arXiv subjects

Zhaolin Chen

Publications and source records attributed to Zhaolin Chen.

At least 19 recordsLinked to original sources

arXiv2026

Multi-Expert Conformal Risk Control for Pairwise LLM Judging in Open-Ended Dialogue

In this paper, we explore multi-expert Conformal Risk Control (CRC) algorithms for pairwise LLM-as-a-Judge evaluation in open-ended dialogue. Our core insight is that multi-expert aggregation offers a complementary remedy to CRC: whereas CRC controls risk at the decision threshold through abstention, aggregation sanitizes the scoring function at its source. Guided by this, we first design two multi-expert CRC methods: Score Averaging and Decision Voting, which aggregate at the score and decision levels, respectively. While both strategies outperform single-expert methods on homogeneous expert panels, on heterogeneous LLM judges they remain risk-valid but recover only limited coverage, because a uniform threshold cannot match the experts' distinct scoring scales. To resolve this issue, we further propose Marginal-Calibrated Conformal Consensus (MC3): it captures distinct per-expert scales via initial threshold ratios, while jointly tuning a unified decision function $C_t(x)$ applied identically in both calibration and test, thereby preserving exchangeability. To evaluate our framework, we construct Panel, a 1,800-pair human pairwise-preference benchmark for open-ended dialogue. It is built on responses generated by four open-weight LLMs over dialogue contexts from three domains (ESConv, MSC, DREAM), with full logit access. In experiments, we find that both Score Averaging and Decision Voting substantially improve accuracy and acceptance rate on homogeneous panels. Notably, MC3 extends these gains to heterogeneous panels by accommodating distinct per-expert scoring scales across all three datasets.

cs.CL↗

arXiv2026

Steer the Sampling, Not the Kernel Grid: Geometry-Guided Sampling Operator for Volumetric Segmentation

Accurate 3D segmentation is central to quantitative lesion assessment and anatomy mapping for clinical planning and follow-up. Thin, elongated, and fine anatomical/pathological structures (e.g., vessels) are a particularly challenging case: a one-voxel boundary error can disconnect a branch and change clinically relevant topology. In encoder-decoder networks (e.g., U-Net), repeated downsampling and fixed-grid convolution blur or alias fine structures and weaken orientation cues, so early mistakes propagate across scales. We propose a geometry-guided local operator that steers where features are sampled, rather than deforming convolutional kernels, under a single formulation for both feature refinement (stride 1) and resolution reduction (stride > 1). At each voxel, it predicts a local orientation and bounded step sizes, samples symmetrically along these directions, and transforms paired samples into compact geometric and boundary cues with lightweight mixing; a cross-scale consensus aligns encoder and decoder features at skip connections to reduce geometric mismatch. Replacing all stride 1 and stride 2 operators in a 3D U-Net yields consistent improvements on BraTS, MSD Hepatic Vessel, and TDSC-ABUS, with notably better boundary metrics (e.g., BraTS Dice 86.1 to 88.9, HD95 7.1 to 6.2; TDSC-ABUS HD95 39.1 to 27.8) while reducing parameters from 2.3M to 0.8M. We further demonstrate that the operator can be integrated into other backbones (e.g., nnU-Net, Swin-UNETR, and MedNeXt) without changing their macro-architectures while providing consistent performance gains.

cs.CV↗

arXiv2026

Boot-and-Feedback Framework for Generalist-Expert Model Collaboration in Breast Ultrasound Diagnosis

Breast ultrasound (BUS) is widely used for breast cancer diagnosis yet remains operator-dependent. While deep learning shows promise, ensuring diagnostic reliability and interpretability is challenging. Recent Multimodal Large Language Models (MLLMs) often generate spurious descriptions due to limited domain knowledge, which mislead downstream expert models and compromise clinical validity. To address these challenges, we propose the Boot-and-Feedback (BooF) model collaboration framework for synergistic MLLM-expert interaction. Specifically, in the Boot Stage, the MLLM is guided by the BI-RADS lexicon and preliminary benign-malignant vision-expert predictions, enabling it to transfer general reasoning to BUS analysis while avoiding hallucinations. Subsequently, the Feedback Stage integrates these descriptions with visual features via a lightweight Attention-Gated Cross-Modality Fusion Module. This allows the expert to leverage textual feedback while adaptively filtering noise. Extensive experiments on multiple BUS datasets demonstrate that BooF substantially outperforms state-of-the-art methods in terms of diagnostic accuracy and interpretability.

