Pub Date : 2026-03-10DOI: 10.1109/tmi.2026.3672432
Lin Teng,Shen Zhao,Jiadong Zhang,Feng Shi,Dinggang Shen
Accurate and automated reconstruction of cortical surfaces across the human lifespan is essential for studying brain development, aging, and the early diagnosis of neurological disorders. However, traditional neuroimaging pipelines require hours per subject, limiting scalability. Existing deep learning methods typically target narrow age ranges, struggling to generalize due to substantial age-related anatomical variability. This leads to inaccurate quantification of cortical properties, such as curvature and cortical thickness, thereby undermining their potential as reliable biomarkers for routine clinical brain analysis. To address these challenges, we present uBrainSurf, a unified curvature-aware deformation framework for lifespan cortical surface reconstruction. Specifically, uBrainSurf learns a sequence of stationary velocity fields (SVFs) from volumetric MR images, gradually deforming a smooth template mesh to subject-specific white-matter and pial surfaces through a coarse-to-fine strategy. To enhance the reconstruction accuracy, we introduce an auxiliary curvature prediction branch that provides an anatomical prior, guiding the model to prioritize anatomically important regions. Furthermore, we propose a novel curvature-driven loss function that encourages consistency between the curvatures of corresponding points on predicted and target surfaces, ensuring the reconstructed surfaces are directly suitable for downstream analyses. The uBrainSurf is evaluated on a large-scale brain dataset comprising 2,132 subjects spanning 0-100 years. Experimental results demonstrate that uBrainSurf achieves superior performance and generalizability while being several orders of magnitude faster than traditional pipelines. Our code is available at https://github.com/TL9792/CCF.
{"title":"uBrainSurf: Unified Curvature-aware Deformation Framework for Lifespan Brain Cortical Surface Reconstruction.","authors":"Lin Teng,Shen Zhao,Jiadong Zhang,Feng Shi,Dinggang Shen","doi":"10.1109/tmi.2026.3672432","DOIUrl":"https://doi.org/10.1109/tmi.2026.3672432","url":null,"abstract":"Accurate and automated reconstruction of cortical surfaces across the human lifespan is essential for studying brain development, aging, and the early diagnosis of neurological disorders. However, traditional neuroimaging pipelines require hours per subject, limiting scalability. Existing deep learning methods typically target narrow age ranges, struggling to generalize due to substantial age-related anatomical variability. This leads to inaccurate quantification of cortical properties, such as curvature and cortical thickness, thereby undermining their potential as reliable biomarkers for routine clinical brain analysis. To address these challenges, we present uBrainSurf, a unified curvature-aware deformation framework for lifespan cortical surface reconstruction. Specifically, uBrainSurf learns a sequence of stationary velocity fields (SVFs) from volumetric MR images, gradually deforming a smooth template mesh to subject-specific white-matter and pial surfaces through a coarse-to-fine strategy. To enhance the reconstruction accuracy, we introduce an auxiliary curvature prediction branch that provides an anatomical prior, guiding the model to prioritize anatomically important regions. Furthermore, we propose a novel curvature-driven loss function that encourages consistency between the curvatures of corresponding points on predicted and target surfaces, ensuring the reconstructed surfaces are directly suitable for downstream analyses. The uBrainSurf is evaluated on a large-scale brain dataset comprising 2,132 subjects spanning 0-100 years. Experimental results demonstrate that uBrainSurf achieves superior performance and generalizability while being several orders of magnitude faster than traditional pipelines. Our code is available at https://github.com/TL9792/CCF.","PeriodicalId":13418,"journal":{"name":"IEEE Transactions on Medical Imaging","volume":"195 1","pages":""},"PeriodicalIF":10.6,"publicationDate":"2026-03-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147383478","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 : 2026-03-10DOI: 10.1109/tmi.2026.3672477
Zhaocan Yang,Yan Li,Yang Liu
Retinal diseases (RD) are major causes of global vision impairment. Automated diagnosis using fundus images has significant clinical value, particularly in multi-label classification of RD. Recently, hierarchy-aware methods have shown potential in improving classification performance by leveraging hierarchical relationships among disease categories. However, implicit hierarchy-aware methods often fail to capture complex semantic relationships required to make multi-level predictions, while explicit hierarchy-aware methods fail to maintain hierarchy level consistency. Additionally, existing approaches do not sufficiently integrate knowledge from expert domains. Accordingly, in this paper, we introduce a novel framework, namely Hierarchy-Aware and Knowledge-Guided Learning (HAKGL), for diagnosing RD from fundus images. It establishes complex relationships among diseases by employing a hierarchical Transformer for making multi-level predictions by maximally exploiting the visual information. Besides, we utilize feature similarities to establish correlations among hierarchy levels, which offer additional supervision signals to align hierarchical feature representations. This strategy explicitly maintains hierarchical consistency, thereby improving the performance of the model. Furthermore, we propose a correlation learning strategy for aligning image correlations with expert textual knowledge extracted from retinal foundation models, thus enabling the model to learn more generalizable representations. The superiority of the proposed HAKGL approach has been validated through extensive experiments in multi-label classification of RD. Code is available at https://github.com/YZC-99/HAKGL.
