pengfei Oh, Yujin Teng, Lin Wang, Yulin Sun, Yiqun Chen, Ling Li, Xiang Li, Quanzheng Guo, Ning Shen, Dinggang School of Biomedical Engineering State Key Laboratory of Advanced Medical Materials and Devices ShanghaiTech University China Center of Advanced Medical Computing and Analysis Massachusetts General Hospital Harvard Medical School Harvard University United States Shanghai United Imaging Intelligence Co. Ltd China Shanghai Clinical Research and Trial Center China

Accurate tumor segmentation is crucial for cancer diagnosis and treatment. While foundation models have advanced general-purpose segmentation, existing methods still struggle with: (1) limited incorporation of medical priors, (2) imbalance between generic and tumor-specific features, and (3) high computational costs for clinical adaptation. To address these challenges, we propose MAST-Pro (Mixture-of-experts for Adaptive Segmentation of pan-Tumors with knowledge-driven Prompts), a novel framework that integrates dynamic Mixture-of-Experts (D-MoE) and knowledge-driven prompts for pan-tumor segmentation. Specifically, text and anatomical prompts provide domain-specific priors, guiding tumor representation learning, while D-MoE dynamically selects experts to balance generic and tumor-specific feature learning, improving segmentation accuracy across diverse tumor types. To enhance efficiency, we employ Parameter-Efficient Fine-Tuning (PEFT), optimizing MAST-Pro with significantly reduced computational overhead. Experiments on multi-anatomical tumor datasets demonstrate that MAST-Pro outperforms state-of-the-art approaches, achieving up to a 5.20% improvement in average DSC while reducing trainable parameters by 91.04%, without compromising accuracy. © 2025, CC BY-NC-ND.

关键词： Image segmentation

来源：评论

学校读者我要写书评

暂无评论

An Ordinary Differential Equation Sampler with Stochastic Start for Diffusion Bridge Models

SSRN

引用

SSRN 2025年

作者： Wang, Yuang Jin, Pengfei Zhang, Li Li, Quanzheng Chen, Zhiqiang Wu, Dufan The Department of Engineering Physics Tsinghua University 30 Shuangqing Road Haidian Beijing100084 China Center for Advanced Medical Computing and Analysis Massachusetts General Hospital Harvard Medical School 399 Revolution Dr SomervilleMA02145 United States

Diffusion bridge models have demonstrated promising performance in conditional image generation tasks, such as image restoration and translation, by initializing the generative process from corrupted images instead of pure Gaussian noise. However, existing diffusion bridge models often rely on Stochastic Differential Equation (SDE) samplers, which result in slower inference speed compared to diffusion models that employ high-order Ordinary Differential Equation (ODE) solvers for acceleration. To mitigate this gap, we propose a high-order ODE sampler with a stochastic start for diffusion bridge models. To overcome the singular behavior of the probability flow ODE (PF-ODE) at the beginning of the reverse process, a posterior sampling approach was introduced at the first reverse step. The sampling was designed to ensure a smooth transition from corrupted images to the generative trajectory while reducing discretization errors. Following this stochastic start, Heun's second-order solver is applied to solve the PF-ODE, achieving high perceptual quality with significantly reduced neural function evaluations (NFEs). Our method is fully compatible with pretrained diffusion bridge models and requires no additional training. Extensive experiments on image restoration and translation tasks, including super-resolution, JPEG restoration, Edges-to-Handbags, and DIODE-Outdoor, demonstrated that our sampler outperforms state-of-the-art methods in both visual quality and Frechet Inception Distance (FID). © 2025, The Authors. All rights reserved.

关键词： Stochastic systems

来源：评论

学校读者我要写书评

暂无评论

Cine Cardiac Magnetic Resonance Segmentation using Temporal-spatial Adaptation of Prompt-enabled Segment-Anything-Model: A Feasibility Study

引用

Journal of cardiovascular magnetic resonance : official journal of the Society for Cardiovascular Magnetic Resonance 2025年 101909页

作者： Zhennong Chen Sekeun Kim Hui Ren Sunghwan Kim Siyeop Yoon Quanzheng Li Xiang Li From Center for Advanced Medical Computing and Analysis Massachusetts General Hospital and Harvard Medical School Boston MA 02114. From Center for Advanced Medical Computing and Analysis Massachusetts General Hospital and Harvard Medical School Boston MA 02114. Electronic address: xli60@mgh.harvard.edu.