cs.CV↗

arXiv2026

SinoDiff: Physics-Consistent Self-Supervised Diffusion for Unified Low-Dose to Standard-Dose PET Sinogram Recovery

Low-dose positron emission tomography (LD-PET) reduces radiation exposure but leads to poor image quality and hinders diagnostic confidence. Existing supervised LD to standard-dose (SD) PET recovery methods often fail to generalise across dose variations, while dose-agnostic supervised methods require paired LD-SD data. Current self-supervised methods, although more flexible, typically produce inferior results, including loss of anatomical details and oversmoothing pathological features. These pose major limitations for practical applications. To overcome these limitations, we propose SinoDiff, a novel self-supervised, physics-consistent diffusion framework for the recovery of PET sinograms across multiple predefined dose levels. Unlike the noise simulation in traditional diffusion methods, SinoDiff integrates the PET acquisition model into the forward diffusion process via Poisson thinning, enabling physically consistent sampling/modelling of dose-dependent count statistics. During the reverse diffusion process, SinoDiff estimates the incremental change in PET signals from predefined dose levels. Therefore, SinoDiff is a single, unified model that requires no retraining across multiple dose levels. To consider the characteristics of PET sinogram, we incorporate a frequency-domain convolution to capture long-range dependencies across projection angles and detector bins. Experiments on [18F]-FDG and [18F]-FDOPA datasets demonstrate that SinoDiff achieves competitive performance against supervised and self-supervised baselines across multiple dose levels.

eess.IV↗

arXiv2026

Virtual Full-stack Scanning of Brain MRI via Imputing Any Quantised Code

Magnetic resonance imaging (MRI) is a powerful and versatile imaging technique, offering a wide spectrum of information about the anatomy by employing different acquisition modalities. However, in the clinical workflow, it is impractical to collect all relevant modalities due to the scan time and cost constraints. Virtual full-stack scanning aims to impute missing MRI modalities from available but incomplete acquisitions, offering a cost-efficient solution to enhance data completeness and clinical usability. Existing imputation methods often depend on global conditioning or modality-specific designs, which limit their generalisability across patient cohorts and imaging protocols. To address these limitations, we propose CodeBrain, a unified framework that reformulates various ``any-to-any'' imputation tasks as a region-level full-stack code prediction problem. CodeBrain adopts a two-stage pipeline: (1) it learns the compact representation of a complete MRI modality set by encoding it into scalar-quantised codes at the region level, enabling high-fidelity image reconstruction after decoding these codes along with modality-agnostic common features; (2) it trains a projection encoder to predict the full-stack code map from incomplete modalities via a grading-based design for diverse imputation scenarios. Extensive experiments on two public brain MRI datasets, i.e., IXI and BraTS 2023, demonstrate that CodeBrain consistently outperforms state-of-the-art methods, establishing a new benchmark for unified brain MRI imputation and enabling virtual full-stack scanning. Our code will be released at https://github.com/ycwu1997/CodeBrain.

cs.CV↗

arXiv2026

ReDiff: Reliability-Guided Diffusion for Trustworthy Ultra-Low-Field to High-Field MRI Synthesis

Low-field to high-field MRI synthesis has emerged as a promising strategy to improve image quality when access to high-field scanners is limited. However, in ultra-low-field settings, the degradation of anatomical detail is spatially heterogeneous: structurally ambiguous regions are more susceptible to unstable high-frequency generation, which may produce anatomically inconsistent textures and boundaries. This issue is particularly problematic when synthesized images are used for downstream quantitative analysis. We therefore study how to make diffusion-based LF-to-HF synthesis more spatially reliable, rather than only sharper on average. To this end, we propose a reliability-guided diffusion framework (ReDiff) with two complementary inference-time mechanisms. First, a reliability-guided sampling strategy attenuates unstable reverse-diffusion updates in regions with weak low-field support. Second, an uncertainty-aware candidate selection scheme aggregates multiple stochastic reconstructions according to spatial consensus and predictive uncertainty. Beyond aggregate image quality, we test whether the uncertainty is itself a usable reliability signal. Experiments on paired 64mT$\rightarrow$3T MRI datasets show that ReDiff attains the lowest LPIPS across three contrasts and two datasets while remaining competitive on PSNR and SSIM, and downstream segmentation analysis indicates better preservation of anatomical structure.

