Hybrid hardware and computer vision-based tracking system and method
The present disclosure relates to a tracking system for augmented reality in a clinical setting. Specifically, the present disclosure relates to an approach for combining hardware-based tracking and computer vision-based tracking in order to accurately overlay a projected image onto a video image.
1. A tracking system for augmented reality, the tracking system comprising:
an imaging modality configured to obtain a medical image;
a display configured to display the medical image; and
an image control device including processing circuitry configured to
acquire spatial data of a real object, the real object being a medical surgical tool observed in the medical image of the imaging modality, via a first tracking method,
determine a projection matrix based upon the acquired spatial data of the real object, the acquired spatial data corresponding to points observed on the real object,
project a virtual object onto the medical image of the imaging modality, the virtual object being a projection of a shape model of the real object according to the projection matrix,
calculate a correction matrix based on the projection of the virtual object,
update the projection of the virtual object according to the correction matrix, and
modify the display of the medical image based on the updated projection, the medical image including an ultrasound image overlayed on a camera image.
2. The tracking system according to claim 1 , wherein the projection of the virtual object is fused with a time-matched medical image of the imaging modality.
3. The tracking system according to claim 1 , wherein the first tracking method is an electromagnetic-based tracking system.
4. The tracking system according to claim 1 , wherein calculation of the correction matrix includes optimization of at least one cost function.
5. The tracking system according to claim 4 , wherein the at least one cost function is based upon a comparison of a position of one or more features of the virtual object and a position of a corresponding one or more features of the real object.
6. The tracking system according to claim 5 , wherein the position of the one or more features of the virtual object and the position of the corresponding one or more features of the real object is determined according to a second tracking method.
7. The tracking system according to claim 1 , wherein calculation of the correction matrix includes determining a spatial relationship between one or more features of the virtual object and a corresponding one or more features of the real object, the one or more features of the virtual object and the corresponding one or more features of the real object including a user-defined pattern.
8. The tracking system according to claim 1 , wherein calculation of the correction matrix includes determining a spatial relationship between one or more features of the virtual object and a corresponding one or more features of the real object, wherein the one or more features of the virtual object and the corresponding one or more features of the real object are determined based upon a classifier, the classifier being trained to detect the real object.
9. A method for augmented reality, comprising:
acquiring, via processing circuitry, spatial data of a real object, the real object being a medical surgical tool observed in a medical image of an imaging modality via a first tracking method;
determining, via the processing circuitry, a projection matrix based upon the acquired spatial data of the real object, the acquired spatial data corresponding to points observed on the real object;
projecting, via the processing circuitry, a virtual object onto the medical image of the imaging modality, the virtual object being a projection of a shape model of the real object according to the projection matrix;
calculating, via the processing circuitry, a correction matrix based on the projection of the virtual object;
updating, via the processing circuitry, the projection of the virtual object according to the correction matrix; and
displaying, via the processing circuitry, the medical image, wherein the displayed medical image includes an ultrasound image overlayed on a camera image and wherein the displayed medical image is modified based upon the updated projection.
10. The method according to claim 9 , wherein calculating the correction matrix includes optimization of at least one cost function.
11. The method according to claim 10 , wherein the at least one cost function is based upon a comparison of a position of one or more features of the virtual object and a position of a corresponding one or more features of the real object.
12. The method according to claim 11 , wherein the position of the one or more features of the virtual object and the position of the corresponding one or more features of the real object is determined according to a second tracking method.
13. The method according to claim 9 , wherein calculating the correction matrix includes determining a spatial relationship between one or more features of the virtual object and a corresponding one or more features of the real object, the one or more features of the virtual object and the corresponding one or more features of the real object including a user-defined pattern.
14. The method according to claim 9 , wherein calculating the correction matrix includes determining a spatial relationship between one or more features of the virtual object and a corresponding one or more features of the real object, wherein the one or more features of the virtual object and the corresponding one or more features of the real object are determined based upon a classifier, the classifier being trained to detect the real object.
15. A non-transitory computer-readable storage medium storing computer-readable instructions that, when executed by a computer, cause the computer to perform a method of augmented reality, comprising:
acquiring spatial data of a real object, the real object being a medical surgical tool observed in a medical image of an imaging modality, via a first tracking method;
determining a projection matrix based upon the acquired spatial data of the real object the acquired spatial data corresponding to points observed on the real object;
projecting a virtual object onto the medical image of the imaging modality, the virtual object being a projection of a shape model of the real object according to the projection matrix;
calculating a correction matrix based on the projection of the virtual object;
updating the projection of the virtual object according to the correction matrix; and
displaying the medical image, wherein the displayed medical image is modified based upon the updated projection and wherein the medical image includes an ultrasound image overlayed on a camera image.
16. The non-transitory computer-readable storage medium according to claim 15 , wherein calculating the correction matrix includes optimization of at least one cost function.
17. The non-transitory computer-readable storage medium according to claim 16 , wherein the at least one cost function is based upon a comparison of a position of one or more features of the virtual object and a position of a corresponding one or more features of the real object.
18. The non-transitory computer-readable storage medium according to claim 17 , wherein the position of the one or more features of the virtual object and the position of the corresponding one or more features of the real object is determined according to a second tracking method.
19. The non-transitory computer-readable storage medium according to claim 15 , wherein calculating the correction matrix includes determining a spatial relationship between one or more features of the virtual object and a corresponding one or more features of the real object, the one or more features of the virtual object and the corresponding one or more features of the real object including a user-defined pattern.
20. The non-transitory computer-readable storage medium according to claim 15 , wherein calculating the correction matrix includes determining a spatial relationship between one or more features of the virtual object and a corresponding one or more features of the real object, wherein the one or more features of the virtual object and the corresponding one or more features of the real object are determined based upon a classifier, the classifier being trained to detect the real object.