Providing a classification explanation and a generative function

An explanation system for providing a classification explanation, comprising:

• an input interface configured to receive an acquired medical image corresponding to an examination volume;
• a computation device including at least one processor,
• wherein at least one of the input interface and the computation unit device is configured to receive or determine an actual classification value, the actual classification value classifying an anatomical structure within the acquired medical image,
• wherein at least one of the input interface and the computation device are configured to receive or determine a first classification value, different from the actual classification value, the first classification value being a potential classification value of the anatomical structure within the acquired medical image,
• wherein the computation device is configured to apply a first generative function to the acquired medical image to generate a first synthetic image, the first generative function being based on the first classification value and being configured to modify the anatomical structure to exhibit the first classification value;
• and
• an output interface configured to provide a visualization based on the acquired medical image and the first synthetic image.

A computer-implemented method for providing a first generative function, comprising:

• receiving a first training image including a first training anatomical structure, the first training anatomical structure being classified by a first training classification value;
• receiving or determining a second training classification value, different from the first training classification value, the second training classification value being a potential classification value of the first anatomical structure within the first training medical image;
• applying a first generative function to the first training image to generate a modified first training image, the first generative function being based on the second training classification value;
• receiving a first input image, the first input image including a first input anatomical structure, the first input anatomical structure being classified by the second training classification value;
• applying a first discriminative function to the first input image to generate a first discriminator value;
• applying the first discriminative function to the modified first training image to generate a second discriminator value;
• adapting a parameter of at least one of the first generative function and the first discriminative function, based on an generative-adversarial loss function, the generative-adversarial loss function being based on the first discriminator value and the second discriminator value;
• and
• providing at least one of the first generative function and the first discriminative function.