cs.CV↗

arXiv2026

M2Diff: Multi-Modality Multi-Task Enhanced Diffusion Model for MRI-Guided Low-Dose PET Enhancement

Positron emission tomography (PET) scans expose patients to radiation, which can be mitigated by reducing the dose, albeit at the cost of diminished quality. This makes low-dose (LD) PET recovery an active research area. Previous studies have focused on standard-dose (SD) PET recovery from LD PET scans and/or multi-modal scans, e.g., PET/CT or PET/MRI, using deep learning. While these studies incorporate multi-modal information through conditioning in a single-task model, such approaches may limit the capacity to extract modality-specific features, potentially leading to early feature dilution. Although recent studies have begun incorporating pathology-rich data, challenges remain in effectively leveraging multi-modality inputs for reconstructing diverse features, particularly in heterogeneous patient populations. To address these limitations, we introduce a multi-modality multi-task diffusion model (M2Diff) that processes MRI and LD PET scans separately to learn modality-specific features and fuse them via hierarchical feature fusion to reconstruct SD PET. This design enables effective integration of complementary structural and functional information, leading to improved reconstruction fidelity. We have validated the effectiveness of our model on both healthy and Alzheimer's disease brain datasets. The M2Diff achieves superior qualitative and quantitative performance on both datasets.

eess.IV↗

arXiv2025

Optimize Flip Angle Schedules In MR Fingerprinting Using Reinforcement Learning

Magnetic Resonance Fingerprinting (MRF) leverages transient-state signal dynamics generated by the tunable acquisition parameters, making the design of an optimal, robust sequence a complex, high-dimensional sequential decision problem, such as optimizing one of the key parameters, flip angle. Reinforcement learning (RL) offers a promising approach to automate parameter selection, to optimize pulse sequences that maximize the distinguishability of fingerprints across the parameter space. In this work, we introduce an RL framework for optimizing the flip-angle schedule in MRF and demonstrate a learned schedule exhibiting non-periodic patterns that enhances fingerprint separability. Additionally, an interesting observation is that the RL-optimized schedule may enable a reduction in the number of repetition time, potentially accelerate MRF acquisitions.

cs.LG↗

arXiv2025

Style Content Decomposition-based Data Augmentation for Domain Generalizable Medical Image Segmentation

Due to domain shifts across diverse medical imaging modalities, learned segmentation models often suffer significant performance degradation during deployment. We posit that these domain shifts can generally be categorized into two main components: 1) "style" shifts, referring to global disparities in image properties such as illumination, contrast, and color; and 2) "content" shifts, which involve local discrepancies in anatomical structures. To address the domain shifts in medical image segmentation, we first factorize an image into style codes and content maps, explicitly modeling the "style" and "content" components. Building on this, we introduce a Style-Content decomposition-based data augmentation algorithm (StyCona), which performs augmentation on both the global style and local content of source-domain images, enabling the training of a well-generalized model for domain generalizable medical image segmentation. StyCona is a simple yet effective plug-and-play module that substantially improves model generalization without requiring additional training parameters or modifications to segmentation model architectures. Experiments on cardiac magnetic resonance imaging and fundus photography segmentation tasks, with single and multiple target domains respectively, demonstrate the effectiveness of StyCona and its superiority over state-of-the-art domain generalization methods. The code is available at https://github.com/Senyh/StyCona.

eess.IV↗

arXiv2025

Segment Any Tumour: An Uncertainty-Aware Vision Foundation Model for Whole-Body Analysis

