Medical image analysis最新文献

{"title":"Transfer learning from 2D natural images to 4D fMRI brain images via geometric mapping","authors":"Kai Gao ,&nbsp;Lubin Wang ,&nbsp;Liang Li ,&nbsp;Xiao Chen ,&nbsp;Bin Lu ,&nbsp;Yu-Wei Wang ,&nbsp;Xue-Ying Li ,&nbsp;Zi-Han Wang ,&nbsp;Hui-Xian Li ,&nbsp;Yi-Fan Liao ,&nbsp;Li-Ping Cao ,&nbsp;Guan-Mao Chen ,&nbsp;Jian-Shan Chen ,&nbsp;Tao Chen ,&nbsp;Tao-Lin Chen ,&nbsp;Yan-Rong Chen ,&nbsp;Yu-Qi Cheng ,&nbsp;Zhao-Song Chu ,&nbsp;Shi-Xian Cui ,&nbsp;Xi-Long Cui ,&nbsp;Dewen Hu","doi":"10.1016/j.media.2026.103949","DOIUrl":"10.1016/j.media.2026.103949","url":null,"abstract":"<div><div>Functional magnetic resonance imaging (fMRI) allows real-time observation of brain activity through blood oxygen level-dependent (BOLD) signals and is extensively used in studies related to sex classification, age estimation, behavioral measurements prediction, and mental disorder diagnosis. However, the application of deep learning techniques to brain fMRI analysis is hindered by the small sample size of fMRI datasets. Transfer learning offers a solution to this problem, but most existing approaches are designed for large-scale 2D natural images. The heterogeneity between 4D fMRI data and 2D natural images makes direct model transfer infeasible. This study proposes a novel geometric mapping-based fMRI transfer learning method that enables transfer learning from 2D natural images to 4D fMRI brain images, bridging the transfer learning gap between fMRI data and natural images. The proposed Multi-scale Multi-domain Feature Aggregation (MMFA) module extracts effective aggregated features and reduces the dimensionality of fMRI data to 3D space. By treating the cerebral cortex as a folded Riemannian manifold in 3D space and mapping it into 2D space using surface geometric mapping, we make the transfer learning from 2D natural images to 4D brain images possible. Moreover, the topological relationships of the cerebral cortex are maintained with our method, and calculations are performed along the Riemannian manifold of the brain, effectively addressing signal interference problems. The experimental results based on the Human Connectome Project (HCP) dataset demonstrate the effectiveness of the proposed method. Our method achieved state-of-the-art performance in sex classification, age estimation, and behavioral measurement prediction tasks. Moreover, we propose a cascaded transfer learning approach for depression diagnosis, and proved its effectiveness on 23 depression datasets. In summary, the proposed fMRI transfer learning method, which accounts for the structural characteristics of the brain, is promising for applying transfer learning from natural images to brain fMRI images, significantly enhancing the performance in various fMRI analysis tasks.</div></div>","PeriodicalId":18328,"journal":{"name":"Medical image analysis","volume":"110 ","pages":"Article 103949"},"PeriodicalIF":11.8,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145995243","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"DDTracking: A diffusion model-based deep generative framework with local-global spatiotemporal modeling for diffusion MRI tractography","authors":"Yijie Li ,&nbsp;Wei Zhang ,&nbsp;Xi Zhu ,&nbsp;Ye Wu ,&nbsp;Yogesh Rathi ,&nbsp;Lauren J. O'Donnell ,&nbsp;Fan Zhang","doi":"10.1016/j.media.2026.103967","DOIUrl":"10.1016/j.media.2026.103967","url":null,"abstract":"<div><div>Diffusion MRI (dMRI) tractography is an advanced technique that uniquely enables in vivo mapping of brain fiber pathways. Traditional methods rely on tissue modeling to estimate fiber orientations for streamline propagation, which are computationally intensive and remain sensitive to noise and artifacts. Recent deep learning-based approaches enable data-driven fiber tracking by directly mapping dMRI signals to orientations, demonstrating both improved efficiency and accuracy. However, existing methods typically operate by either leveraging local signal information or learning global dependencies along streamlines. This paper presents DDTracking, a deep generative framework for tractography. One key innovation is the reformulation of streamline propagation as a conditional denoising diffusion process. To the best of our knowledge, this is the first work to apply diffusion models for fiber tracking. Our network architecture incorporates two new designs, including: (1) a dual-pathway encoding scheme that extracts complementary local spatial features and global temporal context, and (2) a conditional diffusion model module that integrates the spatiotemporal features to predict propagation orientations. All components are trained jointly in an end-to-end manner without any pretraining. In this way, DDTracking can capture fine-scale structural details at each point while ensuring long-range consistency across the entire streamline. We conduct a comprehensive evaluation across diverse datasets, including both synthetic and clinical data. Experiments