Method and apparatus for automated regularization tuning in magnetic resonance imaging (MRI) using compressed sensing (CS)
US11389077B2 · kind B2 · utility
Assignee
Inventors
Key dates
| Filing date | May 6, 2021 |
| Grant date | Jul 19, 2022 |
| Priority date | — |
| Expiry date | May 6, 2041 |
Classification
- Technology area (CPC G)Physics
- CPC primaryG16H30/40
- WIPO fieldMedical technology
- WIPO sectorInstruments
Abstract
An apparatus and method are provided to simultaneously provide good image quality and fast image reconstruction from magnetic resonance imaging (MRI) data by selecting an appropriate value for the regularization parameter used in compressed sensing (CS) image reconstruction. In CS reconstruction a high-resolution image can be reconstructed from randomized undersampled data by imposing sparsity in multi-scale transformation (e.g., wavelet) domain. Further, in the transformation domain, a threshold can be determined between signal and noise levels of the transform coefficients. A regularization parameter based on this threshold scales the regularization term, which imposes sparsity, relative to the data fidelity term in an objective function, thereby balancing the tradeoff between noise and smoothing.
Source: USPTO / EPO open patent data. Objective bibliographic and citation counts.