Prompt-driven vision foundation models, such as the Segment Anything Model, have recently demonstrated remarkable adaptability in computer vision. However, their direct application to medical imaging remains challenging due to heterogeneous tissue structures, imaging artefacts, and low-contrast boundaries, particularly in tumours and cancer primaries leading to suboptimal segmentation in ambiguous or overlapping lesion regions. Here, we present Segment Any Tumour 3D (SAT3D), a lightweight volumetric foundation model designed to enable robust and generalisable tumour segmentation across diverse medical imaging modalities. SAT3D integrates a shifted-window vision transformer for hierarchical volumetric representation with an uncertainty-aware training pipeline that explicitly incorporates uncertainty estimates as prompts to guide reliable boundary prediction in low-contrast regions. Adversarial learning further enhances model performance for the ambiguous pathological regions. We benchmark SAT3D against three recent vision foundation models and nnUNet across 11 publicly available datasets, encompassing 3,884 tumour and cancer cases for training and 694 cases for in-distribution evaluation. Trained on 17,075 3D volume-mask pairs across multiple modalities and cancer primaries, SAT3D demonstrates strong generalisation and robustness. To facilitate practical use and clinical translation, we developed a 3D Slicer plugin that enables interactive, prompt-driven segmentation and visualisation using the trained SAT3D model. Extensive experiments highlight its effectiveness in improving segmentation accuracy under challenging and out-of-distribution scenarios, underscoring its potential as a scalable foundation model for medical image analysis.

eess.IV↗

arXiv2025

Cyclic Self-Supervised Diffusion for Ultra Low-field to High-field MRI Synthesis

Synthesizing high-quality images from low-field MRI holds significant potential. Low-field MRI is cheaper, more accessible, and safer, but suffers from low resolution and poor signal-to-noise ratio. This synthesis process can reduce reliance on costly acquisitions and expand data availability. However, synthesizing high-field MRI still suffers from a clinical fidelity gap. There is a need to preserve anatomical fidelity, enhance fine-grained structural details, and bridge domain gaps in image contrast. To address these issues, we propose a \emph{cyclic self-supervised diffusion (CSS-Diff)} framework for high-field MRI synthesis from real low-field MRI data. Our core idea is to reformulate diffusion-based synthesis under a cycle-consistent constraint. It enforces anatomical preservation throughout the generative process rather than just relying on paired pixel-level supervision. The CSS-Diff framework further incorporates two novel processes. The slice-wise gap perception network aligns inter-slice inconsistencies via contrastive learning. The local structure correction network enhances local feature restoration through self-reconstruction of masked and perturbed patches. Extensive experiments on cross-field synthesis tasks demonstrate the effectiveness of our method, achieving state-of-the-art performance (e.g., 31.80 $\pm$ 2.70 dB in PSNR, 0.943 $\pm$ 0.102 in SSIM, and 0.0864 $\pm$ 0.0689 in LPIPS). Beyond pixel-wise fidelity, our method also preserves fine-grained anatomical structures compared with the original low-field MRI (e.g., left cerebral white matter error drops from 12.1$\%$ to 2.1$\%$, cortex from 4.2$\%$ to 3.7$\%$). To conclude, our CSS-Diff can synthesize images that are both quantitatively reliable and anatomically consistent.

cs.CV↗

arXiv2025

MedNet-PVS: A MedNeXt-Based Deep Learning Model for Automated Segmentation of Perivascular Spaces

