D2IP: Deep Dynamic Image Prior for 3D Time-sequence Pulmonary Impedance Imaging

2025-07-18

AI Summary

To address the high computational cost and limited feasibility of 3D time-resolved dynamic imaging in unsupervised tomographic reconstruction, this paper proposes D2IP—a novel unsupervised framework for dynamic electrical impedance tomography (EIT). D2IP reduces iteration count via unsupervised parameter warm-starting, enforces inter-frame consistency through a temporal parameter propagation mechanism, and employs a lightweight 3D-FastResUNet architecture for efficient reconstruction. Integrating deep image prior, temporal modeling, and 3D convolutions—without requiring any labeled training data—it significantly improves both reconstruction speed and fidelity. Evaluated on simulated and clinical pulmonary impedance datasets, D2IP achieves a 7.1× speedup over state-of-the-art unsupervised methods, with average MSSIM improved by 24.8% and ERR reduced by 8.1%. To our knowledge, D2IP is the first unsupervised method enabling millisecond-level 3D dynamic reconstruction while preserving high structural fidelity.

Technology Category

• Computer Vision: 3D Computer Vision
• Intelligent Robots: State Estimation
• Machine Learning: Unsupervised & Self-Supervised Learning

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Responsible Web: Data and user privacy-enhancing technologies for the Web
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets

Abstract

Unsupervised learning methods, such as Deep Image Prior (DIP), have shown great potential in tomographic imaging due to their training-data-free nature and high generalization capability. However, their reliance on numerous network parameter iterations results in high computational costs, limiting their practical application, particularly in complex 3D or time-sequence tomographic imaging tasks. To overcome these challenges, we propose Deep Dynamic Image Prior (D2IP), a novel framework for 3D time-sequence imaging. D2IP introduces three key strategies - Unsupervised Parameter Warm-Start (UPWS), Temporal Parameter Propagation (TPP), and a customized lightweight reconstruction backbone, 3D-FastResUNet - to accelerate convergence, enforce temporal coherence, and improve computational efficiency. Experimental results on both simulated and clinical pulmonary datasets demonstrate that D2IP enables fast and accurate 3D time-sequence Electrical Impedance Tomography (tsEIT) reconstruction. Compared to state-of-the-art baselines, D2IP delivers superior image quality, with a 24.8% increase in average MSSIM and an 8.1% reduction in ERR, alongside significantly reduced computational time (7.1x faster), highlighting its promise for clinical dynamic pulmonary imaging.

Problem

• Accelerates 3D time-sequence imaging convergence
• Reduces computational costs in dynamic tomography
• Improves pulmonary impedance reconstruction accuracy

Innovation

• Unsupervised Parameter Warm-Start accelerates convergence
• Temporal Parameter Propagation enforces temporal coherence
• 3D-FastResUNet backbone improves computational efficiency