Learning Data Triage: Linear Decoding Works for Compressive MRI
This approach could reduce computational costs for medical imaging applications, though it appears incremental as it builds on existing compressive sampling frameworks.
The paper tackles the problem of compressive MRI reconstruction by learning optimal sub-sampling patterns from training data instead of relying on known signal structures, and shows that simple linear reconstruction achieves competitive results with theoretical guarantees and experimental validation on real-world MRI data.
The standard approach to compressive sampling considers recovering an unknown deterministic signal with certain known structure, and designing the sub-sampling pattern and recovery algorithm based on the known structure. This approach requires looking for a good representation that reveals the signal structure, and solving a non-smooth convex minimization problem (e.g., basis pursuit). In this paper, another approach is considered: We learn a good sub-sampling pattern based on available training signals, without knowing the signal structure in advance, and reconstruct an accordingly sub-sampled signal by computationally much cheaper linear reconstruction. We provide a theoretical guarantee on the recovery error, and show via experiments on real-world MRI data the effectiveness of the proposed compressive MRI scheme.