demonstrate that DDTracking outperforms traditional model-based and state-of-the-art deep learning-based methods in terms of tracking accuracy and computational efficiency. Furthermore, our results highlight DDTracking’s high generalizability across heterogeneous datasets, spanning varying health conditions, age groups, imaging protocols, and scanner types. Code is available at: <span><span>https://github.com/yishengpoxiao/DDTracking.git</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":18328,"journal":{"name":"Medical image analysis","volume":"110 ","pages":"Article 103967"},"PeriodicalIF":11.8,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146071490","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"MIL-Adapter: Coupling multiple instance learning and vision-language adapters for few-shot slide-level classification","authors":"Pablo Meseguer ,&nbsp;Rocío del Amor ,&nbsp;Valery Naranjo","doi":"10.1016/j.media.2026.103964","DOIUrl":"10.1016/j.media.2026.103964","url":null,"abstract":"<div><div>Contrastive language-image pretraining has greatly enhanced visual representation learning and enabled zero-shot classification. Vision-language language models (VLM) have succeeded in few-shot learning by leveraging adaptation modules fine-tuned for specific downstream tasks. In computational pathology (CPath), accurate whole-slide image (WSI) prediction is crucial for aiding in cancer diagnosis, and multiple instance learning (MIL) remains essential for managing the gigapixel scale of WSIs. In the intersection between CPath and VLMs, the literature still lacks specific adapters that handle the particular complexity of the slides. To solve this gap, we introduce MIL-Adapter, a novel approach designed to obtain consistent slide-level classification under few-shot learning scenarios. In particular, our framework is the first to combine trainable MIL aggregation functions and lightweight visual-language adapters to improve the performance of histopathological VLMs. MIL-Adapter relies on textual ensemble learning to construct discriminative zero-shot prototypes. It is serves as a solid starting point, surpassing MIL models with randomly initialized classifiers in data-constrained settings. With our experimentation, we demonstrate the value of textual ensemble learning and the robust predictive performance of MIL-Adapter through diverse datasets and configurations of few-shot scenarios, while providing crucial insights on model interpretability. The code is publicly accessible in <span><span>https://github.com/cvblab/MIL-Adapter</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":18328,"journal":{"name":"Medical image analysis","volume":"110 ","pages":"Article 103964"},"PeriodicalIF":11.8,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146048254","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"An efficient, scalable, and adaptable plug-and-play temporal attention module for motion-guided cardiac segmentation with sparse temporal labels","authors":"Md Kamrul Hasan ,&nbsp;Guang Yang ,&nbsp;Choon Hwai Yap","doi":"10.1016/j.media.2026.103981","DOIUrl":"10.1016/j.media.2026.103981","url":null,"abstract":"<div><div>Cardiac anatomy segmentation is essential for clinical assessment of cardiac function and disease diagnosis to inform treatment and intervention. Deep learning (DL) has improved cardiac anatomy segmentation accuracy, especially when information on cardiac motion dynamics is integrated into the networks. Several methods for incorporating motion information have been proposed; however, existing methods are not yet optimal: adding the time dimension to input data causes high computational costs, and incorporating registration into the segmentation network remains computationally costly and can be affected by errors of registration, especially with non-DL registration. While attention-based motion modeling is promising, suboptimal design constrains its capacity to learn the complex and coherent temporal interactions inherent in cardiac image sequences. Here, we propose a novel approach to incorporating motion information in the DL segmentation networks: a computationally efficient yet robust Temporal Attention Module (TAM), modeled as a small, multi-headed, cross-temporal attention module, which can be plug-and-play inserted into a broad range of segmentation networks (CNN, transformer, or hybrid) without a drastic architecture modification. Extensive experiments on multiple cardiac imaging datasets, such as 2D echocardiography (CAMUS and EchoNet-Dynamic), 3D echocardiography (MITEA), and 3D cardiac MRI (ACDC), confirm that TAM consistently improves segmentation performance across datasets when added to a range of networks, including UNet, FCN8s, UNetR, SwinUNetR, and the recent I²UNet and DT-VNet. Integrating TAM into SAM yields a temporal SAM that reduces Hausdorff distance (HD) from 3.99 mm to 3.51 mm on the CAMUS dataset, while integrating TAM into a pre-trained MedSAM reduces HD