numpy arrays are the common data format at runtime. a lot of libraries can deal with numpy arrays.
numpy itself lets you take a slice of an array. that is cropping.
OpenCV may not know how to read your type of medical image. there are other libraries specifically for reading and writing various image types.
OpenCV is also not an app you can interact with. if you need to drag a rectangle on your image for the cropping, that can be done because OpenCV supports some GUI by itself but that too isn’t its purpose and the GUI facilities provided by OpenCV are bare, not user-friendly.
are you saying it’s an image file with text drawn in it? why is text kept in an image file? is this hand-written or not?
if you need to perform optical character recognition, you’d best look for libraries for the purpose. OpenCV, among the contrib modules, has a text detector and an interface to tesseract for the OCR.