Results
- Showing results for:
- Reset all filters
Search results
-
Journal articleYu Q, Zhang C, Jin G, et al., 2026,
StealthMark: Harmless and Stealthy Ownership Verification for Medical Segmentation via Uncertainty-Guided Backdoors.
, IEEE Trans Image Process, Vol: 35, Pages: 1290-1304Annotating medical data for training AI models is often costly and limited due to the shortage of specialists with relevant clinical expertise. This challenge is further compounded by privacy and ethical concerns associated with sensitive patient information. As a result, well-trained medical segmentation models on private datasets constitute valuable intellectual property requiring robust protection mechanisms. Existing model protection techniques primarily focus on classification and generative tasks, while segmentation models-crucial to medical image analysis-remain largely underexplored. In this paper, we propose a novel, stealthy, and harmless method, StealthMark, for verifying the ownership of medical segmentation models under closed-box conditions. Our approach subtly modulates model uncertainty without altering the final segmentation outputs, thereby preserving the model's performance. To enable ownership verification, we incorporate model-agnostic explanation methods, e.g. LIME, to extract feature attributions from the model outputs. Under specific triggering conditions, these explanations reveal a distinct and verifiable watermark. We further design the watermark as a QR code to facilitate robust and recognizable ownership claims. We conducted extensive experiments across four medical imaging datasets (CMR dataset from UK Biobank, the SEG fundus dataset, the EchoNet echocardiography dataset, and the PraNet colonoscopy dataset) and five mainstream segmentation models. The results demonstrate the effectiveness, stealthiness, and harmlessness of our method on the original model's segmentation performance. For example, when applied to the SAM model, StealthMark consistently achieved attack success rates (ASR) above 95% across various datasets while maintaining less than a 1% drop in Dice and AUC scores-significantly outperforming backdoor-based watermarking methods and highlighting its strong potential for practical deployment. Our implementation code is made
-
Journal articleZhang Z, Zhang L, Cheng Y, et al., 2026,
From Coarse to Continuous: Progressive Refinement Implicit Neural Representation for Motion-Robust Anisotropic MRI Reconstruction.
, IEEE Trans Image Process, Vol: 35, Pages: 3550-3565In motion-robust magnetic resonance imaging (MRI), slice-to-volume reconstruction is critical for recovering anatomically consistent 3D brain volumes from 2D slices, especially under accelerated acquisitions or patient motion. However, this task remains challenging due to hierarchical structural disruptions. It includes local detail loss from k-space undersampling, global structural aliasing caused by motion, and volumetric anisotropy. Therefore, we propose a progressive refinement implicit neural representation (PR-INR) framework. Our PR-INR unifies motion correction, structural refinement, and volumetric synthesis within a geometry-aware coordinate space. Specifically, a motion-aware diffusion module is first employed to generate coarse volumetric reconstructions that suppress motion artifacts and preserve global anatomical structures. Then, we introduce an implicit detail restoration module that performs residual refinement by aligning spatial coordinates with visual features. It corrects local structures and enhances boundary precision. Further, a voxel continuous-aware representation module represents the image as a continuous function over 3D coordinates. It enables accurate inter-slice completion and high-frequency detail recovery. We evaluate PR-INR on five public MRI datasets under various motion conditions (3% and 5% displacement), undersampling rates (4x and 8x) and slice resolutions (scale = 5). Experimental results demonstrate that PR-INR outperforms state-of-the-art methods in both quantitative reconstruction metrics and visual quality. It further shows generalization and robustness across diverse unseen domains.
