Transferable Optimization Network for Cross-Domain Image Reconstruction

TL;DR

This paper introduces a transfer learning framework with bi-level optimization to improve image reconstruction quality in data-scarce scenarios by leveraging heterogeneous datasets across domains.

Contribution

The paper presents a novel two-step transfer learning approach with bi-level optimization for cross-domain image reconstruction, enabling effective use of limited data.

Findings

High-quality reconstruction with limited data achieved

Universal feature extractor trained on diverse datasets

Effective domain adaptation demonstrated in MR imaging

Abstract

We develop a novel transfer learning framework to tackle the challenge of limited training data in image reconstruction problems. The proposed framework consists of two training steps, both of which are formed as bi-level optimizations. In the first step, we train a powerful universal feature-extractor that is capable of learning important knowledge from large, heterogeneous data sets in various domains. In the second step, we train a task-specific domain-adapter for a new target domain or task with only a limited amount of data available for training. Then the composition of the adapter and the universal feature-extractor effectively explores feature which serve as an important component of image regularization for the new domains, and this leads to high-quality reconstruction despite the data limitation issue. We apply this framework to reconstruct under-sampled MR images with limited data by using a collection of diverse data samples from different domains, such as images of other anatomies, measurements of various sampling ratios, and even different image modalities, including natural images. Experimental results demonstrate a promising transfer learning capability of the proposed method.