Enlarged perivascular spaces (PVS) are increasingly recognized as biomarkers of cerebral small vessel disease, Alzheimer's disease, stroke, and aging-related neurodegeneration. However, manual segmentation of PVS is time-consuming and subject to moderate inter-rater reliability, while existing automated deep learning models have moderate performance and typically fail to generalize across diverse clinical and research MRI datasets. We adapted MedNeXt-L-k5, a Transformer-inspired 3D encoder-decoder convolutional network, for automated PVS segmentation. Two models were trained: one using a homogeneous dataset of 200 T2-weighted (T2w) MRI scans from the Human Connectome Project-Aging (HCP-Aging) dataset and another using 40 heterogeneous T1-weighted (T1w) MRI volumes from seven studies across six scanners. Model performance was evaluated using internal 5-fold cross validation (5FCV) and leave-one-site-out cross validation (LOSOCV). MedNeXt-L-k5 models trained on the T2w images of the HCP-Aging dataset achieved voxel-level Dice scores of 0.88+/-0.06 (white matter, WM), comparable to the reported inter-rater reliability of that dataset, and the highest yet reported in the literature. The same models trained on the T1w images of the HCP-Aging dataset achieved a substantially lower Dice score of 0.58+/-0.09 (WM). Under LOSOCV, the model had voxel-level Dice scores of 0.38+/-0.16 (WM) and 0.35+/-0.12 (BG), and cluster-level Dice scores of 0.61+/-0.19 (WM) and 0.62+/-0.21 (BG). MedNeXt-L-k5 provides an efficient solution for automated PVS segmentation across diverse T1w and T2w MRI datasets. MedNeXt-L-k5 did not outperform the nnU-Net, indicating that the attention-based mechanisms present in transformer-inspired models to provide global context are not required for high accuracy in PVS segmentation.

cs.CV↗

arXiv2025

D2Diff : A Dual Domain Diffusion Model for Accurate Multi-Contrast MRI Synthesis

Multi contrast MRI synthesis is inherently challenging due to the complex and nonlinear relationships among different contrasts. Each MRI contrast highlights unique tissue properties, but their complementary information is difficult to exploit due to variations in intensity distributions and contrast specific textures. Existing methods for multi contrast MRI synthesis primarily utilize spatial domain features, which capture localized anatomical structures but struggle to model global intensity variations and distributed patterns. Conversely, frequency domain features provide structured inter contrast correlations but lack spatial precision, limiting their ability to retain finer details. To address this, we propose a dual domain learning framework that integrates spatial and frequency domain information across multiple MRI contrasts for enhanced synthesis. Our method employs two mutually trained denoising networks, one conditioned on spatial domain and the other on frequency domain contrast features through a shared critic network. Additionally, an uncertainty driven mask loss directs the models focus toward more critical regions, further improving synthesis accuracy. Extensive experiments show that our method outperforms SOTA baselines, and the downstream segmentation performance highlights the diagnostic value of the synthetic results.

eess.IV↗

arXiv2025

PeFoMed: Parameter Efficient Fine-tuning of Multimodal Large Language Models for Medical Imaging

Multimodal large language models (MLLMs) represent an evolutionary expansion in the capabilities of traditional large language models, enabling them to tackle challenges that surpass the scope of purely text-based applications. It leverages the knowledge previously encoded within these language models, thereby enhancing their applicability and functionality in the reign of multimodal contexts. Recent works investigate the adaptation of MLLMs as a universal solution to address medical multi-modal problems as a generative task. In this paper, we propose a parameter efficient framework for fine-tuning MLLMs, specifically validated on medical visual question answering (Med-VQA) and medical report generation (MRG) tasks, using public benchmark datasets. We also introduce an evaluation metric using the 5-point Likert scale and its weighted average value to measure the quality of the generated reports for MRG tasks, where the scale ratings are labelled by both humans manually and the GPT-4 model. We further assess the consistency of performance metrics across traditional measures, GPT-4, and human ratings for both VQA and MRG tasks. The results indicate that semantic similarity assessments using GPT-4 align closely with human annotators and provide greater stability, yet they reveal a discrepancy when compared to conventional lexical similarity measurements. This questions the reliability of lexical similarity metrics for evaluating the performance of generative models in Med-VQA and report generation tasks. Besides, our fine-tuned model significantly outperforms GPT-4v. This indicates that without additional fine-tuning, multi-modal models like GPT-4v do not perform effectively on medical imaging tasks. The code will be available here: https://github.com/jinlHe/PeFoMed.

cs.CL↗

arXiv2024

A Physics-based Generative Model to Synthesize Training Datasets for MRI-based Fat Quantification