{"title":"Hierarchy-Aware and Knowledge-Guided Learning for Multi-Label Classification of Retinal Diseases from Fundus Images.","authors":"Zhaocan Yang,Yan Li,Yang Liu","doi":"10.1109/tmi.2026.3672477","DOIUrl":"https://doi.org/10.1109/tmi.2026.3672477","url":null,"abstract":"Retinal diseases (RD) are major causes of global vision impairment. Automated diagnosis using fundus images has significant clinical value, particularly in multi-label classification of RD. Recently, hierarchy-aware methods have shown potential in improving classification performance by leveraging hierarchical relationships among disease categories. However, implicit hierarchy-aware methods often fail to capture complex semantic relationships required to make multi-level predictions, while explicit hierarchy-aware methods fail to maintain hierarchy level consistency. Additionally, existing approaches do not sufficiently integrate knowledge from expert domains. Accordingly, in this paper, we introduce a novel framework, namely Hierarchy-Aware and Knowledge-Guided Learning (HAKGL), for diagnosing RD from fundus images. It establishes complex relationships among diseases by employing a hierarchical Transformer for making multi-level predictions by maximally exploiting the visual information. Besides, we utilize feature similarities to establish correlations among hierarchy levels, which offer additional supervision signals to align hierarchical feature representations. This strategy explicitly maintains hierarchical consistency, thereby improving the performance of the model. Furthermore, we propose a correlation learning strategy for aligning image correlations with expert textual knowledge extracted from retinal foundation models, thus enabling the model to learn more generalizable representations. The superiority of the proposed HAKGL approach has been validated through extensive experiments in multi-label classification of RD. Code is available at https://github.com/YZC-99/HAKGL.","PeriodicalId":13418,"journal":{"name":"IEEE Transactions on Medical Imaging","volume":"1 1","pages":""},"PeriodicalIF":10.6,"publicationDate":"2026-03-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147383477","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 : 2026-03-09DOI: 10.1109/tmi.2026.3672198
Josué Ruano-Balseca, Diego Bravo, Diana L. Giraldo, Martín Gómez, Fabio A. Gómez, Eduardo Romero
{"title":"Spatio-temporal characterization of gastric distensibility in upper endoscopy identifies the presence of Helicobacter pylori","authors":"Josué Ruano-Balseca, Diego Bravo, Diana L. Giraldo, Martín Gómez, Fabio A. Gómez, Eduardo Romero","doi":"10.1109/tmi.2026.3672198","DOIUrl":"https://doi.org/10.1109/tmi.2026.3672198","url":null,"abstract":"","PeriodicalId":13418,"journal":{"name":"IEEE Transactions on Medical Imaging","volume":"40 1","pages":""},"PeriodicalIF":10.6,"publicationDate":"2026-03-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147380687","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 : 2026-03-09DOI: 10.1109/tmi.2026.3671423
Ziang Xu, Bin Li, Yang Hu, Chenyu Zhang, James East, Sharib Ali, Jens Rittscher
{"title":"Self-supervised Monocular Depth and Pose Estimation for Endoscopy with Latent Priors","authors":"Ziang Xu, Bin Li, Yang Hu, Chenyu Zhang, James East, Sharib Ali, Jens Rittscher","doi":"10.1109/tmi.2026.3671423","DOIUrl":"https://doi.org/10.1109/tmi.2026.3671423","url":null,"abstract":"","PeriodicalId":13418,"journal":{"name":"IEEE Transactions on Medical Imaging","volume":"25 1","pages":""},"PeriodicalIF":10.6,"publicationDate":"2026-03-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147380688","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 : 2026-03-06DOI: 10.1109/tmi.2026.3671287
Wenwen Zhang,Zhenyu Tang,Hao Zhang,Shaohao Rui,Z Jane Wang,Xiaosong Wang
Federated learning (FL) enables collaborative model training across decentralized medical datasets while preserving data privacy. Its practical adoption remains limited due to data heterogeneity, specifically, differences in input imaging modality (e.g., CT or MRI) and client task (e.g., segmentation or classification) across participating institutions (clients). Such data heterogeneity poses significant challenges for jointly learning a unified global model that generalizes across clients with different input modality and task. To address this, we propose FedCMT, a modality-agnostic FL framework that adaptively aggregates heterogeneous client models. FedCMT supports flexible input modalities and diverse local tasks by incorporating group-wise adapters and personalized decoders that capture modality- and task-specific features. To enhance collaboration across clients, FedCMT employs a conflict-averse module that extracts modality-invariant representations and mitigates inter-client feature conflicts. FedCMT also integrates a global-to-local knowledge distillation mechanism to balance global consistency and local specialization. The proposed FedCMT maintains stability while fostering shared knowledge in diverse medical imaging modalities. We evaluate FedCMT on ten CT and MR datasets involving up to eight federated clients performing segmentation or classification tasks. Experimental results show that FedCMT consistently outperforms state-of-the-art FL baselines, yielding an average improvement of 4.76% over state-of-the-art methods and 4.01% over standalone training. These results demonstrate FedCMT as a promising adaptable FL for real-world medical image analysis.
