Patent · US Active

Reconstruction of magnetic-resonance datasets using machine learning

US11360176B2 · kind B2 · utility

1Cited by

0References

19Claims

0Family size

Assignees

Siemens Healthcare Limited · DE
The General Hospital Corporation · US

Inventors

Daniel Polak · Cambridge, US
Kawin Setsompop · Boston, US

Key dates

Filing date	Jan 31, 2020
Grant date	Jun 14, 2022
Priority date	—
Expiry date	Apr 27, 2040

Classification

Technology area (CPC G)Physics
CPC primaryG06N3/08
WIPO fieldComputer technology
WIPO sectorElectrical engineering

Abstract

A method includes determining an initial magnetic resonance imaging, MRI, dataset (201) in image domain based on an initial reconstruction of MRI measurement data obtained using an undersampling scheme (400); and determining patches (231-233) of the initial MRI dataset (201) in accordance with a patching scheme, the patching scheme depending on the undersampling scheme (400); and, for each one of the patches (231-233): applying a machine-learned algorithm to obtain a respective patch (231-233) of a reconstructed MRI dataset, the machine-learned algorithm depending on the undersampling scheme (400); and combining the patches (231-233) of the reconstructed MRI dataset.

Source: USPTO / EPO open patent data. Objective bibliographic and citation counts.