Method and device for segmenting medical image data
The present invention relates to a method and device for automatical segmentation of medical images of a non-planar surface of an object, in particular of a heart valve, including detecting a surface boundary which delimits the non-planar surface of the object, creating a model surface, which is spanned between the surface boundary, correcting the model surface by means of distance information containing information about the distance between the model surface and the non-planar surface of the object, until a corrected model surface is generated, and depicting the corrected model surface, where the correction of the model surface is preferably carried out by means of three- or four-dimensional image data sets, by orienting the three-dimensional volume-rendering of the object essentially perpendicular to the model surface, so that the information about the distance between the model surface and the non-planar surface of the object can be evaluated, until the corrected model surface has been produced.
1. A method for segmenting medical images of a non-planar surface of an object, the medical images being available as three or four-dimensional image sets, the method comprising the following steps:
detecting a surface boundary in three- or four-dimensional image data sets, which delimits the non-planar surface of the object,
creating a model surface, which is spanned between the surface boundary,
correcting the model surface by means of distance information, which contains information about distance(s) between the model surface and the non-planar surface of the object, until a corrected model surface is generated, wherein correcting the model surface by means of distance information is carried out by orienting a three-dimensional volume-rendering of the object essentially perpendicular or acute-angled to the model surface, wherein the volume renderer performing the volume rendering produces a depth-map, which is used to provide the distance information between the model surface and the non-planar surface of the object, and
depicting the corrected model surface,
wherein the correction of the model surface by means of the distance information is carried out iteratively, until the corrected model surface is generated, and
the distance information is at least partially obtained along tilted vectors which are acute-angled to the model surface, wherein the three-dimensional volume-rendering of the object is then oriented essentially parallel to the tilted vectors.
2. The method of claim 1 , wherein
the non-planar surface comprises at least one first subsurface and a second subsurface, which each are curved and separated from each other by means of a partition line, and wherein the corrected model surface is separated along the partition line into the at least two sub-surfaces prior to the depiction.
3. The method of claim 1 , wherein
the surface boundary is a closed line, so that the model surface spans the non-planar surface.
4. The method according to claim 1 , wherein the detection of the surface boundary is done by detecting land marks which delimit the non-planar surface and by interpolating these land marks to a contiguous line.
5. The method according to claim 1 , wherein
the correction of the model surface by means of distance information is carried out by orienting a three-dimensional volume-rendering of the object essentially perpendicular to the model surface, so that the information about the distance between the model surface and the non-planar surface of the object can be evaluated, until the corrected model surface is generated.
6. The method according to claim 1 , wherein
the distance information is gained along vectors standing essentially perpendicular to the model surface, by evaluating a volume rendering of a three-dimensional image data set of the non-planar surface accordingly.
7. The method to claim 1 , wherein the corrected model surface is separated prior to the depiction into the at least two sub-surfaces, wherein the separation is carried out along areas of greatest curvature, greatest distance to the model surface and/or along similar or essentially orthogonal surface normals.
8. The method of claim 7 ,
wherein the depiction of the separated, corrected model surfaces is carried out by means of a wire frame model, wherein also a transformation in time of these surfaces is continually depicted.
9. A non-transitory computer program product, containing program code stored on a non-transitory computer readable medium, the program code effecting the performance of a method when the program code is executed on a computer, the method comprising:
detecting a surface boundary in three- or four-dimensional image data sets, which delimits the non-planar surface of an object,
creating a model surface, which is spanned between the surface boundary,
correcting a surface of the model by means of distance information, which contains information about distance(s) between the model surface and the non-planar surface of the object, until a corrected model surface is generated, wherein correcting the model surface by means of distance information is carried out by orienting a three-dimensional volume-rendering of the object essentially perpendicular or acute-angled to the model surface, wherein the volume renderer performing the volume rendering produces a depth-map, which is used to provide the distance information between the model surface and the non-planar surface of the object, and
depicting the corrected model surface,
wherein the correction of the model surface by means of the distance information is carried out iteratively, until the corrected model surface is generated, and
the distance information is at least partially obtained along tilted vectors which are acute-angled to the model surface, wherein the three-dimensional volume-rendering of the object is then oriented essentially parallel to the tilted vectors.
10. A computer for segmenting medical images of a non-planar surface of an object by executing program code stored in a non-transitory computer-readable medium, the segmenting of the medical images performed according to a method comprising:
detecting, by the computer, a surface boundary in three- or four-dimensional image data sets, which delimits the non-planar surface of the object,
creating, by the computer, a model surface, which is spanned between the surface boundary,
correcting, by the computer, a surface of the model by means of distance information, which contains information about distance(s) between the model surface and the non-planar surface of the object, until a corrected model surface is generated, wherein correcting the model surface by means of distance information is carried out by orienting a three-dimensional volume-rendering of the object essentially perpendicular or acute-angled to the model surface, wherein the volume renderer performing the volume rendering produces a depth-map, which is used to provide the distance information between the model surface and the non-planar surface of the object, and
depicting, by the computer, the corrected model surface,
wherein the correction of the model surface by means of the distance information is carried out iteratively, until the corrected model surface is generated, and
the distance information is at least partially obtained along tilted vectors which are acute-angled to the model surface, wherein the three-dimensional volume-rendering of the object is then oriented essentially parallel to the tilted vectors.