FedSegNet: A Federated Learning Framework for 3D Medical Image Segmentation

Gliomas, the most aggressive primary tumors in the central nervous system, are characterized by high morphological heterogeneity and diffusely infiltrating boundaries. Such complexity poses significant challenges for accurate segmentation in clinical practice. Although deep learning methods have shown promising results, they often struggle to achieve a satisfactory trade-off among precise boundary delineation, robust multi-scale feature representation, and computational efficiency, particularly when processing high-resolution three-dimensional (3D) magnetic resonance imaging (MRI) data. Therefore, the aim of this study is to develop a novel 3D segmentation framework that specifically addresses these challenges, thereby improving clinical utility in brain tumor analysis. To accomplish this, we propose a multi-level channel-spatial attention and light-weight scale-fusion network (MCSLF-Net), which integrates a multi-level channel-spatial attention mechanism (MCSAM) and a light-weight scale-fusion module. By strategically enhancing subtle boundary features while maintaining a compact network design, our approach seeks to achieve high accuracy in delineating complex glioma morphologies, reduce computational burden, and provide a more clinically feasible segmentation solution. We propose MCSLF-Net, a network integrating two key components: (I) MCSAM: by strategically inserting a 3D channel-spatial attention module at critical semantic layers, the network progressively emphasizes subtle, infiltrative edges and small, easily overlooked contours. This avoids reliance on an additional edge detection branch while enabling fine-grained localization in ambiguous transitional regions. (II) Light-weight scale fusion unit (LSFU): leveraging depth-wise separable convolutions combined with multi-scale atrous (dilated) convolutions, LSFU enhances computational efficiency and adapts to varying feature requirements at different network depths. In doing so, it effectively captures small infiltrative lesions as well as extensive tumor areas. By coupling these two modules, MCSLF-Net balances global contextual information with local fine-grained features, simultaneously reducing the computational burden typically associated with 3D medical image segmentation. Extensive experiments on the BraTS 2019, BraTS 2020, and BraTS 2021 datasets validated the effectiveness of our approach. On BraTS 2021, MCSLF-Net achieved a mean Dice similarity coefficient (DSC) of 0.8974 and a mean 95th percentile Hausdorff distance (HD95) of 2.52 mm. Notably, it excels in segmenting intricate transitional areas, including the enhancing tumor (ET) region and the tumor core (TC), thereby demonstrating superior boundary delineation and multi-scale feature fusion capabilities relative to existing methods. These findings underscore the clinical potential of deploying multi-level channel-spatial attention and light-weight multi-scale fusion strategies in high-precision 3D glioma segmentation. By striking an optimal balance among boundary accuracy, multi-scale feature capture, and computational efficiency, the proposed MCSLF-Net offers a practical framework for further advancements in automated brain tumor analysis and can be extended to a range of 3D medical image segmentation tasks.