{"title":"Modality-Agnostic Federated Learning with Adaptive Updates for Heterogeneous Medical Image Tasks.","authors":"Wenwen Zhang,Zhenyu Tang,Hao Zhang,Shaohao Rui,Z Jane Wang,Xiaosong Wang","doi":"10.1109/tmi.2026.3671287","DOIUrl":"https://doi.org/10.1109/tmi.2026.3671287","url":null,"abstract":"Federated learning (FL) enables collaborative model training across decentralized medical datasets while preserving data privacy. Its practical adoption remains limited due to data heterogeneity, specifically, differences in input imaging modality (e.g., CT or MRI) and client task (e.g., segmentation or classification) across participating institutions (clients). Such data heterogeneity poses significant challenges for jointly learning a unified global model that generalizes across clients with different input modality and task. To address this, we propose FedCMT, a modality-agnostic FL framework that adaptively aggregates heterogeneous client models. FedCMT supports flexible input modalities and diverse local tasks by incorporating group-wise adapters and personalized decoders that capture modality- and task-specific features. To enhance collaboration across clients, FedCMT employs a conflict-averse module that extracts modality-invariant representations and mitigates inter-client feature conflicts. FedCMT also integrates a global-to-local knowledge distillation mechanism to balance global consistency and local specialization. The proposed FedCMT maintains stability while fostering shared knowledge in diverse medical imaging modalities. We evaluate FedCMT on ten CT and MR datasets involving up to eight federated clients performing segmentation or classification tasks. Experimental results show that FedCMT consistently outperforms state-of-the-art FL baselines, yielding an average improvement of 4.76% over state-of-the-art methods and 4.01% over standalone training. These results demonstrate FedCMT as a promising adaptable FL for real-world medical image analysis.","PeriodicalId":13418,"journal":{"name":"IEEE Transactions on Medical Imaging","volume":"72 1","pages":""},"PeriodicalIF":10.6,"publicationDate":"2026-03-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147368238","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}
Magnetic Particle Imaging (MPI) enables noninvasive temperature imaging without depth limitations. However, due to the lack of effective calibration strategies that can simultaneously address issues such as calibration infeasibility and environmental mismatch, its practical in vivo application remains challenging. In this work, we propose a novel in vivo temperature imaging method based on a dual-mode magnetic particle spectroscopy/magnetic particle imaging (MPS/MPI) system. First, MPS is employed to capture the differences in harmonic phase responses of magnetic nanoparticles (MNPs) under in vivo and in vitro conditions, thereby enabling the construction of calibration functions that are consistent with the in vivo environment. Second, an MLP based calibration strategy is proposed, which accounts for non-ideal deviations from the approximately linear temperature-phase relationship and integrates multi-parameter information into a unified network, thereby enabling accurate and stable temperature mapping. Comprehensive simulation, in vitro, and in vivo experiments demonstrate that, compared with conventional phantom-based temperature mapping methods, the proposed method reduces the in vivo temperature reconstruction error by approximately 17.24% and achieves an average absolute temperature error below 1.257 °C. These results verify the feasibility of accurate in vivo temperature imaging using MPI and provide essential technical support for temperature-sensitive applications, including magnetic hyperthermia.
