arXiv:2607. 24453v1 Announce Type: cross Abstract: Learning from minimal human supervision is a long-standing goal in medical image analysis, where dense expert annotations are costly.
arXiv:2606. 17972v1 Announce Type: cross Abstract: Self-supervised DINO models provide strong transferable visual representations, yet applying them directly to image segmentation remains challenging.
The paper introduces Spatial‑FAD, a few‑shot medical anomaly detection framework that fuses Vision‑Language Model (CLIP) semantics with spatial priors from Vision Foundation Models (DINO). A VFM‑enhanced adapter injects structural affinity into CLIP features, while a sliding‑window aggregation produces high‑resolution embeddings for finer lesion localization. Prototype‑enhanced support memory further improves efficiency and performance, yielding significant gains on Liver CT, Retinal OCT, and Brain MRI datasets, notably an 11.4% Dice improvement in 4‑shot scenarios.
arXiv:2609.18578v1 Announce Type: new Abstract: Pathologists diagnose diseases by first locating suspicious tissue and then examining it at higher magnification, whereas self-supervised vision transf...
arXiv:2608.24281v1 Announce Type: new Abstract: Reducing annotation requirements remains a key challenge in developing robust medical object detectors. To address this, Vision-Language (VL) object de...
Exemplar is a few‑shot segmentation method that combines a frozen DINOv3 backbone with a fixed bank of classical native‑resolution filter responses in a single lightweight head. Trained only from support masks, it achieves a mean foreground intersection‑over‑union of 0.782 across eleven biomedical imaging datasets, outperforming either component alone and surpassing five other few‑shot methods in 54 of 55 comparisons. With a single annotated mask, Exemplar reaches 0.703, higher than a from‑scratch nnU‑Net trained on the same mask, and while nnU‑Net eventually overtakes it with eight masks, it requires 16–77× longer to fit.
arXiv:2606. 03888v1 Announce Type: cross Abstract: Self-supervised learning has enabled large-scale pre-training on 2D natural images, producing general-purpose visual representations that transfer effectively across tasks.
The paper introduces a dual‑input, multi‑task learning framework that jointly segments and classifies bone tumors by applying bidirectional cross‑modal attention between a lesion crop and the full radiograph. Using a YOLO‑based detector and a dual‑stream DenseNet121 architecture, the model fuses fine‑grained lesion detail with global anatomical context through a novel cross‑modal attention fusion strategy and hierarchical multi‑scale feature fusion. On the multi‑institutional Bone Tumor X‑ray Radiograph Dataset, the approach outperforms single‑input baselines, achieving a Dice coefficient of 0.896 and a macro‑averaged F1‑score of 0.928, with an AUC of 0.999 for malignant osteosarcoma.
The paper introduces GAD-MambaUNet, a lightweight medical image segmentation network that integrates efficient local modeling, Direction-Group Graph Selective Scan (DG‑GSS) for structured information exchange, and training‑time supervision from a frozen DINOv3 teacher with Gradient‑Adaptive Distillation. GAD‑MambaUNet demonstrates a strong accuracy‑efficiency trade‑off compared to other lightweight and general segmentation methods, and ablation studies confirm the benefits of DG‑GSS and DINOv3‑GAD supervision. Future work aims to refine teacher‑student alignment and apply the framework to multi‑class and multi‑modal medical segmentation tasks.
The paper introduces an unsupervised domain adaptation framework that aligns redundancy-reducing features to enable accurate 3D segmentation of cone-beam CT (CBCT) without target-domain annotations or inference-time adaptation. The method is architecture-agnostic, working with both CNN-based and ViT-based foundation models, and is evaluated on two liver segmentation benchmarks for interventional vascular procedures and radiation therapy. Results show that even large pretrained segmentation networks need explicit feature-space bridging to generalize across diagnostic CT and CBCT, and the proposed approach consistently outperforms existing pretrained foundation models and UDA strategies.
GazeRefine is a training‑free framework that uses eye‑gaze data as an inference‑time prompt for zero‑shot medical image segmentation. It converts sparse, duration‑weighted fixations into foreground and background priors that initialize semantic prototypes in a frozen DINOv3 feature space, then iteratively refines these prototypes through discrimination, affinity propagation, and anchoring to the gaze guidance. The method achieves strong results on colonoscopy polyp segmentation and competitive performance on prostate MRI, demonstrating that gaze‑guided prototype refinement can enable segmentation without dense expert annotations or model fine‑tuning.
DINO-Med introduces a patch‑based framework that adapts natural‑image foundation models, specifically DINOv3, to multi‑modal medical imaging. The method uses training‑free registration, automated localization, and mask‑filtered patch extraction to aggregate patch‑level features into subject‑level diagnostics. In liver fibrosis staging, DINOv3 outperforms handcrafted radiomics, ResNet, and SAM‑Med2D features, achieving 78.4% accuracy for mild fibrosis (S1) and 75.8% for cirrhosis (S4) on the CARE 2025 cohort.