The convergence of vehicular technology, artificial intelligence (AI), and distributed computing has catalyzed the emergence of the Internet of Vehicles (IoV) as a cornerstone of next-generation intelligent transportation systems (ITS). By enabling vehicle-to-everything (V2X) communication, IoV supports cooperative perception, real-time decision-making, and autonomous driving. However, the reliance on large-scale, data-driven intelligence in IoV exposes systems to critical challenges, including adversarial poisoning, privacy leakage, identity forgery, and the fragility of centralized learning architectures. Federated Learning (FL) has been proposed as a promising paradigm to alleviate some of these issues by enabling distributed model training without centralizing sensitive vehicular data. Nonetheless, conventional FL remains vulnerable to security and trust limitations, particularly in dynamic vehicular environments. This thesis addresses these challenges by designing secure, privacy-preserving, and scalable FL frameworks that leverage distributed ledger technologies and cutting-edge security mechanisms.The thesis advances knowledge through four interconnected contributions. First, two novel optimization-driven poisoning attack models are introduced: PA-PSOSA and PAPSOGA, which combine particle swarm optimization with simulated annealing and genetic algorithms, respectively. These models demonstrate that even a small poisoning budget can substantially degrade global model utility under black-box and clean-label constraints, highlighting the urgency of robust defenses in vehicular FL. Second, a permissioned blockchain-enabled FL (BCFL) framework is proposed, in which consortium edge nodes running Practical Byzantine Fault Tolerance (PBFT) consensus replace the central aggregator. With blockchain integration and data validation mechanisms, this design ensures identity authentication, verifiable audit trails, and improved resilience against poisoning and Sybil attacks, while maintaining high model accuracy under adversarial conditions. Third, the framework is further enhanced to achieve inference-resistance by integrating secure aggregation (SecAgg) and differential privacy (DP), and lightweight with off-chain commitments. This design significantly reduces ledger storage requirements, increases system throughput, and mitigates inference-based privacy risks. Finally, to overcome the scalability limitations of PBFT-based BCFL, a DAG-enabled FL (DFL) framework is developed. By leveraging parallel validation, utility-score-based tip selection, and reputation-weighted aggregation, this framework significantly improves scalability, reduces communication complexity, and enhances robustness in asynchronous vehicular environments.Together, these contributions articulate a coherent progression from exposing vulnerabilities in vehicular FL to constructing secure, privacy-preserving, and scalable frameworks tailored for IoV ecosystems. The findings demonstrate that interdisciplinary integration of optimization theory, cryptography, differential privacy, and distributed ledger technologies is indispensable for trustworthy vehicular intelligence. Beyond theoretical significance, the proposed frameworks offer practical designs for deployment in safety-critical IoV environments. Future research directions include the integration of zero-knowledge proofs (ZKP) for verifiable privacy, adaptive defenses against evolving adversarial strategies, and experimental validation in real-world vehicular testbeds. Collectively, this thesis establishes a foundation for secure federated intelligence in IoV, contributing to the reliability, efficiency, and trustworthiness of next-generation ITS.

Federated learning (FL) offers an effective learning architecture to protect data privacy in a distributed manner. However, the inevitable network asynchrony, overdependence on a central coordinator, and lack of an open and fair incentive mechanism collectively hinder FL's further development. We propose IronForge, a new generation of FL framework, that features a directed acyclic graph (DAG)-based structure, where nodes represent uploaded models, and referencing relationships between models form the DAG that guides the aggregation process. This design eliminates the need for central coordinators to achieve fully decentralized operations. IronForge runs in a public and open network and launches a fair incentive mechanism by enabling state consistency in the DAG. Hence, the system fits in networks where training resources are unevenly distributed. In addition, dedicated defense strategies against prevalent FL attacks on incentive fairness and data privacy are presented to ensure the security of IronForge. Experimental results based on a newly developed test bed FLSim highlight the superiority of IronForge to the existing prevalent FL frameworks under various specifications in performance, fairness, and security. To the best of our knowledge, IronForge is the first secure and fully decentralized FL (DFL) framework that can be applied in open networks with realistic network and training settings.

Despite significant progress in federated learning (FL), persistent challenges, such as stragglers, adversaries, and communication costs remain. To address these issues, we propose FedLSC, a novel FL framework that leverages layer-selected correlation (LSC) to enhance both robustness and efficiency. In contrast to the existing methods, FedLSC does not rely on public data during model training, making it more practical and resilient in real-world scenarios. FedLSC introduces three key innovations: 1) preprocessing of layer selection (LS), which identifies significant layers to reduce communication costs and performance degradation; 2) local updates using LS-based scaled sign-stochastic gradient descent (SSS), introducing a layer-specific scaling mechanism to mitigate performance loss from quantization and significantly reduce communication costs; and 3) model aggregation via LSC-based schemes, which enhances robustness by processing only the significant layers and mitigating the impact of stragglers and adversaries. Furthermore, integrating the SSS scheme into FedLSC reduces communication costs to as little as 0.01% of those in state-of-the-art (SOTA) method while maintaining performance. Evaluations conducted across various FL scenarios show that FedLSC effectively supports robust performance and efficiency, even in bandwidth-constrained environments, thereby confirming its practicality in modern FL applications.

