Radiotherapeutic treatment planning apparatus and methods
A treatment planning apparatus comprises a means for receiving a three dimensional representation of a patient, a means for displaying at least a part of the representation, a means for user-designation of sub-regions of the representation, and a computing means adapted to segment the representation into a region within an anatomical structure and a region outside the anatomical structure on the basis of the locations of the sub-regions and the degree of similarity of voxels to neighboring voxels in the representation. The treatment planning apparatus preferably further comprises means for designating sub-regions as being inside the structure, means for designating sub-regions as being outside the structure, and/or means for designating sub-regions as being at a border of the structure. In the latter case, we prefer that the size of border sub-regions is adjustable. The computing means can segment the representation by minimizing a cost function. We also prefer that the initial values of the function in sub-regions designated as being in a border region are constrained to a different preset value. In an particularly useful aspect of this invention, after minimization of the function the values in sub-regions designated as being in a border region are then un-constrained, the remaining values are constrained, and the function is then minimized a second time.
1. A treatment planning apparatus comprising:
a means for receiving a three dimensional representation of a patient,
a means for displaying at least a part of the representation,
a means for user-designation of sub-regions of the representation, and
a computing means adapted to segment the representation into a region within an anatomical structure and a region outside the anatomical structure on the basis of the locations of the sub-regions and the degree of similarity of voxels to neighbouring voxels in the representation;
wherein the means for user-designation of sub-regions of the representation includes means for user-designation of sub-regions by means of a border stroke specifying the position of an edge of the anatomical structure at a few places, the computing means interpreting the border stroke and determining the edge position.
2. A treatment planning apparatus according to claim 1 in which the at least part of the representation comprises one or more sections through the representation.
3. A treatment planning apparatus according to claim 2 in which the sections are selected from the group consisting of axial, sagittal and coronal views of the patient.
4. A treatment planning apparatus according to claim 1 further comprising means for designating sub-regions as being inside the structure.
5. A treatment planning apparatus according to claim 1 further comprising means for designating sub-regions as being outside the structure.
6. A treatment planning apparatus according to claim 1 in further comprising a means for adjusting the size of a border sub-region.
7. A treatment planning apparatus according to claim 1 in which the computing means segments the representation by minimising a function having the form
C
(
I
)
=
∑
x
f
(
I
(
x
)
-
∑
y
∈
N
(
x
)
ω
xy
I
(
y
)
)
where ƒ is a function, I(x) is an inside/outside value of voxel x, N(x) is a neighbourhood of voxels around x, and ω xy is a measure of the similarity of voxels x and y.
8. A treatment planning apparatus according to claim 7 in which ω xy sums to one for a given x.
9. A treatment planning apparatus according to claim 7 in which the function ƒ(z) is of the form ƒ(z)=z n .
10. A treatment planning apparatus according to claim 9 in which n is 2.
11. A treatment planning apparatus according to claim 7 in which I(x) is 0 for voxels outside the segmented volume.
12. A treatment planning apparatus according to claim 7 in which I(x) is 1 for voxels within the segmented volume.
13. A treatment planning apparatus according to claim 7 in which the initial values of I(x) in the sub-regions prior to minimisation of the function is constrained to a preset value.
14. A treatment planning apparatus according to claim 13 in which the preset value is different for sub-regions inside the structure and sub-regions outside the structure.
15. A treatment planning apparatus according to claim 13 in which the initial values of I(x) in sub-regions designated as being in a border region are constrained to a preset value prior to minimisation of the function.
16. A treatment planning apparatus according to claim 15 in which after minimisation of the function, the values of I(x) in sub-regions designated as being in a border region are un-constrained, the remaining values of I(x) are constrained, and the function is minimised a second time.
17. A treatment planning apparatus according to claim 16 in which the remaining values of I(x) are constrained to their values after the first minimisation.
18. A treatment planning apparatus according to claim 16 in which the remaining values of I(x) are constrained to a preset value based on their value after the first minimisation.
19. A treatment planning apparatus according to claim 18 in which the remaining values of I(x) are constrained to a preset value corresponding to that of an inside voxel or an outside voxel.
20. A treatment planning apparatus according to claim 19 in which the remaining values of I(x) having a value after the first minimisation above a preset threshold are constrained to a preset value corresponding to that of an inside voxel, and the remaining values of I(x) having a value after the first minimisation below a preset threshold are constrained to a preset value corresponding to that of an outside voxel.
21. A method of processing a three dimensional representation of a patient, comprising the steps of:
displaying at least a part of the representation,
designating sub-regions of the representation in which the sub-regions are designated as being one of inside the structure, outside the structure, and containing an edge of the structure, and
segmenting the representation into a region within an anatomical structure and a region outside the anatomical structure,
wherein the segmentation is performed on the basis of the locations of the sub-regions and the degree of similarity of voxels to neighbouring voxels in the representation, and
wherein the designation of sub-regions containing an edge comprises drawing a border stroke specifying the edge of the anatomical structure at a few places, and the segmentation comprises interpreting the border stroke to determine the edge position.
22. A method according to claim 21 in which the segmentation is performed by minimising a function having the form
C
(
I
)
=
∑
x
f
(
I
(
x
)
-
∑
y
∈
N
(
x
)
ω
xy
I
(
y
)
)
where I(x) is an inside/outside value of voxel x, N(x) is a neighbourhood of voxels around x, and ω xy is a measure of the similarity of voxels x and y.
23. A treatment planning apparatus according to claim 22 in which after minimisation of the function, the values of I(x) in any sub-regions designated as being in a border region are un-constrained, the remaining values of I(x) are constrained, and the function is minimised a second time.