U–shaped networks have become prevalent in various medical image tasks such as segmentation, and restoration. However, most existing U-shaped networks rely on centralized learning which raises privacy concerns. To address these issues, federated learning (FL) and split learning (SL) have been proposed. However, achieving a balance between the local computational cost, model privacy, and parallel training remains a challenge. In this paper, we propose a novel hybrid learning paradigm called Dynamic Corrected Split Federated Learning (DC-SFL) for U-shaped medical image networks. To preserve data privacy, including the input, model parameters, label and output simultaneously, we propose to split the network into three parts hosted by different parties. We propose a Dynamic Weight Correction Strategy (DWCS) to stabilize the training process and avoid the model drift problem due to data heterogeneity. To further enhance privacy protection and establish a trustworthy distributed learning paradigm, we propose to introduce additively homomorphic encryption into the aggregation process of client-side model, which helps prevent potential collusion between parties and provides a better privacy guarantee for our proposed method. The proposed DC-SFL is evaluated on various medical image tasks, and the experimental results demonstrate its effectiveness. In comparison with state-of-the-art distributed learning methods, our method achieves competitive performance. Related codes will be released at https://github.com/Zi-YuanYang/DC-SFL .

Palmprint has gained popularity as a biometric modality and has recently attracted significant research interest. The competition-based method is the prevailing approach for hand-crafted palmprint recognition, thanks to its powerful discriminative ability to identify distinctive features. However, the competition mechanism possesses vast untapped advantages that have yet to be fully explored. In this paper, we reformulate the traditional competition mechanism and propose a C omprehensive C ompetition Net work (CCNet). The traditional competition mechanism focuses solely on selecting the winner of different channels without considering the spatial information of the features. Our approach considers the spatial competition relationships between features while utilizing channel competition features to extract a more comprehensive set of competitive features. Moreover, existing methods for palmprint recognition typically focus on first-order texture features without utilizing the higher-order texture feature information. Our approach integrates the competition process with multi-order texture features to overcome this limitation. CCNet incorporates spatial and channel competition mechanisms into multi-order texture features to enhance recognition accuracy, enabling it to capture and utilize palmprint information in an end-to-end manner efficiently. Extensive experimental results have shown that CCNet can achieve remarkable performance on four public datasets, showing that CCNet is a promising approach for palmprint recognition that can achieve state-of-the-art performance. Related codes will be released at https://github.com/Zi-YuanYang/CCNet.

In clinical practice, computed tomography (CT) is an important noninvasive inspection technology to provide patients’ anatomical information. However, its potential radiation risk is an unavoidable problem that raises people’s concerns. Recently, deep learning (DL)-based methods have achieved promising results in CT reconstruction, but these methods usually require the centralized collection of large amounts of data for training from specific scanning protocols, which leads to serious domain shift and privacy concerns. To relieve these problems, in this article, we propose a hypernetwork-based physics-driven personalized federated learning method (HyperFed) for CT imaging. The basic assumption of the proposed HyperFed is that the optimization problem for each domain can be divided into two subproblems: local data adaption and global CT imaging problems, which are implemented by an institution-specific physics-driven hypernetwork and a global-sharing imaging network, respectively. Learning stable and effective invariant features from different data distributions is the main purpose of global-sharing imaging network. Inspired by the physical process of CT imaging, we carefully design physics-driven hypernetwork for each domain to obtain hyperparameters from specific physical scanning protocol to condition the global-sharing imaging network, so that we can achieve personalized local CT reconstruction. Experiments show that HyperFed achieves competitive performance in comparison with several other state-of-the-art methods. It is believed as a promising direction to improve CT imaging quality and personalize the needs of different institutions or scanners without data sharing. Related codes have been released at https://github.com/Zi-YuanYang/HyperFed.

Palmprint recognition achieves high discrimination for identity verification. Compared with handcrafted local texture descriptors, convolutional neural networks (CNNs) can spontaneously learn optimal discriminative features without any prior knowledge. To further enhance the features’ representation and discrimination, we propose a coordinate-aware contrastive competitive neural network (CO ₃ Net) for palmprint recognition. To extract the multi-scale textures, CO ₃ Net consists of three parallel learnable Gabor filters (LGF)-based texture extraction branches that learn the discriminative and robust ordering features. Due to the heterogeneity of palmprints, the effects of different textures on the final recognition performance are inconsistent, and dynamically focusing on the textures is beneficial to the performance improvement. Then, CO ₃ Net introduces the attention modules to explore the spatial information, and selects more robust and discriminative textures. Specifically, coordinate attention is embedded into CO ₃ Net to adaptively focus on the important textures from the positional information. Since it is difficult for the cross-entropy loss to build a compact intra-class and separate inter-class feature space, the contrastive loss is employed to jointly optimize the network. CO ₃ Net is validated on four public datasets, and the results demonstrate the remarkable recognition performance of the proposed CO ₃ Net compared to other state-of-the-art methods.