We systematically evaluate a Deep Learning model in a 3D medical image segmentation task. With our model, we address the flaws of manual segmentation: high inter-rater contouring variability and time consumption of the contouring process. The main extension over the existing evaluations is the careful and detailed analysis that could be further generalized on other medical image segmentation tasks. Firstly, we analyze the changes in the inter-rater detection agreement. We show that the model reduces the number of detection disagreements by [Formula: see text] [Formula: see text]. Secondly, we show that the model improves the inter-rater contouring agreement from [Formula: see text] to [Formula: see text] surface Dice Score [Formula: see text]. Thirdly, we show that the model accelerates the delineation process between [Formula: see text] and [Formula: see text] times [Formula: see text]. Finally, we design the setup of the clinical experiment to either exclude or estimate the evaluation biases; thus, preserving the significance of the results. Besides the clinical evaluation, we also share intuitions and practical ideas for building an efficient DL-based model for 3D medical image segmentation.

Aggregating multi-level feature representation plays a critical role in achieving robust volumetric medical image segmentation, which is important for the auxiliary diagnosis and treatment. Unlike the recent neural architecture search (NAS) methods that typically searched the optimal operators in each network layer, but missed a good strategy to search for feature aggregations, this paper proposes a novel NAS method for 3D medical image segmentation, named UXNet, which searches both the scale-wise feature aggregation strategies as well as the block-wise operators in the encoder-decoder network. UXNet has several appealing benefits. (1) It significantly improves flexibility of the classical UNet architecture, which only aggregates feature representations of encoder and decoder in equivalent resolution. (2) A continuous relaxation of UXNet is carefully designed, enabling its searching scheme performed in an efficient differentiable manner. (3) Extensive experiments demonstrate the effectiveness of UXNet compared with recent NAS methods for medical image segmentation. The architecture discovered by UXNet outperforms existing state-of-the-art models in terms of Dice on several public 3D medical image segmentation benchmarks, especially for the boundary locations and tiny tissues. The searching computational complexity of UXNet is cheap, enabling to search a network with best performance less than 1.5 days on two TitanXP GPUs.

Federated learning (FL) offers a robust approach for privacy-preserving model training across distributed clients, but its performance often declines significantly under non-independent and identically distributed data, a critical challenge in real-world scenarios. Existing methods frequently face issues such as slow convergence, reliance on heuristic regularization, or substantial communication overhead. To address these challenges, we propose SelfDistillCore, a federated learning framework that combines a sparsification technique with a momentum-based aggregation rule using an exponentially weighted moving average. By transmitting only the top K% of parameter updates and using momentum to balance historical and new updates, SelfDistillCore mitigates client drift and reduces communication costs. Experimental results show that SelfDistillCore significantly enhances convergence stability, improves resilience to data heterogeneity, increases scalability, and reduces communication costs by 19% to 70% compared to standard baselines. The code of SelfDistillCore can be found at https://github.com/dongld-2020/selfdistillcore_pj .

Objective. Over the past years, convolutional neural networks based methods have dominated the field of medical image segmentation. But the main drawback of these methods is that they have difficulty representing long-range dependencies. Recently, the Transformer has demonstrated super performance in computer vision and has also been successfully applied to medical image segmentation because of the self-attention mechanism and long-range dependencies encoding on images. To the best of our knowledge, only a few works focus on cross-modalities of image segmentation using the Transformer. Hence, the main objective of this study was to design, propose and validate a deep learning method to extend the application of Transformer to multi-modality medical image segmentation. Approach. This paper proposes a novel automated multi-modal Transformer network termed AMTNet for 3D medical image segmentation. Especially, the network is a well-modeled U-shaped network architecture where many effective and significant changes have been made in the feature encoding, fusion, and decoding parts. The encoding part comprises 3D embedding, 3D multi-modal Transformer, and 3D Co-learn down-sampling blocks. Symmetrically, the 3D Transformer block, upsampling block, and 3D-expanding blocks are included in the decoding part. In addition, a Transformer-based adaptive channel interleaved Transformer feature fusion module is designed to fully fuse features of different modalities. Main results. We provide a comprehensive experimental analysis of the Prostate and BraTS2021 datasets. The results show that our method achieves an average DSC of 0.907 and 0.851 (0.734 for ET, 0.895 for TC, and 0.924 for WT) on these two datasets, respectively. These values show that AMTNet yielded significant improvements over the state-of-the-art segmentation networks. Significance. The proposed 3D segmentation network exploits complementary features of different modalities during the feature extraction process at multiple scales to increase the 3D feature representations and improve the segmentation efficiency. This powerful network enriches the research of the Transformer to multi-modal medical image segmentation.

