Sketched Equivariant Imaging Regularization and Deep Internal Learning for Inverse Problems
This work addresses computational bottlenecks in unsupervised training for inverse imaging problems, offering incremental improvements for researchers and practitioners in medical imaging.
The authors tackled the computational inefficiency of Equivariant Imaging (EI) regularization in unsupervised deep imaging networks by proposing a sketched EI method using randomized sketching for acceleration, achieving significant computational speedups in X-ray CT and MRI reconstruction tasks.
Equivariant Imaging (EI) regularization has become the de-facto technique for unsupervised training of deep imaging networks, without any need of ground-truth data. Observing that the EI-based unsupervised training paradigm currently has significant computational redundancy leading to inefficiency in high-dimensional applications, we propose a sketched EI regularization which leverages the randomized sketching techniques for acceleration. We apply our sketched EI regularization to develop an accelerated deep internal learning framework, which can be efficiently applied for test-time network adaptation. Additionally, for network adaptation tasks, we propose a parameter-efficient approach to accelerate both EI and Sketched-EI via optimizing only the normalization layers. Our numerical study on X-ray CT and multicoil magnetic resonance image reconstruction tasks demonstrate that our approach can achieve significant computational acceleration over standard EI counterpart in single-input setting and network adaptation at test time.