FIRM: Federated Image Reconstruction using Multimodal Tomographic Data

TL;DR

This paper introduces a federated algorithm for reconstructing images from multimodal tomographic data, improving quality and efficiency while maintaining data decentralization.

Contribution

It presents a novel federated approach with an adaptive step-size rule for multimodal image reconstruction, extending to augmented Lagrangian methods.

Findings

Superior computational efficiency demonstrated

Enhanced image reconstruction quality shown

Guaranteed convergence of the method

Abstract

We propose a federated algorithm for reconstructing images using multimodal tomographic data sourced from dispersed locations, addressing the challenges of traditional unimodal approaches that are prone to noise and reduced image quality. Our approach formulates a joint inverse optimization problem incorporating multimodality constraints and solves it in a federated framework through local gradient computations complemented by lightweight central operations, ensuring data decentralization. Leveraging the connection between our federated algorithm and the quadratic penalty method, we introduce an adaptive step-size rule with guaranteed sublinear convergence and further suggest its extension to augmented Lagrangian framework. Numerical results demonstrate its superior computational efficiency and improved image reconstruction quality.