BACKGROUND:We propose an approach to adapt a segmentation foundation model, segment-anything-model (SAM), for cine Cardiac Magnetic Resonance (CMR) segmentation and evaluate its generalization performance on unseen datasets. METHODS:We present our model, cineCMR-SAM, which introduces a temporal-spatial attention mechanism to produce segmentation across one cardiac cycle. We freeze the pre-trained SAM's weights to leverage SAM's generalizability while fine-tuning the rest of the model on two public cine CMR datasets. Our model also enables text prompts to specify the view type (short-axis or long-axis) of the input slices and box prompts to guide the segmentation region. We evaluated our model's generalization performance on three external testing datasets including a public multi-center, multi-vendor testing dataset of 136 cases and two retrospectively collected in-house datasets from two different centers with specific pathologies: aortic stenosis (40 cases) and heart failure with preserved ejection fraction (HFpEF) (53 cases). RESULTS:Our approach achieved superior generalization in both the public testing dataset (Dice for LV = 0.94 and for myocardium = 0.86) and two in-house datasets (Dice ≥ 0.90 for LV and ≥ 0.82 for myocardium) compared to existing CMR deep learning segmentation methods. Clinical parameters derived from automatic and manual segmentations showed a strong correlation (r ≥ 0.90). The use of both text prompts and box prompts enhanced the segmentation accuracy. CONCLUSION:cineCMR-SAM effectively adapts SAM for cine CMR segmentation, achieving high generalizability and superior accuracy on unseen datasets.

关键词： CMR Deep learning Segment-anything-model Segmentation

来源：评论

学校读者我要写书评

暂无评论

LLM-guided Decoupled Probabilistic Prompt for Continual Learning in medical Image Diagnosis

引用

IEEE Transactions on medical Imaging 2025年 PP卷 PP页

作者： Luo, Yiwen Li, Wuyang Chen, Cheng Li, Xiang Liu, Tianming Niu, Tianye Yuan, Yixuan City University of Hong Kong Department of Electrical Engineering Hong Kong Chinese University of Hong Kong Department of Electronic Engineering Hong Kong Massachusetts General Hospital Harvard Medical School Center of Advanced Medical Computing and Analysis BostonMA02114 United States University of Georgia School of Computing AthensGA30602 United States University of Science and Technology of China School of Information Science and Technology Hefei230026 China Department of Electronic Engineering Chinese University of Hong Kong Hong Kong CUHK Shenzhen Research Institute Shenzhen518172 China

Deep learning-based traditional diagnostic models typically exhibit limitations when applied to dynamic clinical environments that require handling the emergence of new diseases. Continual learning (CL) offers a promising solution, aiming to learn new knowledge while preserving previously learned knowledge. Though recent rehearsal-free CL methods employing prompt tuning (PT) have shown promise, they rely on deterministic prompts that struggle to handle diverse fine-grained knowledge. Moreover, existing PT methods utilize randomly initialized prompts that are trained under standard classification constraints, impeding expert knowledge integration and optimal performance acquisition. In this paper, we propose an LLM-guided Decoupled Probabilistic Prompt (LDPP) for Continual Learning in medical image diagnosis. Specifically, we develop an Expert Knowledge Generation (EKG) module that leverages LLM to acquire decoupled expert knowledge and comprehensive category descriptions. Then, we introduce a Decoupled Probabilistic Prompt pool (DePP) to construct a shared decoupled probabilistic prompt pool, which constructs a shared prompt pool with probabilistic prompts derived from the expert knowledge set. These prompts dynamically provide diverse and flexible descriptions for input images. Finally, We design a Steering Prompt Pool (SPP) to enhance intra-class compactness and promote model performance by learning nonshared prompts. With extensive experimental validation, LDPP consistently sets state-of-the-art performance under the challenging class-incremental setting in CL. © 1982-2012 IEEE.