Deep learning-based techniques have potential to optimize scan and post-processing times required for MRI-based fat quantification, but they are constrained by the lack of large training datasets. Generative models are a promising tool to perform data augmentation by synthesizing realistic datasets. However no previous methods have been specifically designed to generate datasets for quantitative MRI (q-MRI) tasks, where reference quantitative maps and large variability in scanning protocols are usually required. We propose a Physics-Informed Latent Diffusion Model (PI-LDM) to synthesize quantitative parameter maps jointly with customizable MR images by incorporating the signal generation model. We assessed the quality of PI-LDM's synthesized data using metrics such as the Fréchet Inception Distance (FID), obtaining comparable scores to state-of-the-art generative methods (FID: 0.0459). We also trained a U-Net for the MRI-based fat quantification task incorporating synthetic datasets. When we used a few real (10 subjects, $~200$ slices) and numerous synthetic samples ($>3000$), fat fraction at specific liver ROIs showed a low bias on data obtained using the same protocol than training data ($0.10\%$ at $\hbox{ROI}_1$, $0.12\%$ at $\hbox{ROI}_2$) and on data acquired with an alternative protocol ($0.14\%$ at $\hbox{ROI}_1$, $0.62\%$ at $\hbox{ROI}_2$). Future work will be to extend PI-LDM to other q-MRI applications.

eess.SP↗

arXiv2024

Bilateral Hippocampi Segmentation in Low Field MRIs Using Mutual Feature Learning via Dual-Views

Accurate hippocampus segmentation in brain MRI is critical for studying cognitive and memory functions and diagnosing neurodevelopmental disorders. While high-field MRIs provide detailed imaging, low-field MRIs are more accessible and cost-effective, which eliminates the need for sedation in children, though they often suffer from lower image quality. In this paper, we present a novel deep-learning approach for the automatic segmentation of bilateral hippocampi in low-field MRIs. Extending recent advancements in infant brain segmentation to underserved communities through the use of low-field MRIs ensures broader access to essential diagnostic tools, thereby supporting better healthcare outcomes for all children. Inspired by our previous work, Co-BioNet, the proposed model employs a dual-view structure to enable mutual feature learning via high-frequency masking, enhancing segmentation accuracy by leveraging complementary information from different perspectives. Extensive experiments demonstrate that our method provides reliable segmentation outcomes for hippocampal analysis in low-resource settings. The code is publicly available at: https://github.com/himashi92/LoFiHippSeg.

eess.IV↗

arXiv2024

Data and Physics driven Deep Learning Models for Fast MRI Reconstruction: Fundamentals and Methodologies

Magnetic Resonance Imaging (MRI) is a pivotal clinical diagnostic tool, yet its extended scanning times often compromise patient comfort and image quality, especially in volumetric, temporal and quantitative scans. This review elucidates recent advances in MRI acceleration via data and physics-driven models, leveraging techniques from algorithm unrolling models, enhancement-based methods, and plug-and-play models to the emerging full spectrum of generative model-based methods. We also explore the synergistic integration of data models with physics-based insights, encompassing the advancements in multi-coil hardware accelerations like parallel imaging and simultaneous multi-slice imaging, and the optimization of sampling patterns. We then focus on domain-specific challenges and opportunities, including image redundancy exploitation, image integrity, evaluation metrics, data heterogeneity, and model generalization. This work also discusses potential solutions and future research directions, with an emphasis on the role of data harmonization and federated learning for further improving the general applicability and performance of these methods in MRI reconstruction.

eess.SP↗

arXiv2024

Deep kernel representations of latent space features for low-dose PET-MR imaging robust to variable dose reduction

Low-dose positron emission tomography (PET) image reconstruction methods have potential to significantly improve PET as an imaging modality. Deep learning provides a promising means of incorporating prior information into the image reconstruction problem to produce quantitatively accurate images from compromised signal. Deep learning-based methods for low-dose PET are generally poorly conditioned and perform unreliably on images with features not present in the training distribution. We present a method which explicitly models deep latent space features using a robust kernel representation, providing robust performance on previously unseen dose reduction factors. Additional constraints on the information content of deep latent features allow for tuning in-distribution accuracy and generalisability. Tests with out-of-distribution dose reduction factors ranging from $\times 10$ to $\times 1000$ and with both paired and unpaired MR, demonstrate significantly improved performance relative to conventional deep-learning methods trained using the same data. Code:https://github.com/cameronPain

cs.CV↗