Virtual coronary intervention planning (VCIP) aims to optimize the hemodynamic outcomes of percutaneous coronary intervention (PCI) in patients with coronary stenosis. However, its clinical adoption remains constrained by the computational burden associated with evaluating numerous combinatorial intervention strategies, leading to time-consuming workflows and potentially suboptimal decisions in the catheterization laboratory. While conventional deep reinforcement learning (DRL) offers a path to automated VCIP, it often explores state-action-reward space inefficiently. In this study, we propose an Informed-Exploration Reinforcement Learning framework that concentrates the search on clinically meaningful interventions by integrating historical intervention experience with patient-specific anatomical and physiological information to guide the generation of functionally informed stent strategies. Extensive experiments on 172 vessels from 146 patients show that IERL achieves high agreement (r = 0.815) with real interventions and excellent computational efficiency with an average run time of 2.1 seconds. By aligning exploration with both prior experience and patient context, IERL provides objective, reproducible, and near-real-time VCIP decision support, enabling timely and interpretable recommendations compatible with catheterization workflows. The code and models are available at: https://github.com/HIC-SYSU/IERL/tree/main.
{"title":"Informed-Exploration Reinforcement Learning for Automated Virtual Coronary Intervention Planning.","authors":"Anbang Wang,Ming Lei,Heye Zhang,Zhifan Gao,Qi Zhang,Zhihui Zhang,Ping Zhu,Dan Deng,Lingyun Zu,Guang Yang,Xiujian Liu","doi":"10.1109/tmi.2026.3707748","DOIUrl":"https://doi.org/10.1109/tmi.2026.3707748","url":null,"abstract":"Virtual coronary intervention planning (VCIP) aims to optimize the hemodynamic outcomes of percutaneous coronary intervention (PCI) in patients with coronary stenosis. However, its clinical adoption remains constrained by the computational burden associated with evaluating numerous combinatorial intervention strategies, leading to time-consuming workflows and potentially suboptimal decisions in the catheterization laboratory. While conventional deep reinforcement learning (DRL) offers a path to automated VCIP, it often explores state-action-reward space inefficiently. In this study, we propose an Informed-Exploration Reinforcement Learning framework that concentrates the search on clinically meaningful interventions by integrating historical intervention experience with patient-specific anatomical and physiological information to guide the generation of functionally informed stent strategies. Extensive experiments on 172 vessels from 146 patients show that IERL achieves high agreement (r = 0.815) with real interventions and excellent computational efficiency with an average run time of 2.1 seconds. By aligning exploration with both prior experience and patient context, IERL provides objective, reproducible, and near-real-time VCIP decision support, enabling timely and interpretable recommendations compatible with catheterization workflows. The code and models are available at: https://github.com/HIC-SYSU/IERL/tree/main.","PeriodicalId":13418,"journal":{"name":"IEEE Transactions on Medical Imaging","volume":"52 1","pages":""},"PeriodicalIF":10.6,"publicationDate":"2026-06-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148335904","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}
Pub Date : 2026-06-25DOI: 10.1109/tmi.2026.3707404
Xu Wang,Shuai Zhang,Baoru Huang,Jialang Xu,Danail Stoyanov,Evangelos B Mazomenos
Reconstructing dynamic surgical scenes from endoscopic videos remains a fundamental challenge in robot-assisted surgery. Existing methods primarily focus on deformable tissues, overlooking the presence of articulated instruments. To bridge this gap, we present EndoLRMGS, the first unified framework capable of reconstructing both deformable tissue and articulated instruments in a modular approach from monocular video and depth priors. We introduce Frequency-Modulated Gaussian Splatting (FMGS), for deformable tissue reconstruction, which modulates the spatial frequency of Gaussian primitives according to the Nyquist-Shannon sampling theorem, improving the visual fidelity while maintaining robust geometric accuracy. For instrument reconstruction, we leverage the Large Reconstruction Model (LRM) to generate high-quality, watertight 3D models from single images, and introduce a novel Orthographic and Perspective joint Projection Optimization (OPjPO) module to recover metric scale and spatial alignment. Extensive experiments on public datasets demonstrate the effectiveness of EndoLRMGS, achieving PSNR values from 28.4981 to 38.4179, with Chamfer distance ranging from 1.43 to 4.71 mm in tissue reconstruction. For instrument reconstruction, EndoLRMGS achieves PSNR values ranging from 19.7341 to 22.4393 on left views and 17.7442 to 20.8436 on right views. In terms of spatial alignment accuracy, it attains IoU values between 71.56% and 85.82%, with Chamfer distance ranging from 12.59 to 17.38 mm. These results highlight EndoLRMGS as a powerful and versatile solution for accurate, complete, and photorealistic 3D reconstruction of surgical scenes. Code is available at: EndoLRMGS.