关键词： Deep learning

来源：评论

学校读者我要写书评

暂无评论

The Evolution of Image Reconstruction in PET: From Filtered Back-Projection to Artificial Intelligence

引用

PET Clinics 2021年第4期16卷 533-542页

作者： Gong, Kuang Kim, Kyungsang Cui, Jianan Wu, Dufan Li, Quanzheng Department of Radiology Center for Advanced Medical Computing and Analysis Gordon Center for Medical Imaging Massachusetts General Hospital Harvard Medical School Boston MA United States

PET can provide functional images revealing physiologic processes in vivo. Although PET has many applications, there are still some limitations that compromise its precision: the absorption of photons in the body causes signal attenuation; the dead-time limit of system components leads to the loss of the count rate; the scattered and random events received by the detector introduce additional noise; the characteristics of the detector limit the spatial resolution; and the low signal-to-noise ratio caused by the scan-time limit (eg, dynamic scans) and dose concern. The early PET reconstruction methods are analytical approaches based on an idealized mathematical model.

关键词： Artificial intelligence Deep neural network Image reconstruction

来源：评论

学校读者我要写书评

暂无评论

Projection Restoration from Filtered-Backprojection Images of Sparse-View CT

Projection Restoration from Filtered-Backprojection Images o...

引用

2022 IEEE Nuclear Science Symposium, medical Imaging Conference, and Room Temperature Semiconductor Detector Conference, IEEE NSS MIC RTSD 2022

作者： Wu, Dufan Kim, Kyungsang Li, Quanzheng Massachusetts General Hospital Harvard Medical School Center for Advanced Medical Computing and Analysis BostonMA02114 United States

ISBN: (纸本)9781665488723

Projection data has been considered essential to improve the image quality of filtered-backprojection (FBP) results in computed tomography (CT), especially when the projection is corrupted. In recent years, the emerging deep learning (DL)-based algorithms made it possible to improve the FBP image quality without any projection data by training deep neural networks for image-to-image mapping. However, networks that incorporate both projection and image data still generally show better performance compared to image-only networks. Despites of the advantages, projection data are usually hard to acquire for most researchers and costly to store and process. In this work, we propose a novel method to restore the projection data from the FBP images. The method iteratively optimizes the projection such that its FBP reconstruction matches the original, corrupted FBP image. Furthermore, the restored projection can be used to improve the DL-FBP images using iterative reconstruction (IR). Sparse-view CT with parallel beam was simulated from both Forbild head phantom a Mayo Low-dose Challenge dataset to validate the algorithm. Results showed that the proposed method reduced the projection estimation error by 90% compared to forward projection from the sparse-view FBP and 50% compared to that from the DL results. Post-DL IR using the restored projection also improved the image textures and reduced the residual streaking artifacts. © 2022 IEEE.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Uncertainty Prediction for Deep Learning-based Image Denoising in Low-dose CT Imaging

Uncertainty Prediction for Deep Learning-based Image Denoisi...

引用

2021 IEEE Nuclear Science Symposium and medical Imaging Conference, NSS/MIC 2021

作者： Wu, Dufan Xie, Yutong Li, Quanzheng Massachusetts General Hospital Harvard Medical School Center for Advanced Medical Computing and Analysis Gordon Center for Medical Imaging BostonMA02114 United States Peking University Academy for Advanced Interdisciplinary Studies Beijing1000871 China

ISBN: (纸本)9781665421133

Deep learning based low-dose CT image denoising demonstrates good performance, but it remains a question how certain the results are for a trained denoising network. Existing methods to quantify the uncertainty of denoising networks either assume simple noise models or need Monte-Carlo sampling during testing. In this work, we proposed a simple but effective method to directly predict the uncertainty for a given denoising network. An uncertainty prediction network with a basic UNet structure is trained on the training dataset to predict the squared error between the denoising results and label images using L2-norm. Then given a single noisy image and its denoising result, the network will give the expected error between the denoising result and possible clean images. The proposed method does not have any constraints to the structure of the denoising network and is very easy to implement. We validated the efficacy of the proposed method with both simulated and real data. © 2021 IEEE.

关键词： Image denoising

来源：评论

学校读者我要写书评

暂无评论

MR guided PET image denoising based on denoising diffusion probabilistic model and data consistency constraint

MR guided PET image denoising based on denoising diffusion p...