{"title":"Phase-lag Based MPS/MPI Dual-mode Precise in vivo Temperature Imaging Technique.","authors":"Siao Lei,Wenxuan Zou,Yanjun Liu,Guanghui Li,Gen Shi,Jiaqian Li,Jie He,Guangxing Zhou,Yang Jing,Yu An,Jie Tian","doi":"10.1109/tmi.2026.3670844","DOIUrl":"https://doi.org/10.1109/tmi.2026.3670844","url":null,"abstract":"Magnetic Particle Imaging (MPI) enables noninvasive temperature imaging without depth limitations. However, due to the lack of effective calibration strategies that can simultaneously address issues such as calibration infeasibility and environmental mismatch, its practical in vivo application remains challenging. In this work, we propose a novel in vivo temperature imaging method based on a dual-mode magnetic particle spectroscopy/magnetic particle imaging (MPS/MPI) system. First, MPS is employed to capture the differences in harmonic phase responses of magnetic nanoparticles (MNPs) under in vivo and in vitro conditions, thereby enabling the construction of calibration functions that are consistent with the in vivo environment. Second, an MLP based calibration strategy is proposed, which accounts for non-ideal deviations from the approximately linear temperature-phase relationship and integrates multi-parameter information into a unified network, thereby enabling accurate and stable temperature mapping. Comprehensive simulation, in vitro, and in vivo experiments demonstrate that, compared with conventional phantom-based temperature mapping methods, the proposed method reduces the in vivo temperature reconstruction error by approximately 17.24% and achieves an average absolute temperature error below 1.257 °C. These results verify the feasibility of accurate in vivo temperature imaging using MPI and provide essential technical support for temperature-sensitive applications, including magnetic hyperthermia.","PeriodicalId":13418,"journal":{"name":"IEEE Transactions on Medical Imaging","volume":"1 1","pages":""},"PeriodicalIF":10.6,"publicationDate":"2026-03-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147359389","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 : 2026-03-04DOI: 10.1109/tmi.2026.3670643
C Ross Schmidtlein,Jin Ren,Andrzej Krol,Howard C Gifford,Joseph A O'Donoghue,Lisa Bodei,Yuesheng Xu
Targeted Alpha Therapy (TAT), using alpha-emitting radionuclides (AER) such as 225Ac, shows promise for the treatment of advanced and refractory cancers. Currently, TAT is prescribed on the basis of activity (e.g., MBq, kBq/kg), with no account taken of individual biodistribution or kinetics. The delivery of patient-specific treatment, based on absorbed dose criteria, requires in-vivo imaging of the AER biodistribution, a challenging scenario due to the scarcity of imageable photons. To address this, we present a novel computed quantitative planar (CQP) imaging method that reconstructs a coronal projection of the 3D AER distribution from anterior/posterior scintigraphy coregistered with CT. The model is regularized using maximum a posteriori estimation with sparse ℓ1 tight-framelet transforms and solved via a convergence-guaranteed fixed-point proximity algorithm. To experimentally evaluate our approach, we built a modular slab phantom containing a known distribution of 225Ac vitrified in epoxy. CQP reconstruction was characterized by significantly reduced bias and noise, improved spatial resolution, and better signal-to-noise ratios, compared to geometric mean methods. The CQP approach is clinically implementable with conventional SPECT/CT systems, without need for hardware additions or modifications, and can assist dosimetry workflows, especially where 3D SPECT/PET is impractical.