-
Journal articleMarkus JE, Cristinacce PLH, Punwani S, et al., 2025,
Steps on the Path to Clinical Translation-A British and Irish Chapter ISMRM Workshop Survey of the UK MRI Community
, MAGNETIC RESONANCE IN MEDICINE, ISSN: 0740-3194 -
Journal articleTänzer M, Scott AD, Khalique Z, et al., 2025,
Accelerating cDTI with deep learning-based tensor de-noising and breath hold reduction. a step towards improved efficiency and clinical feasibility
, Journal of Cardiovascular Magnetic Resonance, Vol: 27, ISSN: 1097-6647BackgroundCardiac Diffusion Tensor Imaging (cDTI) non-invasively provides unique insights into cardiac microstructure. Current protocols require multiple breath-hold repetitions to achieve adequate signal-to-noise ratio, resulting in lengthy scan times. The aim of this study was to develop a cDTI de-noising method that would enable the reduction of repetitions while preserving image quality.MethodsWe present a novel de-noising framework for cDTI acceleration centred on three fundamental advances: (1) a paradigm shift from image-based to tensor-space de-noising that better preserves structural information, (2) an ensemble of Vision Transformer-based models specifically optimised for tensor processing through adversarial training, and (3) a sophisticated data augmentation strategy that maximises training data utilisation through dynamic repetition selection.ResultsOur approach reduces scan times by a factor of up to 4 while achieving a 20% reduction in cDTI maps errors over existing de-noising methods (Table 1) and preserving anatomical features such as infarct characterisation and transmural cardiomyocyte orientation patterns. Crucially, our proposed method succeeds in clinical cases where other algorithms previously failed.ConclusionsThis demonstrates substantial improvements in cDTI acquisition efficiency, achieving up to 4-fold scan time reduction (3-5 breath-holds) while maintaining diagnostic accuracy across diverse cardiac pathologies.
-
Journal articleTeh I, Moulin K, Ferreira PF, et al., 2025,
Multi-center investigation of cardiac diffusion tensor imaging in healthy volunteers by the Society of Cardiovascular Magnetic Resonance Cardiac Diffusion Special Interest Group NETwork (SIGNET)
, Journal of Cardiovascular Magnetic Resonance, Vol: 27, ISSN: 1097-6647BackgroundCardiac diffusion tensor imaging (cDTI) is an emerging technique for microstructural characterization of the heart and has shown clinical potential in a range of cardiomyopathies. However, there is substantial variation reported for in vivo cDTI results across the literature, and sensitivity of cDTI to differences in imaging sites, scanners, acquisition protocols, and post-processing methods remains incompletely understood.MethodsSIGNET is a prospective multi-center, observational study in traveling and non-traveling healthy volunteers. The study was initiated by the executive board of the Society of Cardiovascular Magnetic Resonance (SCMR) Cardiac Diffusion Special Interest Group (SIG) as a follow-up to a previous multi-center study on phantom validation of cardiac DTI and a recently published SCMR consensus statement on cardiac diffusion MRI. The study has been developed by the Project Management Committee in consultation with the SCMR cardiac diffusion SIG, which includes international experts in cardiac diffusion MRI. To date, more than 20 international institutions have engaged with the study, including sites that are new to cardiac DTI, making this the largest collaborative effort in the field.DiscussionSIGNET will provide important information about the key sources of variation in cardiac DTI. This will help rationalize strategies for addressing and minimizing such variation. Harmonization of protocols in this and future studies will underpin efforts to translate cardiac DTI for clinical application.
-
Journal articleJin W, Tian X, Wang N, et al., 2025,
Representation-driven sampling and adaptive policy resetting for improving multi-Agent reinforcement learning
, NEURAL NETWORKS, Vol: 192, ISSN: 0893-6080 -
Journal articleHao P, Wang H, Yang G, et al., 2025,
Enhancing Visual Reasoning With LLM-Powered Knowledge Graphs for Visual Question Localized-Answering in Robotic Surgery
, IEEE JOURNAL OF BIOMEDICAL AND HEALTH INFORMATICS, Vol: 29, Pages: 9027-9040, ISSN: 2168-2194- Cite
- Citations: 2
-
Journal articleWang H, Chen Y, Chen W, et al., 2025,
Serp-Mamba: Advancing High-Resolution Retinal Vessel Segmentation With Selective State-Space Model.
