The current factory mechanism takes the first registered file reader/writer that can be found for a desired file extension. For many extensions, the ITK readers/writers are chosen since they support a variety of image formats.
In order give other file-readers the chance to process a file, they should be prioritized over the default itk mechanism.
Current workaround for this problem is the usage of self-made file extensions that are only recognized by the own file io. However, many reader implementations might just be wrappers around itk-implementations (just producing mitk-objects) and thus should basically handle the same set of extensions.