US 12,315,044 B2
Deep learning-based realtime reconstruction
Thomas Benkert, Neunkirchen am Brand (DE); Marcel Dominik Nickel, Herzogenaurach (DE); Simon Arberet, Princeton, NJ (US); Boris Mailhe, Plainsboro, NJ (US); and Mahmoud Mostapha, Princeton, NJ (US)
Assigned to Siemens Healthineers AG, Forchheim (DE)
Filed by Siemens Healthineers AG, Forchheim (DE)
Filed on Sep. 13, 2021, as Appl. No. 17/473,229.
Prior Publication US 2023/0084413 A1, Mar. 16, 2023
Int. Cl. G06T 11/00 (2006.01); G06N 3/08 (2023.01); G16H 30/20 (2018.01)
CPC G06T 11/003 (2013.01) [G06N 3/08 (2013.01); G16H 30/20 (2018.01); G06T 2210/41 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A method of reconstruction for a medical imaging system, the method comprising:
scanning, using a plurality of repetitions of a pulse sequence, a patient by the medical imaging system without the patient leaving the medical imaging system, the scanning acquiring in sequence at least a first subset of data using a first repetition of the plurality of repetitions of the pulse sequence and a second subset of data using a second repetition of the plurality of repetitions;
first reconstructing, after the first repetition is complete and during a same time period as when the second repetition of the scanning by the medical imaging system is performed, a first partial representation of an object of the patient from the first subset of the scan data, the first reconstructing being by, at least in part, a machine-learned model, the machine-learned model having an architecture for separate reconstruction of the first subset of data and the second subset of data;
second reconstructing, after the second repetition acquires the second subset of data, a second partial representation of the object from the second subset of the scan data, the second reconstruction being by, at least in part, the machine-learned model; and
generating an image of the object from the first partial representation and the second partial representation.