Detection of anatomical landmarks

Patent metadata

Field	Value
Publication number	US-9842394-B2
Application number	US-201715471421-A
Country	US
Kind code	B2
Filing date	Mar 28, 2017
Priority date	Aug 13, 2010
Publication date	Dec 12, 2017
Grant date	Dec 12, 2017

Abstract

A method including accessing image data representing tissue and identifying one or more features of the tissue indicated by the image data. A model is selected for the tissue based on the one or more identified features. The image data is segmented and, using the model, one or more anatomical landmarks of the tissue indicated by the segmented image data are identified.

First claim

What is claimed is:

1. A method, comprising:
accessing image data representing tissue;
identifying one or more features of the tissue indicated by the image data;
accessing model data including information relating to each of a plurality of reference models, wherein each of the reference models corresponds to a different one of a plurality of patient populations, and wherein each of the reference models includes a set of attributes related to a corresponding one of the patient populations;
selecting a model for the tissue, including:
comparing the identified one or more features of the tissue with the set of attributes for each of the reference models to determine which set of attributes most closely matches the identified one or more features; and
selecting the reference model including the set of attributes that most closely matches the identified one or more features;

segmenting the image data; and
identifying, using the selected reference model, one or more anatomical landmarks of the tissue indicated by the segmented image data.

2. The method of claim 1 , wherein the accessing the image data, the identifying the one or more features of the tissue, the accessing the model data, the selecting the model, the segmenting the image data, and identifying the one or more anatomical landmarks are performed by one or more computing devices.

3. The method of claim 1 , wherein the identifying the one or more features of the tissue indicated by the image data comprises performing shape recognition on the image data to identify the tissue and orientation of the tissue.

4. The method of claim 1 , wherein each of the patient populations corresponds to at least one of a medical condition, a medical deformity, size, age, and sex;
wherein each of the reference models is representative of tissue for the corresponding patient population; and
wherein selecting the reference model for the tissue further comprises determining that the one or more identified features are correlated with the medical condition, medical deformity, size, age, or sex that corresponds to the set of attributes that most closely matches the one or more identified features.

5. The method of claim 1 , wherein the segmenting the image data comprises defining a plurality of boundaries in the image data, and wherein defining the plurality of boundaries comprises applying one or more thresholds to the image data.

6. The method of claim 5 , wherein each of the reference models further comprises threshold information, and wherein applying one or more thresholds to the image data comprises applying the one or more thresholds based upon the threshold information of the selected model.

7. The method of claim 5 , wherein the defining the plurality of boundaries in the image data further comprises detecting edges of the image data.

8. The method of claim 1 , wherein each of the reference models further includes a template indicating locations of one or more anatomical features of the corresponding patient population, and wherein the identifying one or more anatomical landmarks of the tissue includes identifying the one or more anatomical landmarks based upon the template of the selected reference model.

9. The method of claim 8 , further comprising registering the image data with the template of the selected model prior to the identifying the one or more anatomical landmarks.

10. A method, comprising:
accessing image data representing tissue;
identifying a feature set of the tissue including one or more features of the tissue indicated by the image data;
accessing model data including a plurality of reference models, wherein each of the reference models corresponds to a different one of a plurality of patient populations, wherein each of the reference models includes a set of image processing parameters and a template for detecting one or more anatomical landmarks;
determining that the identified feature set is correlated with one of the patient populations based upon a comparison of the one or more identified features with the template of each of the reference models;
selecting the reference model corresponding to the patient population with which the feature set is correlated;
segmenting the image data based upon the set of image processing parameters of the selected reference model; and
identifying one or more anatomical landmarks of the tissue based upon the segmented image data and the template of the selected reference model.

11. The method of claim 10 , wherein the template associated with each reference model includes attribute information related to attributes characteristic of the patient population corresponding to the reference model, and wherein the comparison of the one or more identified features with the template of each reference model includes a comparison of the feature set with the attribute information of each template.

12. The method of claim 11 , wherein the template of each reference model further includes location information relating to one or more anatomical features of the corresponding patient population, and wherein the identifying the one or more anatomical landmarks based upon the segmented image data and the template of the selected reference model includes identifying the one or more anatomical features based upon the location information of the template.

13. The method of claim 10 , wherein identifying the one or more anatomical landmarks comprises:
identifying slices of the image data corresponding to a region of interest;
selecting one of the slices that has the highest contrast for the region of interest; and
identifying the one or more anatomical landmarks based on the segmented image data corresponding to the selected slice.

14. The method of claim 13 , wherein the image processing information for each of the reference models includes one or more intensity or contrast thresholds, and wherein the segmenting the image data includes applying the one or more intensity or contrast thresholds to the image data.

15. The method of claim 10 , wherein the set of image processing parameters for each reference model is tailored to the patient population corresponding to the reference model.

16. The method of claim 10 , further comprising:
identifying a region of the segmented image data containing an error in a segmented boundary corresponding to a predetermined anatomical feature; and
correcting the segmented boundary based on the selected reference model to produce corrected segmented image data; and
wherein the identifying one or more anatomical landmarks in the segmented image data using the selected reference model comprises identifying one or more anatomical landmarks in the corrected segmented image data using the selected reference model.

17. The method of claim 10 , wherein the identifying the feature set comprises:
segmenting the image data to produce initial segmented image data;
identifying the one or more features of the tissue based on the initial segmented image data; and
identifying the feature set as correlated with a particular medical condition or deformity.

18. A method, comprising:
accessing image data representing tissue;
accessing attribute information including a plurality of attribute sets, wherein each of the attribute sets is characteristic of a corresponding patient population;
accessing model information including a plurality of reference models, wherein each of the reference models corresponds to a respective one of the patient populations;
identifying, using the image data, one or more features of the tissue;
selecting

Assignees

• Smith & Nephew Inc

Inventors

• Buisseret Mark Joseph
• Nadzadi Mark Ellsworth
• Rumery Megan Patrice
• Agnihotri Aashiish
• Mckinnon Brian W

Classifications

• G06T7/0012 Primary

  Biomedical image inspection · CPC title

• G06T7/12 Primary

  Edge-based segmentation · CPC title

• G06T7/60

  Analysis of geometric attributes · CPC title

• G06T7/149

  involving deformable models, e.g. active contour models · CPC title

• G06T2207/10088

  Magnetic resonance imaging [MRI] · CPC title