DGCM-SSFA is a compact multi‑domain attention framework that synthesizes high‑fidelity 7T MRI from routine 3T scans using only 11 million parameters. It combines a Dual‑Granularity Cross‑Modulated Feature Modulator (DGCM) and a Sparse Strip Frequency Attention (SSFA) module within a U‑Net Transformer backbone to enhance local edge details and capture global spatial context efficiently. Experiments on UNC and BNU 3T‑to‑7T datasets show competitive structural fidelity, achieving SSIM scores of 0.8431 (BNU T2w) and 0.8100 (BNU T1w) while remaining far smaller than Transformer‑based alternatives.
arXiv:2605.05522v3 Announce Type: replace-cross Abstract: Although self-supervised pretraining is expected to learn broadly transferable representations, its effectiveness across imaging modalities s...
SRUG is a supervised, fusion‑driven generator network designed for paired MRI sequence synthesis. It eliminates the generator‑discriminator competition by directly learning a deterministic source‑to‑target mapping, using a residual encoder, full‑resolution stem, and a nested multi‑scale decoder to preserve structural fidelity. Experiments on BraTS 2023 and IXI demonstrate competitive reconstruction quality and cross‑dataset transferability, supporting direct supervised generation as a practical approach for medical image translation.
arXiv:2605.24625v2 Announce Type: replace Abstract: Ultra-low-field (ULF) MRI offers portable and accessible neuroimaging but suffers from reduced signal-to-noise ratio and limited spatial resolution...
The paper presents a method for synthesizing contrast‑enhanced MRI (CE‑MRI) from non‑contrast sequences using a novel Mixture of Feature Difference (MoFD) module. MoFD dynamically weights and aggregates features from individual sequences and their explicit differences, improving image fidelity and perceptual quality over baseline approaches. Experiments on UCSF‑PDGM and UPenn‑GBM datasets show that the synthesized images reconstruct tumor regions well and yield segmentation performance comparable to real CE‑MRI, with regression‑based training achieving higher structural fidelity.
AGGR-Mamba is a lightweight output‑stage refinement network for undersampled brain MRI reconstruction that builds on a Mamba backbone. The method first generates an initial reconstruction and then applies an Attention Refine Head to recalibrate local anatomical structures, edge textures, and residual‑error regions, followed by a gated residual refinement branch that selectively preserves informative residual features while suppressing redundant responses. Experiments on the IXI brain MRI dataset with T1‑ and T2‑weighted images at acceleration factors R=4 and R=8 show that AGGR‑Mamba achieves competitive PSNR and SSIM performance across different reconstruction settings.
The paper introduces FAD‑ViT, a vision transformer that decouples spatial and feature axes of global attention, making feature‑axis attention the primary operator. Experiments on 23 medical imaging tasks show that shifting global attention from spatial to feature leaves from‑scratch performance unchanged but significantly boosts transfer learning, achieving up to three times the gain with fewer parameters and lower compute. The study demonstrates that prioritizing feature‑axis attention is a simple, cost‑effective design choice for improving medical image transfer.
The paper introduces MR‑DiffuSR, a 3‑D latent diffusion framework that uses high‑resolution T1w structural priors to guide super‑resolution of thick‑slice FLAIR MRI scans. By applying cross‑modality structural swin attention and a mixed‑scale degradation strategy, the method avoids hallucinations and remains robust across varying slice thicknesses. On ADNI datasets, MR‑DiffuSR outperforms CNN and 2‑D diffusion baselines, achieving high PSNR, SSIM, and low LPIPS, and maintains strong white‑matter hyperintensity segmentation performance even at 7 mm equivalent slice thickness.
arXiv:2505.05855v4 Announce Type: replace Abstract: Multi-contrast super-resolution (MCSR) is crucial for enhancing MRI but current deep learning methods are limited. They typically require large, pa...
The paper introduces Neuro‑JEPA, a sparse multimodal foundation model that learns unified representations of brain MRI across T1w, T2w, and FLAIR sequences using a latent predictive objective and a Mixture‑of‑Experts architecture. It was pretrained on over 1.5 million scans from 428,647 studies and systematically evaluates architectural, masking, objective, and sparsity choices for robust multimodal representation learning. Across 47 tasks from three health systems and 12 public datasets, Neuro‑JEPA consistently outperforms a simple CNN baseline, demonstrating its effectiveness for both clinical and research applications.
arXiv:2607. 10988v1 Announce Type: cross Abstract: Perineural invasion (PNI) is a critical prognostic factor in cholangiocarcinoma.
arXiv:2606. 11107v1 Announce Type: cross Abstract: Clinicians diagnose brain tumors by synthesizing patient symptoms, medical history, and quantitative imaging data from modalities such as MRI and CT scans into a unified clinical judgement.
The paper introduces MSM‑Seg, a dual‑memory segmentation framework for 3D multi‑modal brain tumor segmentation. It combines a modality‑and‑slice memory attention module to capture cross‑modal and spatial‑slice dependencies, a multi‑scale category‑agnostic prompt encoder for whole‑tumor guidance, and a modality‑adaptive fusion decoder to integrate complementary decoding information. Experiments on various MRI datasets show that MSM‑Seg surpasses state‑of‑the‑art methods for metastases and glioma tumor segmentation.