{"title":"EndoLRMGS: Combining Large Reconstruction Modelling and Gaussian Splatting for Complete Endoscopic Scene Reconstruction.","authors":"Xu Wang,Shuai Zhang,Baoru Huang,Jialang Xu,Danail Stoyanov,Evangelos B Mazomenos","doi":"10.1109/tmi.2026.3707404","DOIUrl":"https://doi.org/10.1109/tmi.2026.3707404","url":null,"abstract":"Reconstructing dynamic surgical scenes from endoscopic videos remains a fundamental challenge in robot-assisted surgery. Existing methods primarily focus on deformable tissues, overlooking the presence of articulated instruments. To bridge this gap, we present EndoLRMGS, the first unified framework capable of reconstructing both deformable tissue and articulated instruments in a modular approach from monocular video and depth priors. We introduce Frequency-Modulated Gaussian Splatting (FMGS), for deformable tissue reconstruction, which modulates the spatial frequency of Gaussian primitives according to the Nyquist-Shannon sampling theorem, improving the visual fidelity while maintaining robust geometric accuracy. For instrument reconstruction, we leverage the Large Reconstruction Model (LRM) to generate high-quality, watertight 3D models from single images, and introduce a novel Orthographic and Perspective joint Projection Optimization (OPjPO) module to recover metric scale and spatial alignment. Extensive experiments on public datasets demonstrate the effectiveness of EndoLRMGS, achieving PSNR values from 28.4981 to 38.4179, with Chamfer distance ranging from 1.43 to 4.71 mm in tissue reconstruction. For instrument reconstruction, EndoLRMGS achieves PSNR values ranging from 19.7341 to 22.4393 on left views and 17.7442 to 20.8436 on right views. In terms of spatial alignment accuracy, it attains IoU values between 71.56% and 85.82%, with Chamfer distance ranging from 12.59 to 17.38 mm. These results highlight EndoLRMGS as a powerful and versatile solution for accurate, complete, and photorealistic 3D reconstruction of surgical scenes. Code is available at: EndoLRMGS.","PeriodicalId":13418,"journal":{"name":"IEEE Transactions on Medical Imaging","volume":"106 1","pages":""},"PeriodicalIF":10.6,"publicationDate":"2026-06-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148322677","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}
Pub Date : 2026-06-23DOI: 10.1109/tmi.2026.3706567
Jinyue Guo,Yanchao Zhang,Hao Zhai,Yi Jiang,Qi Zhang,Yunfeng Hua,Jing Liu,Hua Han
Volume electron microscopy (vEM) has revolutionized the nanoscale reconstruction of synapses in neural circuits. However, large-scale vEM techniques relying on serial sectioning suffer from severe anisotropy, where axial resolution is far worse than lateral resolution. This anisotropic imaging induces discontinuities in biological architectures across 3D space, compromising reconstruction accuracy and instance segmentation of synapses. Although synapse reconstruction can be realized via aggregation of segmented voxels or detected superpixels, conventional semantic and instance-level models fail to learn voxel instance attributes robustly from strong anisotropic datasets. Here, we present SynReEM, a dedicated framework for synapse reconstruction. Specifically, we first conduct structural encoding on synapse annotations to optimize structural components, making instance segmentation feasible within a semantic context. Then, we incorporate biological priors to impose continuity and inclusion constraints on model outputs, leveraging online pseudo-labels to enhance model convergence. Furthermore, we design a dual-headed branch for simultaneous semantic and instance decoding from shared feature maps, fuse the multi-task outputs, and adopt the watershed algorithm to achieve accurate instance reconstruction. Comprehensive evaluations on three vEM datasets containing synapses (Synapse178, AC3/AC4, and SynWTAD) consistently confirm the superior performance of our proposed SynReEM method.
