Hi Shuichi,
Check out this Documentation Page for more information about creating image datastores for image classification: https://www.mathworks.com/help/deeplearning/ug/create-and-explore-datastore-for-image-classification.html
One potential solution is to store your images in a folder containing subfolders for each class (i.e. diagnosis) in your data set. The name of each subfolder should match the desired label for the class. Store all images corresponding to a each class in the correct subdirectory. Then in MATLAB, when you create the imageDatastore object, specify the IncludeSubfolders name-value argument as true and the LabelSource name-value argument as "foldernames". This will label each image in the datastore with the name of the subfolder it is in. The labels will be visible in the Labels property.