from 3.04 to 2.06 pixels after fine-tuning on the EchoNet-Dynamic dataset. On the ACDC 3D dataset, our TAM-UNet and TAM-DT-VNet achieve substantial reductions in HD, from 7.97 mm to 4.23 mm and 6.87 mm to 4.74 mm, respectively. Additionally, TAM’s training does not require segmentation of ground truths from all time frames and can be achieved with sparse temporal annotation. TAM is thus a robust, generalizable, and adaptable solution for motion-awareness enhancement that is easily scaled from 2D to 3D. The code is available at <span><span>https://github.com/kamruleee51/TAM</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":18328,"journal":{"name":"Medical image analysis","volume":"110 ","pages":"Article 103981"},"PeriodicalIF":11.8,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146146638","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A hypergraph-based model for tumor prognosis using local and global information fusion on H&E-stained histology images","authors":"Chao Tang ,&nbsp;Jun Liu ,&nbsp;Yanfen Cui ,&nbsp;Zhenhui Li ,&nbsp;Xiuming Zhang ,&nbsp;Su Yao ,&nbsp;Huan Lin ,&nbsp;Dacheng Yang ,&nbsp;Zhishun Liu ,&nbsp;Wei Zhao ,&nbsp;Shiwei Luo ,&nbsp;Ke Zhao ,&nbsp;Yun Zhu ,&nbsp;Guangjun Yang ,&nbsp;Lixu Yan ,&nbsp;Shuting Chen ,&nbsp;Xiangtian Zhao ,&nbsp;Yingqiu Huo ,&nbsp;Zhiyang Chen ,&nbsp;Hongbo Liu ,&nbsp;Cheng Lu","doi":"10.1016/j.media.2026.103991","DOIUrl":"10.1016/j.media.2026.103991","url":null,"abstract":"<div><div>Prognostic variables play a critical role in guiding clinical treatment decisions for cancer patients. However, extracting prognostic information from gigapixel histopathology slides remains a significant challenge. While attention-based deep learning models trained on histologic images have been extensively investigated, existing approaches often fail to effectively model slide-level contextual information or demonstrate generalizability across diverse cancer types and multi-center datasets. We propose a Hypergraph-based Multi-instance Contrastive Reinforcement learning model (HeMiCoRe), which integrates cluster-restricted local features and cross-cluster global representations from 5196 H&amp;E-stained slides across 10 cancer types, leveraging both morphological and spatial relationships. HeMiCoRe employs hypergraph neural networks to predict patient survival outcomes and achieves state-of-the-art (SOTA) performance on 8 cancer types, demonstrating superior generalization compared to existing weakly supervised methods. This framework holds promise for clinical adoption, offering a robust tool for cancer prognosis and supporting treatment decision-making.</div></div>","PeriodicalId":18328,"journal":{"name":"Medical image analysis","volume":"110 ","pages":"Article 103991"},"PeriodicalIF":11.8,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146152675","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Channel-wise joint disentanglement representation learning for B-mode and super-resolution ultrasound based CAD of breast cancer","authors":"Yuhang Zheng ,&nbsp;Jiale Xu ,&nbsp;Qing Hua ,&nbsp;Xiaohong Jia ,&nbsp;Xueqin Hou ,&nbsp;Yanfeng Yao ,&nbsp;Zheng Wei ,&nbsp;Yulu Zhang ,&nbsp;Fanggang Wu ,&nbsp;Wei Guo ,&nbsp;Yuan Tian ,&nbsp;Jun Wang ,&nbsp;Shujun Xia ,&nbsp;Yijie Dong ,&nbsp;Jun Shi ,&nbsp;Jianqiao Zhou","doi":"10.1016/j.media.2026.103957","DOIUrl":"10.1016/j.media.2026.103957","url":null,"abstract":"<div><div>B-mode ultrasound (BUS) is widely used in breast cancer diagnosis, while the emerging super-resolution ultrasound (SRUS) provides microvascular information with high spatial resolution, which has shown great potential in improving breast cancer diagnosis. However, as a new ultrasound modality, its diagnosis remains highly dependent on the clinical experience of sonologists, highlighting the need for reliable computer-aided diagnosis (CAD) approaches. In this work, a novel dual-branch network with a <strong>C</strong>hannel-<strong>W</strong>ise <strong>J</strong>oint <strong>D</strong>isentanglement <strong>R</strong>epresentation <strong>L</strong>earning (CW-JDRL) method is proposed for the multimodal ultrasound-based CAD of breast cancer, where one branch processes BUS and the other analyzes multimodal SRUS data. The CW-JDRL is implemented on the SRUS branch by grouping the final-layer network channels to capture both common and specific properties. It consists of two modules, namely Gradient-guided Disentanglement (GD) module and Gramian-based Contrastive Learning Disentanglement (GCLD) module. The former disentangles with gradient guidance to encourage consistency among common channels and distinctiveness among specific ones, and the latter disentangles common and specific representations by integrating them into a unified contrastive objective. Extensive experiments on a multicenter SRUS dataset demonstrate that the proposed dual-branch network with CW-JDRL achieves superior performance over the compared algorithms and maintains robust generalizability