{"title":"SynReEM: Synapse Reconstruction via Instance Structure Encoding in Anisotropic Electron Microscopic Volumes.","authors":"Jinyue Guo,Yanchao Zhang,Hao Zhai,Yi Jiang,Qi Zhang,Yunfeng Hua,Jing Liu,Hua Han","doi":"10.1109/tmi.2026.3706567","DOIUrl":"https://doi.org/10.1109/tmi.2026.3706567","url":null,"abstract":"Volume electron microscopy (vEM) has revolutionized the nanoscale reconstruction of synapses in neural circuits. However, large-scale vEM techniques relying on serial sectioning suffer from severe anisotropy, where axial resolution is far worse than lateral resolution. This anisotropic imaging induces discontinuities in biological architectures across 3D space, compromising reconstruction accuracy and instance segmentation of synapses. Although synapse reconstruction can be realized via aggregation of segmented voxels or detected superpixels, conventional semantic and instance-level models fail to learn voxel instance attributes robustly from strong anisotropic datasets. Here, we present SynReEM, a dedicated framework for synapse reconstruction. Specifically, we first conduct structural encoding on synapse annotations to optimize structural components, making instance segmentation feasible within a semantic context. Then, we incorporate biological priors to impose continuity and inclusion constraints on model outputs, leveraging online pseudo-labels to enhance model convergence. Furthermore, we design a dual-headed branch for simultaneous semantic and instance decoding from shared feature maps, fuse the multi-task outputs, and adopt the watershed algorithm to achieve accurate instance reconstruction. Comprehensive evaluations on three vEM datasets containing synapses (Synapse178, AC3/AC4, and SynWTAD) consistently confirm the superior performance of our proposed SynReEM method.","PeriodicalId":13418,"journal":{"name":"IEEE Transactions on Medical Imaging","volume":"22 1","pages":""},"PeriodicalIF":10.6,"publicationDate":"2026-06-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148304878","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}
Pub Date : 2025-12-11DOI: 10.1109/tmi.2025.3642381
Jiaqi Zhang,Xiuzhe Wu,Jiahui Liu,Chunyu Zou,Fengze Nie,Zicheng Sun,Xiaojuan Qi,Jiang Liu
High-fidelity reconstruction of the Posterior Eyeball Shape (PES) is crucial for early diagnosis and timely intervention of sight-threatening diseases such as high myopia, diabetic retinopathy, and glaucoma. However, existing magnetic resonance imaging (MRI)- and optical coherence tomography (OCT)-based methods either provide only coarse scleral geometry or suffer from suboptimal PES representations due to limited field of view (FOV) and detail loss, hindering accurate assessment of intact retinal pigment epithelium (RPE) abnormalities. In this study, we propose the Polar Subarea-Aware Fusion Net (PSAFNet), a novel end-to-end framework that reconstructs complete and high-fidelity PES directly from a single local OCT scan, even under clinically common settings with only 6.25% FOV. To avoid information loss, we reformulate PES reconstruction as a 2D dense regression task and introduce the Ocular Shape Map (OSM), an innovative lossless 2D representation that encodes 3D coordinate attributes into corresponding image channels. PSAFNet then leverages three dedicated modules-Subarea Feature Embedding Module (SFEM), Channel- and Patch-wise Fusion Blocks (CFB/PFB), and Reassemble and Up-sample Module (RUM)-to enhance positional awareness, integrate local-global features, and achieve high-resolution OSM prediction. Furthermore, we construct two large-scale datasets, POSDiag and PESGen, comprising 794 ultra-widefield OCT scans from diverse health conditions and imaging devices, providing a comprehensive benchmark for PES reconstruction. Extensive experiments demonstrate that PSAFNet consistently outperforms existing methods (e.g., EMD=5.58, AAL=97.3%) and exhibits strong clinical relevance, validated by superior performance in downstream disease classification and ophthalmologist evaluations (Expert-Score=82.78%). The source code of the proposed PSAFNet is released at https://github.com/HKUZJ77/PSAFNet.