, IEEE Trans Med Imaging, Vol: 44, Pages: 4811-4825Ultra-Wide-Field Scanning Laser Ophthalmoscopy (UWF-SLO) images capture high-resolution views of the retina with typically spanning 200 degrees. Accurate segmentation of vessels in UWF-SLO images is essential for detecting and diagnosing fundus disease. Recent studies highlight that Mamba's selective State Space Model (SSM) excels in modeling long-range dependencies with linear computational complexity, making it highly suitable for preserving the continuity of elongated vessel structures, especially for high-resolution UWF images. Inspired by this, we propose the Serpentine Mamba (Serp-Mamba) network to address this challenging task. Specifically, we recognize the intricate, varied, and delicate nature of the tubular structure of vessels. Furthermore, the high-resolution of UWF-SLO images exacerbates the imbalance between the vessel and background categories. Based on the above observations, we first devise a Serpentine Interwoven Adaptive (SIA) scan mechanism, which scans UWF-SLO images along curved vessel structures in a snake-like crawling manner. This approach, consistent with vascular texture transformations, ensures the effective and continuous capture of curved vascular structure features. Second, we propose an Ambiguity-Driven Dual Recalibration (ADDR) module to address the category imbalance problem intensified by high-resolution images. Our ADDR module delineates pixels by two learnable thresholds and refines ambiguous pixels through a dual-driven strategy, thereby accurately distinguishing vessels and background regions. Experiment results on three datasets demonstrate the superior performance of our Serp-Mamba on high-resolution vessel segmentation. We also conduct a series of ablation studies to verify the impact of our designs. Our code will be released upon publication (https://github.com/whq-xxh/Serp-Mamba).
-
Journal articleWang Z, Xiao M, Zhou Y, et al., 2025,
Deep Separable Spatiotemporal Learning for Fast Dynamic Cardiac MRI
, IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, Vol: 72, Pages: 3642-3654, ISSN: 0018-9294- Cite
- Citations: 4
-
Journal articleConnor S, Lally P, Pai I, et al., 2025,
7-Tesla sodium magnetic resonance imaging of the inner ears in unilateral Ménière’s disease and endolymphatic hydrops: an exploratory study
, BMC Medical Imaging, Vol: 25, ISSN: 1471-2342BackgroundWhilst delayed post-gadolinium MRI has led to a shift in the diagnostic paradigm of Meniere’s Disease (MD), there remains a strong desire to develop a non-contrast enhanced MRI technique to detect and monitor MD. The endolymphatic space (ES) undergoes hydropic expansion in Ménière’s Disease (MD) and the concentration of sodium ions in the endolymph is at least 10 times lower than that in the perilymph. It was hypothesised that the lower sodium (23Na) concentration in the endolymph relative to the surrounding perilymph would result in a differential reduction in 23Na-MRI signal in inner ears with endolymphatic hydrops (EH). This proof of principle study explored the feasibility of 7-Tesla (7T) 23Na-MRI to lateralise EH ears in unilateral MD.MethodsIn this prospective study, 7T 23Na-MRI was performed in participants with both unilateral definite MD and severe vestibulo-cochlear EH on a delayed post-gadolinium real inversion recovery sequence. Two blinded independent observers qualitatively graded the visibility and anatomical compatibility of inner ear 23Na MRI signal intensity (NaSI), before and after registering to 3D T2-weighted (T2w) MRI and determined the certainty of EH laterality. The internal auditory meatus (IAM), cochlea and vestibule were segmented using 3D Slicer and NaSI was quantified. Inner ear median NaSI were scaled to the adjacent IAM median NaSI and compared between the two ears.ResultsIn 4 unilateral MD participants (mean age 60.3 years, 2 men), both observers correctly predicted EH laterality in 1/4 before and 3/4 participants after fusion to 3D T2w MRI. There was no incorrect lateralisation of EH by either observer, either before or after registration and fusion. In the 3 participants correctly lateralised, quantitative analysis revealed the median inner ear NaSI scaled to the ipsilateral IAM was 1.2–2.8 times higher in the normal cochlea and 1.9–2.9 times higher in the vestibule, compared to
This data is extracted from the Web of Science and reproduced under a licence from Thomson Reuters. You may not copy or re-distribute this data in whole or in part without the written consent of the Science business of Thomson Reuters.
Contact
For enquiries about the MRI Physics Collective, please contact:
Mary Finnegan
Senior MR Physicist at the Imperial College Healthcare NHS Trust
Pete Lally
Assistant Professor in Magnetic Resonance (MR) Physics at Imperial College
Jan Sedlacik
MR Physicist at the Robert Steiner MR Unit, Hammersmith Hospital Campus