Federated Learning (FL) aims to protect data privacy by enabling clients to collectively train machine learning models without sharing their raw data. However, recent studies demonstrate that information exchanged during FL is subject to Gradient Inversion Attacks (GIA) and, consequently, a variety of privacy-preserving methods have been integrated into FL to thwart such attacks, such as Secure Multi-party Computing (SMC), Homomorphic Encryption (HE), and Differential Privacy (DP). Despite their ability to protect data privacy, these approaches inherently involve substantial privacy-utility trade-offs. By revisiting the key to privacy exposure in FL under GIA, which lies in the frequent sharing of model gradients that contain private data, we take a new perspective by designing a novel privacy preserve FL framework that effectively ``breaks the direct connection'' between the shared parameters and the local private data to defend against GIA. Specifically, we propose a Hypernetwork Federated Learning (HyperFL) framework that utilizes hypernetworks to generate the parameters of the local model and only the hypernetwork parameters are uploaded to the server for aggregation. Theoretical analyses demonstrate the convergence rate of the proposed HyperFL, while extensive experimental results show the privacy-preserving capability and comparable performance of HyperFL.

Three-dimensional (3D) medical image segmentation plays a crucial role in medical care applications. Although various two-dimensional (2D) and 3D neural network models have been applied to 3D medical image segmentation and achieved impressive results, a trade-off remains between efficiency and accuracy. To address this issue, a novel mixture convolutional network (MixConvNet) is proposed, in which traditional 2D/3D convolutional blocks are replaced with novel MixConv blocks. In the MixConv block, 3D convolution is decomposed into a mixture of 2D convolutions from different views. Therefore, the MixConv block fully utilizes the advantages of 2D convolution and maintains the learning ability of 3D convolution. It acts as 3D convolutions and thus can process volumetric input directly and learn intra-slice features, which are absent in the traditional 2D convolutional block. By contrast, the proposed MixConv block only contains 2D convolutions; hence, it has significantly fewer trainable parameters and less computation budget than a block containing 3D convolutions. Furthermore, the proposed MixConvNet is pre-trained with small input patches and fine-tuned with large input patches to improve segmentation performance further. In experiments on the Decathlon Heart dataset and Sliver07 dataset, the proposed MixConvNet outperformed the state-of-the-art methods such as UNet3D, VNet, and nnUnet.

Federated learning (FL) has emerged as a promising framework for training shared models across diverse participants, ensuring data remains securely stored on local devices. Despite its potential, FL still faces some critical challenges, including data heterogeneity, privacy risks, and substantial communication overhead. Current privacy-preserving FL research frequently fails to tackle complexities posed by heterogeneous data adequately, hence increasing communication expenses. To tackle these issues, we propose a top-k shuffled differential privacy FL (TopkSDP-FL) framework tailored to heterogeneous data environments. To address the model drift issue effectively, we design a novel regularization for local training, drawing inspiration from contrastive learning. To enhance efficiency, we propose a bidirectional top-k communication mechanism that reduces uplink and downlink overhead while strengthening privacy protection through double amplification with the shuffle model. Additionally, we shuffle all local gradient parameters at the layer level to address privacy budget concerns associated with high-dimensional aggregation and repeated iterations. Finally, a formal privacy analysis confirms the privacy amplification effect of TopkSDP-FL. The experimental results further demonstrate its superiority over other state-of-the-art FL methods, with an average accuracy improvement of 3% compared to FedAvg and other leading algorithms under the non-IID scenario, while also reducing communication costs by over 90%.

