Proposed concepts aim to provide schemes, solutions, concepts, designs, methods, systems, and a computer program pertaining to predicting a field perturbation map for magnetic resonance imaging (MRI) of a subject. In particular, the invention aims to provide a field perturbation map of the subject without the need for additional time-consuming scans of the subject. An accurate field perturbation map is necessary in order to obtain an MRI scan of the subject of high quality. Accordingly, a synthetic computed tomography (CT) image is generated by inputting an initial MRI magnitude image of the subject to an image conversion machine learning algorithm. Subsequently, a weighted susceptibility map of the subject is determined based on the synthetic CT image and the initial MRI magnitude image, which is in turn used to determine the field perturbation map of the subject.