According to an embodiment, an image processing device includes an obtainer, a determiner, a controller, and a generator. The obtainer obtains a position of an object to be observed in volume data of a medical image. The determiner determines a region of interest by using the position of the object and an instructed region inputted by a user in the volume data so that the region of interest includes at least part of the object. The controller controls a relation between the region of interest and a display range that indicates a range allowed to be displayed stereoscopically on a display. The generator generates a stereoscopic image of the volume data according to the relation between the region of interest and the display range.
