RealSyn: An Effective and Scalable Multimodal Interleaved Document Transformation Paradigm

After pre-training on extensive image-text pairs, Contrastive Language-Image Pre-training (CLIP) demonstrates promising performance on a wide variety of benchmarks. However, a substantial volume of non-paired data, such as multimodal interleaved documents, remains underutilized for vision-language representation learning. To fully leverage these unpaired documents, we initially establish a Real-World Data Extraction pipeline to extract high-quality images and texts. Then we design a hierarchical retrieval method to efficiently associate each image with multiple semantically relevant realistic texts. To further enhance fine-grained visual information, we propose an image semantic augmented generation module for synthetic text production. Furthermore, we employ a semantic balance sampling strategy to improve dataset diversity, enabling better learning of long-tail concepts. Based on these innovations, we construct RealSyn, a dataset combining realistic and synthetic texts, available in three scales: 15M, 30M, and 100M. Extensive experiments demonstrate that RealSyn effectively advances vision-language representation learning and exhibits strong scalability. Models pre-trained on RealSyn achieve state-of-the-art performance on multiple downstream tasks. To facilitate future research, the RealSyn dataset and pre-trained model weights are released at https://github.com/deepglint/RealSyn.

Medical Image Registration Meets Vision Foundation Model: Prototype Learning and Contour Awareness

Medical image registration is a fundamental task in medical image analysis, aiming to establish spatial correspondences between paired images. However, existing unsupervised deformable registration methods rely solely on intensity-based similarity metrics, lacking explicit anatomical knowledge, which limits their accuracy and robustness. Vision foundation models, such as the Segment Anything Model (SAM), can generate high-quality segmentation masks that provide explicit anatomical structure knowledge, addressing the limitations of traditional methods that depend only on intensity similarity. Based on this, we propose a novel SAM-assisted registration framework incorporating prototype learning and contour awareness. The framework includes: (1) Explicit anatomical information injection, where SAM-generated segmentation masks are used as auxiliary inputs throughout training and testing to ensure the consistency of anatomical information; (2) Prototype learning, which leverages segmentation masks to extract prototype features and aligns prototypes to optimize semantic correspondences between images; and (3) Contour-aware loss, a contour-aware loss is designed that leverages the edges of segmentation masks to improve the model's performance in fine-grained deformation fields. Extensive experiments demonstrate that the proposed framework significantly outperforms existing methods across multiple datasets, particularly in challenging scenarios with complex anatomical structures and ambiguous boundaries. Our code is available at https://github.com/HaoXu0507/IPMI25-SAM-Assisted-Registration.

NVS-SQA: Exploring Self-Supervised Quality Representation Learning for Neurally Synthesized Scenes without References

Neural View Synthesis (NVS), such as NeRF and 3D Gaussian Splatting, effectively creates photorealistic scenes from sparse viewpoints, typically evaluated by quality assessment methods like PSNR, SSIM, and LPIPS. However, these full-reference methods, which compare synthesized views to reference views, may not fully capture the perceptual quality of neurally synthesized scenes (NSS), particularly due to the limited availability of dense reference views. Furthermore, the challenges in acquiring human perceptual labels hinder the creation of extensive labeled datasets, risking model overfitting and reduced generalizability. To address these issues, we propose NVS-SQA, a NSS quality assessment method to learn no-reference quality representations through self-supervision without reliance on human labels. Traditional self-supervised learning predominantly relies on the "same instance, similar representation" assumption and extensive datasets. However, given that these conditions do not apply in NSS quality assessment, we employ heuristic cues and quality scores as learning objectives, along with a specialized contrastive pair preparation process to improve the effectiveness and efficiency of learning. The results show that NVS-SQA outperforms 17 no-reference methods by a large margin (i.e., on average 109.5% in SRCC, 98.6% in PLCC, and 91.5% in KRCC over the second best) and even exceeds 16 full-reference methods across all evaluation metrics (i.e., 22.9% in SRCC, 19.1% in PLCC, and 18.6% in KRCC over the second best).