引用

medical Imaging 2023: Physics of medical Imaging

作者： Gong, Kuang Gordon Center for Medical Imaging Massachusetts General Hospital Harvard Medical School BostonMA United States Center for Advanced Medical Computing and Analysis Massachusetts General Hospital Harvard Medical School BostonMA United States

ISBN: (纸本)9781510660311

Due to various physical degradation factors, the signal-to-noise ratio (SNR) and image quality of PET needs further improvements. In this work, we proposed a denoising diffusion probabilistic model-based framework for PET image denoising, where the MR prior image was supplied as the additional network input, and the PET information was included in the iterative refinement steps based on Gaussian distribution assumption. 140 18F-MK- 6240 datasets were used in the evaluation, with 1/4 and 1/8 low-dose levels tested for different methods. Global and regional quantifications show that the proposed framework can outperform the Unet-based denoising and MR-guided nonlocal mean denoising methods. © COPYRIGHT SPIE. Downloading of the abstract is permitted for personal use only.

关键词： Diffusion

来源：评论

学校读者我要写书评

暂无评论

A foundation model for human-AI collaboration in medical literature mining

arXiv

引用

arXiv 2025年

作者： Wang, Zifeng Cao, Lang Jin, Qiao Chan, Joey Wan, Nicholas Afzali, Behdad Cho, Hyun-Jin Choi, Chang-In Emamverdi, Mehdi Gill, Manjot K. Kim, Sun-Hyung Li, Yijia Liu, Yi Ong, Hanley Rousseau, Justin Sheikh, Irfan Wei, Jenny J. Xu, Ziyang Zallek, Christopher M. Kim, Kyungsang Peng, Yifan Lu, Zhiyong Sun, Jimeng School of Computing and Data Science University of Illinois Urbana-Champaign UrbanaIL United States Division of Intramural Research National Library of Medicine National Institutes of Health BethesdaMD United States Kidney Diseases Branch National Institute of Diabetes and Digestive and Kidney Diseases National Institutes of Health BethesdaMD United States Center for Advanced Medical Computing and Analysis Department of Radiology Massachusetts General Hospital Harvard Medical School BostonMA United States National Eye Institute National Institutes of Health BethesdaMD United States Department of Ophthalmology Northwestern University Feinberg School of Medicine ChicagoIL United States Division of Pulmonary and Critical Care Medicine Department of Medicine Chungbuk National University Hospital Chungbuk National University College of Medicine Cheongju Korea Republic of Department of Medicine University of Pittsburgh Medical Center PittsburghPA United States Department of Medicine Weill Cornell Medicine New YorkNY United States Department of Radiology Weill Cornell Medicine New YorkNY United States Department of Neurology UT Southwestern Medical Center DallasTX United States Department of Dermatology University of Washington SeattleWA United States Department of Dermatology NYU Langone Health New YorkNY United States OSF HealthCare Illinois Neurological Institute PeoriaIL United States Department of Population Health Sciences Weill Cornell Medicine New YorkNY United States Carle Illinois College of Medicine University of Illinois Urbana-Champaign UrbanaIL United States

Systematic literature review is essential for evidence-based medicine, requiring comprehensive analysis of clinical trial publications. However, the application of artificial intelligence (AI) models for medical literature mining has been limited by insufficient training and evaluation across broad therapeutic areas and diverse tasks. Here, we present LEADS, an AI foundation model for study search, screening, and data extraction from medical literature. The model is trained on 633,759 instruction data points in LEADSInstruct, curated from 21,335 systematic reviews, 453,625 clinical trial publications, and 27,015 clinical trial registries. We showed that LEADS demonstrates consistent improvements over four cutting-edge generic large language models (LLMs) on six tasks. Furthermore, LEADS enhances expert workflows by providing supportive references following expert requests, streamlining processes while maintaining high-quality results. A study with 16 clinicians and medical researchers from 14 different institutions revealed that experts collaborating with LEADS achieved a recall of 0.81 compared to 0.77 experts working alone in study selection, with a time savings of 22.6%. In data extraction tasks, experts using LEADS achieved an accuracy of 0.85 versus 0.80 without using LEADS, alongside a 26.9% time savings. These findings highlight the potential of specialized medical literature foundation models to outperform generic models, delivering significant quality and efficiency benefits when integrated into expert workflows for medical literature mining. © 2025, CC BY.

关键词： Diagnosis

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：