Computer-implemented method for evaluating magnetic resonance data, magnetic resonance device, computer program, and electronically readable data medium

A magnetic resonance device, for evaluating magnetic resonance data of a scan region of a patient, comprising:

• a main magnet;
• and
• control circuitry configured to:
• acquire two image datasets identified with the scan region of the patient, the two image datasets being acquired using different echo times,
• generate, based upon the two image datasets, a least two combination datasets, each one of the two combination datasets being associated with magnetic resonance signals of a different spin species;
• and
• assign a spin species to each of the two combination datasets by:
• determining an input dataset based upon the two combination datasets;
• and
• evaluating the input dataset via an assignment algorithm that is trained in accordance with a trained function to determine assignment information by assigning, to each of the two combination datasets, a respective spin species,
• wherein the trained function comprises a pre-trained function that is trained with respect to spin species assignment via transfer learning.

A non-transitory computer readable medium having instructions stored thereon that, when executed by one or more processors of a magnetic resonance device, cause the magnetic resonance device to evaluate magnetic resonance data of a scan region of a patient by:

• acquiring two image datasets identified with the scan region of the patient, the two image datasets being acquired using different echo times;
• generating, based upon the two image datasets, two combination datasets, each one of the two combination datasets being associated with magnetic resonance signals of a different spin species;
• and
• assigning a spin species to each of the two combination datasets by:
• determining an input dataset based upon the two combination datasets;
• and
• evaluating the input dataset via an assignment algorithm that is trained in accordance with a trained function to determine assignment information by assigning, to each of the two combination datasets, a respective spin species,
• wherein the trained function comprises a pre-trained function that is trained with respect to spin species assignment via transfer learning.