{"title":"Polar Subarea-Aware Fusion Net for Posterior Eyeball Shape Reconstruction.","authors":"Jiaqi Zhang,Xiuzhe Wu,Jiahui Liu,Chunyu Zou,Fengze Nie,Zicheng Sun,Xiaojuan Qi,Jiang Liu","doi":"10.1109/tmi.2025.3642381","DOIUrl":"https://doi.org/10.1109/tmi.2025.3642381","url":null,"abstract":"High-fidelity reconstruction of the Posterior Eyeball Shape (PES) is crucial for early diagnosis and timely intervention of sight-threatening diseases such as high myopia, diabetic retinopathy, and glaucoma. However, existing magnetic resonance imaging (MRI)- and optical coherence tomography (OCT)-based methods either provide only coarse scleral geometry or suffer from suboptimal PES representations due to limited field of view (FOV) and detail loss, hindering accurate assessment of intact retinal pigment epithelium (RPE) abnormalities. In this study, we propose the Polar Subarea-Aware Fusion Net (PSAFNet), a novel end-to-end framework that reconstructs complete and high-fidelity PES directly from a single local OCT scan, even under clinically common settings with only 6.25% FOV. To avoid information loss, we reformulate PES reconstruction as a 2D dense regression task and introduce the Ocular Shape Map (OSM), an innovative lossless 2D representation that encodes 3D coordinate attributes into corresponding image channels. PSAFNet then leverages three dedicated modules-Subarea Feature Embedding Module (SFEM), Channel- and Patch-wise Fusion Blocks (CFB/PFB), and Reassemble and Up-sample Module (RUM)-to enhance positional awareness, integrate local-global features, and achieve high-resolution OSM prediction. Furthermore, we construct two large-scale datasets, POSDiag and PESGen, comprising 794 ultra-widefield OCT scans from diverse health conditions and imaging devices, providing a comprehensive benchmark for PES reconstruction. Extensive experiments demonstrate that PSAFNet consistently outperforms existing methods (e.g., EMD=5.58, AAL=97.3%) and exhibits strong clinical relevance, validated by superior performance in downstream disease classification and ophthalmologist evaluations (Expert-Score=82.78%). The source code of the proposed PSAFNet is released at https://github.com/HKUZJ77/PSAFNet.","PeriodicalId":13418,"journal":{"name":"IEEE Transactions on Medical Imaging","volume":"38 1","pages":""},"PeriodicalIF":10.6,"publicationDate":"2025-12-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145728474","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}
Pub Date : 2025-07-02DOI: 10.1109/tmi.2025.3584857
Yuxuan Sun, Hao Wu, Chenglu Zhu, Yixuan Si, Qizi Chen, Yunlong Zhang, Kai Zhang, Jingxiong Li, Jiatong Cai, Yuhan Wang, Lin Sun, Tao Lin, Lin Yang
{"title":"PathBench: Advancing the Benchmark of Large Multimodal Models for Pathology Image Understanding at Patch and Whole Slide Level","authors":"Yuxuan Sun, Hao Wu, Chenglu Zhu, Yixuan Si, Qizi Chen, Yunlong Zhang, Kai Zhang, Jingxiong Li, Jiatong Cai, Yuhan Wang, Lin Sun, Tao Lin, Lin Yang","doi":"10.1109/tmi.2025.3584857","DOIUrl":"https://doi.org/10.1109/tmi.2025.3584857","url":null,"abstract":"","PeriodicalId":13418,"journal":{"name":"IEEE Transactions on Medical Imaging","volume":"50 1","pages":""},"PeriodicalIF":10.6,"publicationDate":"2025-07-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144546999","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}
Pub Date : 2025-04-21DOI: 10.1109/tmi.2025.3563081
Dunyuan Xu, Xi Wang, Jinpeng Li, Jingyang Zhang, Pheng-Ann Heng
{"title":"Towards Synchronous Memorizability and Generalizability with Site-Modulated Diffusion Replay for Cross-Site Continual Segmentation","authors":"Dunyuan Xu, Xi Wang, Jinpeng Li, Jingyang Zhang, Pheng-Ann Heng","doi":"10.1109/tmi.2025.3563081","DOIUrl":"https://doi.org/10.1109/tmi.2025.3563081","url":null,"abstract":"","PeriodicalId":13418,"journal":{"name":"IEEE Transactions on Medical Imaging","volume":"15 1","pages":""},"PeriodicalIF":10.6,"publicationDate":"2025-04-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143857709","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":"Development-driven Diffusion Model for Longitudinal Prediction of Fetal Brain MRI with Unpaired Data","authors":"Kai Zhang, Geng Chen, Shijie Huang, Fangmei Zhu, Zhongxiang Ding, Dinggang Shen","doi":"10.1109/tmi.2024.3496860","DOIUrl":"https://doi.org/10.1109/tmi.2024.3496860","url":null,"abstract":"","PeriodicalId":13418,"journal":{"name":"IEEE Transactions on Medical Imaging","volume":"62 1","pages":""},"PeriodicalIF":10.6,"publicationDate":"2024-11-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142610631","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}
Pub Date : 2024-10-17DOI: 10.1109/tmi.2024.3482228
Xiaowei Yu, Lu Zhang, Zihao Wu, Dajiang Zhu
{"title":"Core-Periphery Multi-Modality Feature Alignment for Zero-Shot Medical Image Analysis","authors":"Xiaowei Yu, Lu Zhang, Zihao Wu, Dajiang Zhu","doi":"10.1109/tmi.2024.3482228","DOIUrl":"https://doi.org/10.1109/tmi.2024.3482228","url":null,"abstract":"","PeriodicalId":13418,"journal":{"name":"IEEE Transactions on Medical Imaging","volume":"58 1","pages":""},"PeriodicalIF":10.6,"publicationDate":"2024-10-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142448932","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}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：