Cortical bone segmentation from MR Dixon data

Patent metadata

Field	Value
Publication number	US-9659370-B2
Application number	US-201414893980-A
Country	US
Kind code	B2
Filing date	Jun 19, 2014
Priority date	Jun 20, 2013
Publication date	May 23, 2017
Grant date	May 23, 2017

Abstract

The present invention relates to a method for segmenting MR Dixon image data. A processor and a computer program product are also disclosed for use in connection with the method. The invention finds application in the MR imaging field in general and more specifically may be used in the generation of an attenuation map to correct for attenuation by cortical bone during the reconstruction of PET images. In the method, a surface mesh is adapted to a region of interest by: for each mesh element in the surface mesh: selecting a water target position based on a water image feature response in the MR Dixon water image; selecting a fat target position based on a fat image feature response in the MR Dixon fat image; and displacing each mesh element from its current position to a new position based on both its water target position and its corresponding fat target position.

First claim

The invention claimed is:

1. A method for segmenting MR Dixon image data; the method comprising the steps of:
receiving an MR Dixon water image relating to a region of interest;
receiving an MR Dixon fat image relating to a region of interest;
adapting a surface mesh model having a plurality of mesh elements to the region of interest by:
for each mesh element in the region of interest:
selecting, from the current mesh element position and from a plurality of positions displaced from the current mesh element position, a water target position based on a water image feature response in the MR Dixon water image;
selecting, from the current mesh element position and from a plurality of positions displaced from the current mesh element position, a fat target position based on a fat image feature response in the MR Dixon fat image; and

displacing each mesh element from its current position to a new position based on both its water target position and its corresponding fat target position.

2. A method according to claim 1 wherein:
the water image feature response is based on at least one of: an intensity in the water image, a gradient of the intensity in the water image; and
the fat image feature response is based on at least one of: an intensity in the fat image, a gradient of the intensity in the fat image.

3. A method according to claim 1 wherein:
the water image feature response is based on at least one of: the difference between the intensity in the water image and the intensity in a reference water image, the difference between the gradient of the intensity in the water image and the gradient of the intensity in a reference water image; and
the fat image feature response is based on at least one of: the difference between the intensity in the fat image and the intensity in a reference fat image, the difference between the gradient of the intensity in the fat image and the gradient of the intensity in a reference fat image.

4. A method according to claim 1 wherein the new position minimises the energy for all water target positions and all fat target positions.

5. A method according to claim 1 wherein the new position is based on an energy minimum computed for all water target positions and all fat target positions.

6. A method according to claim 1 wherein the new position is based on an energy minimum computed for all water target positions and all fat target positions; wherein the energy is computed for each water target position and for each fat target position using a least squares method.

7. A method according to claim 1 wherein prior to the step of adapting the surface mesh model:
the surface mesh model is coarsely aligned to the region of interest by:
mapping an anchoring feature in the surface mesh to a corresponding anchoring feature in the region of interest;
wherein the anchoring feature comprises at least a portion of a pelvis, a femur or a lung.

8. A method according to claim 7 wherein the coarse alignment includes the method steps of:
executing a Hough Transform to align each mesh element in the region of interest with an image gradient; and
iteratively displacing the anchoring feature in order to improve alignment between the surface mesh model and the region of interest.

9. A method for obtaining a PET attenuation map; the method comprising the method steps of claim 1 and further including the method steps of:
assigning at least one gamma photon attenuation coefficient to at least one volume bounded by the surface mesh model.

10. A method according to claim 9 wherein a first gamma photon attenuation coefficient corresponding to the attenuation of cortical bone is assigned to at least one cortical bone volume bounded by the surface mesh model; and
a second gamma photon attenuation coefficient corresponding to the attenuation of bone marrow is assigned to at least one bone marrow volume within the cortical bone volume;
wherein the bone marrow volume is distinguished from the cortical bone volume based on the intensity in the fat image in relation to a reference intensity value.

11. A method according to claim 9 wherein a first gamma photon attenuation coefficient corresponding to the attenuation of a first anatomical bone type is assigned to a first volume bounded by the surface mesh model; and
a second gamma photon attenuation coefficient corresponding to the attenuation of a second anatomical bone type is assigned to a second volume bounded by the surface mesh model;
wherein the first and second anatomical bone types are different and wherein the first and second gamma photon attenuation coefficients are different.

12. A method according to claim 9 wherein a spatially-varying gamma photon attenuation coefficient is assigned to a volume bounded by the surface mesh model; wherein the value of the gamma photon attenuation coefficient is determined in accordance with a template distribution for the volume.

13. A method according to claim 1 wherein the segmentation is executed without the acquisition of either an ultrashort echo time (UTE) MR sequence or a T1- or a T2-weighted MR image.

14. A processor configured to execute the method steps of claim 1 .

15. A computer program product on a non-transitory computer readable medium comprising instructions which when executed on a processor cause the processor to carry out the method steps of claim 1 .

Assignees

• Koninklijke Philips Nv

Inventors

• Buerger Christian
• Waechter-Stehle Irina
• Peters Jochen
• Hansis Eberhard Sebastian
• Weber Frank Michael
• Klinder Tobias
• Renisch Steffen

Classifications

• G01R33/481 Primary

  MR combined with positron emission tomography [PET] or single photon emission computed tomography [SPECT] · CPC title

• G06T7/0014 Primary

  using an image reference approach · CPC title

• G06T12/00

  Tomographic reconstruction from projections · CPC title

• G06F18/22

  Matching criteria, e.g. proximity measures · CPC title

• G06F18/24

  Classification techniques · CPC title