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

TL;DR

D2IP is a novel deep learning framework that significantly accelerates and improves 3D time-sequence pulmonary impedance imaging by combining innovative strategies for faster convergence, temporal coherence, and computational efficiency.

Contribution

The paper introduces D2IP, a new framework with strategies like UPWS, TPP, and a lightweight backbone to enhance 3D dynamic imaging speed and quality without extensive training data.

Findings

Achieves 24.8% higher MSSIM than baselines.

Reduces ERR by 8.1% compared to state-of-the-art methods.

Operates 7.1 times faster in reconstruction.

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.