The scarcity data of medical field brings the collaborative training in medical vision-language pre-training (VLP) cross different clients. Therefore, the collaborative training in medical VLP faces two challenges: First, the medical data requires privacy, thus can not directly shared across different clients. Second, medical data distribution across institutes is typically heterogeneous, hindering local model alignment and representation capabilities. To simultaneously overcome these two challenges, we propose the framework called personalized model selector with fused multimodal information (PMS-FM). The contribution of PMS-FM is two-fold: 1) PMS-FM uses embeddings to represent information in different formats, allowing for the fusion of multimodal data. 2) PMS-FM adapts to personalized data distributions by training multiple models. A model selector then identifies and selects the best-performing model for each individual client. Extensive experiments with multiple real-world medical datasets demonstrate the superb performance of PMS-FM over existing federated learning methods on different zero-shot classification tasks.

Medical imaging inherently exhibits multi-scale characteristics, encompassing both global anatomical structures and localized pathological details. However, most existing multimodal large language models (MLLMs) for medical visual question answering (VQA) rely mainly on global features, limiting fine-grained reasoning across spatial levels. To address this, we propose MSFormer (Multi-Scale Transformer), a vision-language architecture that dynamically integrates hierarchical image features across multiple scales. MSFormer extracts multi-resolution embeddings via a vision backbone and refines them using a Multi-Scale Positional Embedding (MSPE) module to maintain spatial alignment. Its core Multi-Scale Grouped Attention (MSGA) mechanism enables learnable queries to jointly attend to features from different scales, adaptively focusing on context relevant to each question. Through contrastive pretraining, instruction tuning, and fine-tuning, MSFormer effectively aligns multi-scale visual and textual representations, substantially improving both open- and closed-ended medical VQA. Extensive experiments show that MSFormer consistently surpasses prior state-of-the-art models, underscoring the value of scale-aware visual prompting for enhanced interpretability and clinical reasoning in multimodal medical AI.

Accurately measuring the evolution of Multiple Sclerosis (MS) with magnetic resonance imaging (MRI) critically informs understanding of disease progression and helps to direct therapeutic strategy. Deep learning models have shown promise for automatically segmenting MS lesions, but the scarcity of accurately annotated data hinders progress in this area. Obtaining sufficient data from a single clinical site is challenging and does not address the heterogeneous need for model robustness. Conversely, the collection of data from multiple sites introduces data privacy concerns and potential label noise due to varying annotation standards. To address this dilemma, we explore the use of the federated learning framework while considering label noise. Our approach enables collaboration among multiple clinical sites without compromising data privacy under a federated learning paradigm that incorporates a noise-robust training strategy based on label correction. Specifically, we introduce a Decoupled Hard Label Correction (DHLC) strategy that considers the imbalanced distribution and fuzzy boundaries of MS lesions, enabling the correction of false annotations based on prediction confidence. We also introduce a Centrally Enhanced Label Correction (CELC) strategy, which leverages the aggregated central model as a correction teacher for all sites, enhancing the reliability of the correction process. Extensive experiments conducted on two multi-site datasets demonstrate the effectiveness and robustness of our proposed methods, indicating their potential for clinical applications in multi-site collaborations to train better deep learning models with lower cost in data collection and annotation.

We address temporal action localization in untrimmed long videos. This is important because videos in real applications are usually unconstrained and contain multiple action instances plus video content of background scenes or other activities. To address this challenging issue, we exploit the effectiveness of deep networks in temporal action localization via three segment-based 3D ConvNets: (1) a proposal network identifies candidate segments in a long video that may contain actions, (2) a classification network learns one-vs-all action classification model to serve as initialization for the localization network, and (3) a localization network fine-tunes the learned classification network to localize each action instance. We propose a novel loss function for the localization network to explicitly consider temporal overlap and achieve high temporal localization accuracy. In the end, only the proposal network and the localization network are used during prediction. On two largescale benchmarks, our approach achieves significantly superior performances compared with other state-of-the-art systems: mAP increases from 1.7% to 7.4% on MEXaction2 and increases from 15.0% to 19.0% on THUMOS 2014.