Cross-View Consistency Regularisation for Knowledge Distillation

Knowledge distillation (KD) is an established paradigm for transferring privileged knowledge from a cumbersome model to a lightweight and efficient one. In recent years, logit-based KD methods are quickly catching up in performance with their feature-based counterparts. However, previous research has pointed out that logit-based methods are still fundamentally limited by two major issues in their training process, namely overconfident teacher and confirmation bias. Inspired by the success of cross-view learning in fields such as semi-supervised learning, in this work we introduce within-view and cross-view regularisations to standard logit-based distillation frameworks to combat the above cruxes. We also perform confidence-based soft label mining to improve the quality of distilling signals from the teacher, which further mitigates the confirmation bias problem. Despite its apparent simplicity, the proposed Consistency-Regularisation-based Logit Distillation (CRLD) significantly boosts student learning, setting new state-of-the-art results on the standard CIFAR-100, Tiny-ImageNet, and ImageNet datasets across a diversity of teacher and student architectures, whilst introducing no extra network parameters. Orthogonal to on-going logit-based distillation research, our method enjoys excellent generalisation properties and, without bells and whistles, boosts the performance of various existing approaches by considerable margins.

Gotta Hear Them All: Sound Source Aware Vision to Audio Generation

Vision-to-audio (V2A) synthesis has broad applications in multimedia. Recent advancements of V2A methods have made it possible to generate relevant audios from inputs of videos or still images. However, the immersiveness and expressiveness of the generation are limited. One possible problem is that existing methods solely rely on the global scene and overlook details of local sounding objects (i.e., sound sources). To address this issue, we propose a Sound Source-Aware V2A (SSV2A) generator. SSV2A is able to locally perceive multimodal sound sources from a scene with visual detection and cross-modality translation. It then contrastively learns a Cross-Modal Sound Source (CMSS) Manifold to semantically disambiguate each source. Finally, we attentively mix their CMSS semantics into a rich audio representation, from which a pretrained audio generator outputs the sound. To model the CMSS manifold, we curate a novel single-sound-source visual-audio dataset VGGS3 from VGGSound. We also design a Sound Source Matching Score to measure localized audio relevance. This is to our knowledge the first work to address V2A generation at the sound-source level. Extensive experiments show that SSV2A surpasses state-of-the-art methods in both generation fidelity and relevance. We further demonstrate SSV2A's ability to achieve intuitive V2A control by compositing vision, text, and audio conditions. Our SSV2A generation can be tried and heard at https://ssv2a.github.io/SSV2A-demo .

Cell as Point: One-Stage Framework for Efficient Cell Tracking

Cellular activities are dynamic and intricate, playing a crucial role in advancing diagnostic and therapeutic techniques, yet they often require substantial resources for accurate tracking. Despite recent progress, the conventional multi-stage cell tracking approaches not only heavily rely on detection or segmentation results as a prerequisite for the tracking stage, demanding plenty of refined segmentation masks, but are also deteriorated by imbalanced and long sequence data, leading to under-learning in training and missing cells in inference procedures. To alleviate the above issues, this paper proposes the novel end-to-end CAP framework, which leverages the idea of regarding Cell as Point to achieve efficient and stable cell tracking in one stage. CAP abandons detection or segmentation stages and simplifies the process by exploiting the correlation among the trajectories of cell points to track cells jointly, thus reducing the label demand and complexity of the pipeline. With cell point trajectory and visibility to represent cell locations and lineage relationships, CAP leverages the key innovations of adaptive event-guided (AEG) sampling for addressing data imbalance in cell division events and the rolling-as-window (RAW) inference method to ensure continuous tracking of new cells in the long term. Eliminating the need for a prerequisite detection or segmentation stage, CAP demonstrates strong cell tracking performance while also being 10 to 55 times more efficient than existing methods. The code and models will be released.