Proposes a methodology for 3D medical image segmentation based on the integration of 3D deformable and Markov random field models. Our method makes use of Markov random field theory to build Gibbs Prior models for the 3D medical image with arbitrary initial parameters to estimate the organ boundary. Then we use a 3D deformable model to fit the estimated boundary under the influence of gradient information in the initial 3D image and the balloon force. The result of the deformable model fit is used to update the Gibbs Prior model parameters, such as the gradient threshold of a boundary. Based on the updated parameters we restart the Gibbs Prior models. By integrating these processes recursively we achieve an automated segmentation of the initial 3D images. Our segmentation solution greatly reduces the time for 3D segmentation process and is capable of getting out of local minim. Results of the method are presented for several examples, including some MRI images with significant amount of noise.

Brain tumor segmentation is the most important way to separate tumor areas from healthy brain tissue in medical imaging. This is necessary for making an accurate diagnosis and planning treatment. But building strong deep learning models is often hard because there isn't much labeled medical data available, and strict privacy rules stop data from being shared in one place. Federated Learning (FL) helps keep patient data private by keeping it local, but its performance often drops when data from different hospitals have big differences in quality, imaging protocols, and distribution. Our research seeks to create a privacy-preserving federated learning framework that adeptly manages significant data heterogeneity while ensuring high segmentation accuracy across various institutions. We propose a new two-stage FL framework that allows multiple institutions to work together while keeping their privacy and effectively dealing with complicated non-IID data distributions. To start, we use a Federated Masked Autoencoder (MAE) for self-supervised pre-training. This lets the model learn strong anatomical features from unlabeled MRI scans. Second, the model is carefully fine-tuned using an Attention ResUNet3D architecture to get very accurate tumor segmentation. We use the SCAFFOLD optimization algorithm to keep training stable across all clients, even when the scanner varies from site to site, thereby directly addressing client drift. We also use strategic foreground-biased sampling and Test-Time Augmentation (TTA) techniques to greatly improve segmentation accuracy in difficult, uneven tumor sub-regions. We ran extensive experiments on the BraTS 2024 dataset in simulated federated settings with 10, 50, and 100 different clients. The Dice coefficients we got were 0.826, 0.824, and 0.818, which demonstrate strong performance. In the end, these strong results show that the suggested method works well on a larger scale and can be used in a clinical setting.

The trade-off between privacy and accuracy presents a challenge for current federated learning (FL) frameworks, hindering their progress from theory to application. The main issues with existing FL frameworks stem from a lack of interpretability and targeted privacy protections. To cope with these, we proposed Disentangled Federated Learning for Privacy (DFLP) which employes disentanglement, one of interpretability techniques, in private FL frameworks. Since sensitive properties are client-specific in nature, our main idea is to turn this feature into a tool that strikes the balance between data privacy and FL model performance, enabling the sensitive attributes to be private. DFLP disentangles the client-specific and class-invariant attributes to mask the sensitive attributes precisely. To our knowledge, this is the first work that successfully integrates disentanglement and the nature of sensitive attributes to achieve privacy protection while ensuring high FL model performance. Extensive experiments validate that disentanglement is an effective method for accuracy-aware privacy protection in FL frameworks.

Alzheimer's disease is a severe neurodegenerative disorder that primarily affects the elderly. Early detection is crucial for enabling timely interventions and slowing the disease's progression. A promising approach for early detection involves analyzing audio data collected from elderly individuals in their homes using Internet of Things (IoT) devices. However, this method presents significant challenges concerning privacy and data security. This paper introduces EDP-AD (Efficient Differential Privacy-Alzheimer's Detection), an efficient and privacy-preserving system. The system utilizes small IoT devices to collect audio data from elderly individuals. By integrating machine learning, federated learning (FL) and differential privacy (DP) techniques, EDP-AD ensures that raw audio data remains local while breaking down data silos, thereby preventing potential privacy leaks during the disease detection process. To enhance system efficiency, a sparse mask updating algorithm based on Top-k is proposed. This algorithm reduces communication overhead by sparsifying the model parameters uploaded by clients within the FL framework. Evaluation on a real-world dataset shows that the system achieves an accuracy rate of 84.48%, reduces communication costs by two-thirds, and provides robust privacy protection while maintaining high efficiency.