IMAGE NORMALIZATION INCREASING ROBUSTNESS OF MACHINE LEARNING APPLICATIONS FOR MEDICAL IMAGES

摘要

The present invention relates to a computer program, a system and a method for normalizing medical images from a predetermined type of image acquisition device using a machine learning unit (ML), the method comprising the steps of:- Receiving (S1) a set of image data (S) with images (II to 19) at a decomposition unit (D), wherein the image data (S) have been generated by being converted from detector signals, acquired at a detector of an image acquisition device, wherein the detector signals are converted by means of applying different settings of the image acquisition device-specific processing algorithms;- At a decomposition unit (D): Decomposing (S2) each of the images (II to 19) of the set of images (S) into components (C1 to Cn) by incorporating at least information from the different settings of the image acquisition device-specific image processing algorithms;- At a normalizing unit (N): Normalizing (S3) each of said components (C1 to Cn) by means of a machine learning unit (ML) by processing at least information from the different settings of the image acquisition device-specific processing algorithms to provide a set of normalized images (CN1 to CNn) with a decreased variability score.