to external data. It suggests not only the effectiveness of SRUS for diagnosis of breast cancer, but also the potential of the proposed CAD model in clinical practice. The codes are publicly available at <span><span>https://github.com/Zyh-AIUltra/CW-JDRL#</span><svg><path></path></svg></span></div></div>","PeriodicalId":18328,"journal":{"name":"Medical image analysis","volume":"110 ","pages":"Article 103957"},"PeriodicalIF":11.8,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146032815","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Physics-informed graph neural networks for flow field estimation in carotid arteries","authors":"Julian Suk ,&nbsp;Dieuwertje Alblas ,&nbsp;Barbara A. Hutten ,&nbsp;Albert Wiegman ,&nbsp;Christoph Brune ,&nbsp;Pim van Ooij ,&nbsp;Jelmer M. Wolterink","doi":"10.1016/j.media.2026.103974","DOIUrl":"10.1016/j.media.2026.103974","url":null,"abstract":"<div><div>Hemodynamic quantities are valuable biomedical risk factors for cardiovascular pathology such as atherosclerosis. Non-invasive, in-vivo measurement of these quantities can only be performed using a select number of modalities that are not widely available, such as 4D flow magnetic resonance imaging (MRI). In this work, we create a surrogate model for hemodynamic flow field estimation, powered by machine learning. We train graph neural networks that include priors about the underlying symmetries and physics, limiting the amount of data required for training. This allows us to train the model using moderately-sized, in-vivo 4D flow MRI datasets, instead of large in-silico datasets obtained by computational fluid dynamics (CFD), as is the current standard. We create an efficient, equivariant neural network by combining the popular PointNet++ architecture with group-steerable layers. To incorporate the physics-informed priors, we derive an efficient discretisation scheme for the involved differential operators. We perform extensive experiments in carotid arteries and show that our model can accurately estimate low-noise hemodynamic flow fields in the carotid artery. Moreover, we show how the learned relation between geometry and hemodynamic quantities transfers to 3D vascular models obtained using a different imaging modality than the training data. This shows that physics-informed graph neural networks can be trained using 4D flow MRI data to estimate blood flow in unseen carotid artery geometries.</div></div>","PeriodicalId":18328,"journal":{"name":"Medical image analysis","volume":"110 ","pages":"Article 103974"},"PeriodicalIF":11.8,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146138282","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"SAM-driven cross prompting with adaptive sampling consistency for semi-supervised medical image segmentation","authors":"Juzheng Miao ,&nbsp;Cheng Chen ,&nbsp;Yuchen Yuan ,&nbsp;Quanzheng Li ,&nbsp;Pheng-Ann Heng","doi":"10.1016/j.media.2026.103973","DOIUrl":"10.1016/j.media.2026.103973","url":null,"abstract":"<div><div>Semi-supervised learning (SSL) has achieved notable progress in medical image segmentation. To achieve effective SSL, a model needs to be able to efficiently learn from limited labeled data and effectively exploit knowledge from abundant unlabeled data. Recent developments in visual foundation models, such as the Segment Anything Model (SAM), have demonstrated remarkable adaptability with improved sample efficiency. To seamlessly harness foundation models in SSL, we propose a SAM-driven cross prompting framework with adaptive sampling and prompt consistency for semi-supervised medical image segmentation, named CPAC-SAM. Our method employs SAM’s unique prompt design and innovates a cross prompting strategy within a dual-branch framework to automatically generate prompts and supervision across two decoder branches, enabling effective learning from both scarce labeled and valuable unlabeled data. To ensure the quality of prompts for unlabeled data and provide meaningful supervision in the cross prompting scheme, we propose an innovative prototype-guided grid sampling strategy with adaptive intervals to simultaneously improve the reliability of the prompt selection area and ensure both adequate prompt density and complete target coverage. We further design a novel prompt consistency regularization to reduce SAM’s prompt sensitivity and to enhance the output invariance under different prompts. We validate our method on five medical image segmentation tasks, encompassing both 2D and 3D scenarios. The extensive experiments with different labeled-data ratios and modalities demonstrate the superiority of our proposed method over the state-of-the-art SSL methods, with more than 4.1% and 3.8% Dice improvement on the breast cancer segmentation task and left atrium segmentation task, respectively. Our code is available at: <span><span>https://github.com/JuzhengMiao/CPAC-SAM</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":18328,"journal":{"name":"Medical