<p>Deep Vein Thrombosis (DVT) is the formation of blood clots in the deep veins of the calf, requiring precise Computer Tomography (CT) scan segmentation for accurate diagnosis and treatment. We proposed and developed an efficient Federated Learning (FedL) architecture using the Federated Averaging (FedAvg) algorithm. Seven distinct local models were designed and trained on non-independent and identically distributed (Non-IID) CT images to maintain data privacy and security, enhancing DVT segmentation efficiency and accuracy. The global model was progressively improved by aggregating the local model’s weights using FedAvg algorithm. Our algorithm was evaluated in three phases using datasets of 1000, 2000, and 3000 samples to assess the global model’s performance. Phase 1 involved three clients, each with unique local models (Convolutional Neural Network (CNN), Sequential, and Semantic). While, Phase 2 expanded to five clients, incorporating additional models (U-Net and VGG Net-19). In Phase 3, scaled to seven clients with advanced models (Modified U-Net and Modified-Net). Empirical results across Phases 1–3 showed significant gains with increasing dataset size –attaining higher Accuracy (<InlineEquation ID="IEq1"> <EquationSource Format="TEX">\(0.9076{\rightarrow }0.9603\)</EquationSource> </InlineEquation>) and F1-score (<InlineEquation ID="IEq2"> <EquationSource Format="TEX">\(0.8889{\rightarrow }0.9521\)</EquationSource> </InlineEquation>), while Tversky Loss decreased to (<InlineEquation ID="IEq3"> <EquationSource Format="TEX">\(0.1111{\rightarrow }0.0479\)</EquationSource> </InlineEquation>). Notably, our framework proved consistent improvement across all phases, achieving a reduction in validation loss from 0.910 to 0.061 and a communication cost increase from 14 MB to 3279 MB with increasing model scales. The average training time rose proportionally (7.67 s <InlineEquation ID="IEq4"> <EquationSource Format="TEX">\({\rightarrow }\)</EquationSource> </InlineEquation> 18,702 s) while maintaining robust differential privacy preservation (<InlineEquation ID="IEq5"> <EquationSource Format="TEX">\(\epsilon\)</EquationSource> </InlineEquation> <InlineEquation ID="IEq6"> <EquationSource Format="TEX">\(\approx 1.24\)</EquationSource> </InlineEquation>) and improved client heterogeneity (<InlineEquation ID="IEq7"> <EquationSource Format="TEX">\(0.13\,{\rightarrow }\,0.51\)</EquationSource> </InlineEquation>), demonstrating our framework’s scalability and stability across heterogeneous environments.</p>

Privacy-aware deep vein thrombosis segmentation using a multi-model federated learning framework with the federated averaging algorithm

期刊论文
出版模式：

开放获取
发表日期：

2026年2月27日

查看全文

Pavihaa Lakshmi B,
Vidhya S

摘要

Deep Vein Thrombosis (DVT) is the formation of blood clots in the deep veins of the calf, requiring precise Computer Tomography (CT) scan segmentation for accurate diagnosis and treatment. We proposed and developed an efficient Federated Learning (FedL) architecture using the Federated Averaging (FedAvg) algorithm. Seven distinct local models were designed and trained on non-independent and identically distributed (Non-IID) CT images to maintain data privacy and security, enhancing DVT segmentation efficiency and accuracy. The global model was progressively improved by aggregating the local model’s weights using FedAvg algorithm. Our algorithm was evaluated in three phases using datasets of 1000, 2000, and 3000 samples to assess the global model’s performance. Phase 1 involved three clients, each with unique local models (Convolutional Neural Network (CNN), Sequential, and Semantic). While, Phase 2 expanded to five clients, incorporating additional models (U-Net and VGG Net-19). In Phase 3, scaled to seven clients with advanced models (Modified U-Net and Modified-Net). Empirical results across Phases 1–3 showed significant gains with increasing dataset size –attaining higher Accuracy ( \(0.9076{\rightarrow }0.9603\) ) and F1-score ( \(0.8889{\rightarrow }0.9521\) ), while Tversky Loss decreased to ( \(0.1111{\rightarrow }0.0479\) ). Notably, our framework proved consistent improvement across all phases, achieving a reduction in validation loss from 0.910 to 0.061 and a communication cost increase from 14 MB to 3279 MB with increasing model scales. The average training time rose proportionally (7.67 s \({\rightarrow }\) 18,702 s) while maintaining robust differential privacy preservation ( \(\epsilon\) \(\approx 1.24\) ) and improved client heterogeneity ( \(0.13\,{\rightarrow }\,0.51\) ), demonstrating our framework’s scalability and stability across heterogeneous environments.