For example, JP2013-165765A discloses a method which detects a cerebral infarction part included in an MRI diffusion-weighted image (DWI), acquires, from an abnormal part of the diffusion-weighted image and a diffusion-weighted image of a healthy person, position conversion data required for anatomical registration therebetween, converts a single photon emission computed tomography (SPECT) image captured by a SPECT apparatus on the basis of the position conversion data such that the position of each tissue of the brain of the patient is matched with the position of each tissue of the brain of the healthy person, and discriminates the cerebral infarction part on the SPECT image. In addition, JP2018-505705A discloses a method which inputs an MR image, applies conversion using machine learning to the input MR image to generate a CT image, and performs diagnosis using the images of a plurality of modalities including the generated CT image.

Further, JP1996-251404A (JP-H08-251404A) discloses a discrimination apparatus that comprises first and second neural networks each of which includes an input layer, an intermediate layer, and an output layer and which are connected to each other such that an output from the input layer to the intermediate layer in the first neural network is input to the input layer of the second neural network. In the discrimination apparatus disclosed in JP1996-251404A (JP-H08-251404A), the discrimination result of a region attribute of image data is output on the basis of the image data input to the first neural network. The use of the discrimination apparatus makes it possible to discriminate specific medical characteristics included in the above-mentioned medical image.