Abstract: The present invention discloses a method for inspecting tubular-shaped structures (1) within a three-dimensional (3D) image data set, e.g. a vessel in a medical image. Initially, there is provided an image data set and performed a visualization of the image data set. Then, an inspection of the image data set is performed. During the inspection the user moves a pointer (P), e.g. via a computer mouse, and a processor performs a local segmentation around the pointer so as to determine a possible shape of a tubular shaped segmented object (1), e.g. a vessel, and the processor also makes a local analysis of the segmented object. Thereafter, a screen displays a view (P1) of the segmented object (1), where the orientation of the first view is derived from the local analysis; the first view can for example be cross-sectional or longitudinal views. The invention may be used directly on the raw image data in a great diversity of visualizations.