ORID: Organ-Regional Information Driven Framework for Radiology Report Generation

The objective of Radiology Report Generation (RRG) is to automatically generate coherent textual analyses of diseases based on radiological images, thereby alleviating the workload of radiologists. Current AI-based methods for RRG primarily focus on modifications to the encoder-decoder model architecture. To advance these approaches, this paper introduces an Organ-Regional Information Driven (ORID) framework which can effectively integrate multi-modal information and reduce the influence of noise from unrelated organs. Specifically, based on the LLaVA-Med, we first construct an RRG-related instruction dataset to improve organ-regional diagnosis description ability and get the LLaVA-Med-RRG. After that, we propose an organ-based cross-modal fusion module to effectively combine the information from the organ-regional diagnosis description and radiology image. To further reduce the influence of noise from unrelated organs on the radiology report generation, we introduce an organ importance coefficient analysis module, which leverages Graph Neural Network (GNN) to examine the interconnections of the cross-modal information of each organ region. Extensive experiments an1d comparisons with state-of-the-art methods across various evaluation metrics demonstrate the superior performance of our proposed method.

AMNCutter: Affinity-Attention-Guided Multi-View Normalized Cutter for Unsupervised Surgical Instrument Segmentation

Surgical instrument segmentation (SIS) is pivotal for robotic-assisted minimally invasive surgery, assisting surgeons by identifying surgical instruments in endoscopic video frames. Recent unsupervised surgical instrument segmentation (USIS) methods primarily rely on pseudo-labels derived from low-level features such as color and optical flow, but these methods show limited effectiveness and generalizability in complex and unseen endoscopic scenarios. In this work, we propose a label-free unsupervised model featuring a novel module named Multi-View Normalized Cutter (m-NCutter). Different from previous USIS works, our model is trained using a graph-cutting loss function that leverages patch affinities for supervision, eliminating the need for pseudo-labels. The framework adaptively determines which affinities from which levels should be prioritized. Therefore, the low- and high-level features and their affinities are effectively integrated to train a label-free unsupervised model, showing superior effectiveness and generalization ability. We conduct comprehensive experiments across multiple SIS datasets to validate our approach's state-of-the-art (SOTA) performance, robustness, and exceptional potential as a pre-trained model. Our code is released at https://github.com/MingyuShengSMY/AMNCutter.

TractShapeNet: Efficient Multi-Shape Learning with 3D Tractography Point Clouds

Brain imaging studies have demonstrated that diffusion MRI tractography geometric shape descriptors can inform the study of the brain's white matter pathways and their relationship to brain function. In this work, we investigate the possibility of utilizing a deep learning model to compute shape measures of the brain's white matter connections. We introduce a novel framework, TractShapeNet, that leverages a point cloud representation of tractography to compute five shape measures: length, span, volume, total surface area, and irregularity. We assess the performance of the method on a large dataset including 1065 healthy young adults. Experiments for shape measure computation demonstrate that our proposed TractShapeNet outperforms other point cloud-based neural network models in both the Pearson correlation coefficient and normalized error metrics. We compare the inference runtime results with the conventional shape computation tool DSI-Studio. Our results demonstrate that a deep learning approach enables faster and more efficient shape measure computation. We also conduct experiments on two downstream language cognition prediction tasks, showing that shape measures from TractShapeNet perform similarly to those computed by DSI-Studio. Our code will be available at: https://github.com/SlicerDMRI/TractShapeNet.

DINeuro: Distilling Knowledge from 2D Natural Images via Deformable Tubular Transferring Strategy for 3D Neuron Reconstruction

Reconstructing neuron morphology from 3D light microscope imaging data is critical to aid neuroscientists in analyzing brain networks and neuroanatomy. With the boost from deep learning techniques, a variety of learning-based segmentation models have been developed to enhance the signal-to-noise ratio of raw neuron images as a pre-processing step in the reconstruction workflow. However, most existing models directly encode the latent representative features of volumetric neuron data but neglect their intrinsic morphological knowledge. To address this limitation, we design a novel framework that distills the prior knowledge from a 2D Vision Transformer pre-trained on extensive 2D natural images to facilitate neuronal morphological learning of our 3D Vision Transformer. To bridge the knowledge gap between the 2D natural image and 3D microscopic morphologic domains, we propose a deformable tubular transferring strategy that adapts the pre-trained 2D natural knowledge to the inherent tubular characteristics of neuronal structure in the latent embedding space. The experimental results on the Janelia dataset of the BigNeuron project demonstrate that our method achieves a segmentation performance improvement of 4.53% in mean Dice and 3.56% in mean 95% Hausdorff distance.