image analysis","volume":"110 ","pages":"Article 103973"},"PeriodicalIF":11.8,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146109925","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Incorporating global-local tissue changes to predict future breast cancer from longitudinal screening mammograms","authors":"Xin Wang ,&nbsp;Tao Tan ,&nbsp;Yuan Gao ,&nbsp;Eric Marcus ,&nbsp;Hong-Yu Zhou ,&nbsp;Chunyao Lu ,&nbsp;Luyi Han ,&nbsp;Antonio Portaluri ,&nbsp;Ruisheng Su ,&nbsp;Tianyu Zhang ,&nbsp;Xinglong Liang ,&nbsp;Regina Beets-Tan ,&nbsp;Katja Pinker ,&nbsp;Yue Sun ,&nbsp;Ritse Mann ,&nbsp;Jonas Teuwen","doi":"10.1016/j.media.2026.103990","DOIUrl":"10.1016/j.media.2026.103990","url":null,"abstract":"<div><div>Early detection of breast cancer (BC) through mammography screening is critical for reducing mortality and improving patient outcomes. However, full-population-based, age-driven screening might not lead to optimal resource use and may enlarge screening associated harms in low risk women. Accurate and interpretable BC risk prediction is essential to improve strategies and make screening more personalized. Although recent deep learning models have shown promise in leveraging mammograms for risk stratification, challenges remain in interpretable modeling of temporal changes, efficiently capturing multi-scale risk tissue features from large-scale images, and precise time prediction to enhance clinical interpretability. In this study, we propose <u>T</u>arcking-<u>A</u>ware <u>Brea</u>st <u>C</u>ancer <u>R</u>isk model (TA-BreaCR), a novel framework that integrates local-to-global multiscale longitudinal tissue changes and explicitly models the ordinal relationship of time to BC events, enabling joint prediction of both future BC risk and estimated time to onset. The model is evaluated on two datasets (In-house and EMBED), outperforming existing and state-of-the-art methods in both risk classification and time-to-event prediction tasks. Visualization analysis reveals consistent attention to high-risk regions over time, enhancing interpretability. These results highlight the potential of TA-BreaCR to support individualized BC screening and prevention.</div></div>","PeriodicalId":18328,"journal":{"name":"Medical image analysis","volume":"110 ","pages":"Article 103990"},"PeriodicalIF":11.8,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146209647","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"4D monocular surgical reconstruction under arbitrary camera motions","authors":"Jiwei Shan ,&nbsp;Zeyu Cai ,&nbsp;Cheng-Tai Hsieh ,&nbsp;Yirui Li ,&nbsp;Hao Liu ,&nbsp;Lijun Han ,&nbsp;Hesheng Wang ,&nbsp;Shing Shin Cheng","doi":"10.1016/j.media.2026.103989","DOIUrl":"10.1016/j.media.2026.103989","url":null,"abstract":"<div><div>Reconstructing deformable surgical scenes from endoscopic videos is a challenging task with important clinical applications. Recent state-of-the-art approaches, such as those based on implicit neural representations or 3D Gaussian splatting, have made notable progress in this area. However, most existing methods are designed for deformable scenes with fixed endoscope viewpoints and rely on stereo depth priors or accurate structure-from-motion for both initialization and optimization. This limits their ability to handle monocular sequences with large camera movements, restricting their use in real clinical settings. To address these limitations, we propose Local-EndoGS, a high-quality 4D reconstruction framework for monocular endoscopic sequences with arbitrary camera motion. Local-EndoGS introduces a progressive, window-based global scene representation that allocates local deformable scene representations for each observed window, enabling scalability to long sequences with substantial camera movement. To overcome unreliable initialization due to the lack of stereo depth or accurate structure-from-motion, we propose a coarse-to-fine initialization strategy that integrates multi-view geometry, cross-window information, and monocular depth priors, providing a robust foundation for subsequent optimization. In addition, we incorporate long-range 2D pixel trajectory constraints and physical motion priors to improve the physical plausibility of the recovered deformations. We comprehensively evaluate Local-EndoGS on three public endoscopic datasets with deformable scenes and varying camera motions. Local-EndoGS achieves superior performance in both appearance quality and geometry, consistently outperforming state-of-the-art methods. Extensive ablation studies further validate the effectiveness of our key designs. Our code will be released upon acceptance at <span><span>https://github.com/IRMVLab/Local-EndoGS</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":18328,"journal":{"name":"Medical image analysis","volume":"110 ","pages":"Article 103989"},"PeriodicalIF":11.8,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146160397","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}