| 摘要 |
<p>The invention relates to a method and an apparatus for segmentation of an object in a 2D or 3D image data set by extracting a path along the object. In order to obtain a higher accuracy and reliability, the method according to the invention comprises the steps of: a) selecting a start point (5) of the path (2) as first active point, b) adapting an adaptable model (40) to the object (1) in a first active section around the start point (5), c) finding the next point (58) of the selected region by the steps of: c1) copying the geometric model parameters of the adapted model (40) of the active point to a plurality of neighboring points (51-58) of the active point (5), c2) orienting a model (32) around each of said plurality of neighboring points (51-58) using said copied model parameters and, searching the closest object points (21-26) around each of said neighboring points (51-58) based on said model, c3) adapting the models to the found object points (21-26) for each neighboring point (51-58), c4) selecting the neighboring point (58) for which the adapted model (48) fits best to the object (1) as next point of the selected region and as next active point, and d) repeating step c) until an end point of the path or a predetermined number of iterations is reached. The invention may preferably be applied for vessel segmentation to create maximum intensity projections with overlapping/occluding vessels removed.</p> |