{"title":"Computed Quantitative Planar Imaging for Targeted Alpha Therapy: Model-Based Sparse Reconstruction Validated with a Novel 225Ac Epoxy Phantom.","authors":"C Ross Schmidtlein,Jin Ren,Andrzej Krol,Howard C Gifford,Joseph A O'Donoghue,Lisa Bodei,Yuesheng Xu","doi":"10.1109/tmi.2026.3670643","DOIUrl":"https://doi.org/10.1109/tmi.2026.3670643","url":null,"abstract":"Targeted Alpha Therapy (TAT), using alpha-emitting radionuclides (AER) such as 225Ac, shows promise for the treatment of advanced and refractory cancers. Currently, TAT is prescribed on the basis of activity (e.g., MBq, kBq/kg), with no account taken of individual biodistribution or kinetics. The delivery of patient-specific treatment, based on absorbed dose criteria, requires in-vivo imaging of the AER biodistribution, a challenging scenario due to the scarcity of imageable photons. To address this, we present a novel computed quantitative planar (CQP) imaging method that reconstructs a coronal projection of the 3D AER distribution from anterior/posterior scintigraphy coregistered with CT. The model is regularized using maximum a posteriori estimation with sparse ℓ1 tight-framelet transforms and solved via a convergence-guaranteed fixed-point proximity algorithm. To experimentally evaluate our approach, we built a modular slab phantom containing a known distribution of 225Ac vitrified in epoxy. CQP reconstruction was characterized by significantly reduced bias and noise, improved spatial resolution, and better signal-to-noise ratios, compared to geometric mean methods. The CQP approach is clinically implementable with conventional SPECT/CT systems, without need for hardware additions or modifications, and can assist dosimetry workflows, especially where 3D SPECT/PET is impractical.","PeriodicalId":13418,"journal":{"name":"IEEE Transactions on Medical Imaging","volume":"130 1","pages":""},"PeriodicalIF":10.6,"publicationDate":"2026-03-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147350524","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 : 2026-03-03DOI: 10.1109/tmi.2026.3670159
Xuan Gong,Jiaqi Li,Yirui Wang,Haoshen Li,Jiawen Yao,Lianzhen Zhong,Dazhou Guo,Ke Yan,David Doermann,Le Lu,Feiran Jiao,Tsung-Ying Ho,Ling Zhang,Abudili Abuduxuku,Haifeng Wang,Xianghua Ye,Dakai Jin,Qifeng Wang
Esophageal cancer is one of the most lethal cancers, with 5-year survival rate of only 20%. Patient outcomes can vary significantly even though they are at the same cancer stage and receive similar treatments. Accurate prognostic prediction for esophageal cancer patients is highly desired to receive personalized precise treatment. Nevertheless, there are very few automated methods yet to fully exploit the preoperative contrast-enhanced computed tomography (CE-CT) imaging for assessing esophageal cancer prognosis. In addition to image patterns, important prognostic factors should encompass tumor size and location, as well as lymph nodes (LNs) involvement, including features such as LN number, size, spatial distribution, and their proximity to tumor. Considering these complexities, we propose a novel Tumor and LN Context-Geometry network for the preoperative prediction of esophageal cancer survival in CE-CT images. Specifically, we (1) focus on learning survival patterns of CT texture via co-attention context modeling at most informative regions, i.e., automatically segmented tumor, LNs and LN-stations; and (2) integrate tumor and LN anatomical and spatial associations into neural geometry modeling for a comprehensive learning of metastatic involvement and tumor invasion to adjacent structures. Empirical studies show our presented framework can improve overall survival prediction performances compared with existing state-of-the-art survival analysis methods, and evidently suggest that incorporating these findings into the existing esophageal cancer staging system would add its clinical values.
{"title":"Preoperative Prediction of Esophageal Cancer Survival in CT via Tumor and Lymph Node Context and Geometry Modeling.","authors":"Xuan Gong,Jiaqi Li,Yirui Wang,Haoshen Li,Jiawen Yao,Lianzhen Zhong,Dazhou Guo,Ke Yan,David Doermann,Le Lu,Feiran Jiao,Tsung-Ying Ho,Ling Zhang,Abudili Abuduxuku,Haifeng Wang,Xianghua Ye,Dakai Jin,Qifeng Wang","doi":"10.1109/tmi.2026.3670159","DOIUrl":"https://doi.org/10.1109/tmi.2026.3670159","url":null,"abstract":"Esophageal cancer is one of the most lethal cancers, with 5-year survival rate of only 20%. Patient outcomes can vary significantly even though they are at the same cancer stage and receive similar treatments. Accurate prognostic prediction for esophageal cancer patients is highly desired to receive personalized precise treatment. Nevertheless, there are very few automated methods yet to fully exploit the preoperative contrast-enhanced computed tomography (CE-CT) imaging for assessing esophageal cancer prognosis. In addition to image patterns, important prognostic factors should encompass tumor size and location, as well as lymph nodes (LNs) involvement, including features such as LN number, size, spatial distribution, and their proximity to tumor. Considering these complexities, we propose a novel Tumor and LN Context-Geometry network for the preoperative prediction of esophageal cancer survival in CE-CT images. Specifically, we (1) focus on learning survival patterns of CT texture via co-attention context modeling at most informative regions, i.e., automatically segmented tumor, LNs and LN-stations; and (2) integrate tumor and LN anatomical and spatial associations into neural geometry modeling for a comprehensive learning of metastatic involvement and tumor invasion to adjacent structures. Empirical studies show our presented framework can improve overall survival prediction performances compared with existing state-of-the-art survival analysis methods, and evidently suggest that incorporating these findings into the existing esophageal cancer staging system would add its clinical values.","PeriodicalId":13418,"journal":{"name":"IEEE Transactions on Medical Imaging","volume":"12 1","pages":""},"PeriodicalIF":10.6,"publicationDate":"2026-03-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147346290","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：