Image registration using a fully convolutional network

1. A method, comprising:

inputting a first image and a second image into a fully convolutional network;

determining an updated fully convolutional network by applying a displacement field for registering the first image and the second image on one or more of a voxel-by-voxel or a pixel-by-pixel basis, and optimizing a similarity metric associated with spatially transforming the first image to match the second image, wherein the similarity metric is a measure of similarity between at least the spatially transformed first image and the second image, and wherein one or more values of the fully convolutional network are adjusted to optimize the similarity metric; and

registering one or more of the first image or the second image based on the updated fully convolutional network.

2. The method of claim 1 , wherein the first image is based on a first imaging modality and the second image is based on a second imaging modality different than the first imaging modality.

3. The method of claim 1 , wherein the first image comprises a magnetic resonance imaging (MRI) scan image and the second image comprises a computed tomography (CT) scan image.

4. The method of claim 1 , wherein the first image represents a first object of interest and the second image represents a second object of interest.

5. The method of claim 4 , wherein the first object of interest is associated with a first patient and the second object of interest is associated with a second patient.

6. The method of claim 1 , wherein the first image represents a first object of interest associated with a first time and the second image represents the first object of interest associated with a second time different than the first time.

7. The method of claim 1 , wherein registering one or more of first image or the second image based on the updated fully convolutional network comprises spatially transforming the first image to at least partially match the second image.

8. The method of claim 1 , wherein the one or more values of the fully convolutional network comprise one or more of a kernel size or a stride of one or more layers of the fully convolutional network.

9. The method of claim 1 , wherein the fully convolutional network comprises a plurality of layers applying operations to one or more of the first image or the second image, and wherein the plurality of layers comprises one or more of a convolution layer, a deconvolution layer, a pooling layer, a normalization layer, an activation layer, or a multi-output regression layer.

10. The method of claim 1 , wherein the fully convolutional network is not trained with training data before inputting the first image and the second image into the fully convolutional network.

11. The method of claim 1 , further comprising:

determining a further updated fully convolutional network by further optimizing the similarity metric based on matching a third image to one or more of the first image, the second image, or a fourth image.

12. The method of claim 1 , wherein determining the updated fully convolutional network by optimizing the similarity metric associated with spatially transforming the first image to match the second image comprises using a self-supervision process in which a processor uses an algorithm to adjust the one or more values of the fully convolutional network until a value of the similarity metric reaches a threshold value.

13. A system, comprising:

a scanning device configured to generate a first image of an object of interest; and

a computing device configured to:

receive the first image;

input the first image and a second image into a fully convolutional network;

determine an updated fully convolutional network by applying a displacement field for registering the first image and the second image on one or more of a voxel-by-voxel or a pixel-by-pixel basis, and optimizing a similarity metric associated with spatially transforming the first image to match the second image, wherein the similarity metric is a measure of similarity between at least the spatially transformed first image and the second image, and wherein one or more values of the fully convolutional network are adjusted to optimize the similarity metric; and

register one or more of the first image or the second image based on the updated fully convolutional network.

14. The system of claim 13 , wherein the first image is based on a first imaging modality and the second image is based on a second imaging modality different than the first imaging modality.

15. The system of claim 13 , wherein the fully convolutional network is not trained with training data before inputting the first image and the second image into the fully convolutional network.

16. The system of claim 13 , wherein the computing device is configured to determine the updated fully convolutional network by optimizing the similarity metric associated with spatially transforming the first image to match the second image comprises using a self-supervision process in which a processor uses an algorithm to adjust the one or more values of the fully convolutional network until a value of the similarity metric reaches a threshold value.

17. A device, comprising:

one or more processors; and

memory storing instructions that, when executed by the one or more processors, cause the device to:

input a first image and a second image into a fully convolutional network;

determine an updated fully convolutional network by applying a displacement field for registering the first image and the second image on one or more of a voxel-by-voxel or a pixel-by-pixel basis, and optimizing a similarity metric associated with spatially transforming the first image to match the second image, wherein the similarity metric is a measure of similarity between at least the spatially transformed first image and the second image, and wherein one or more values of the fully convolutional network are adjusted to optimize the similarity metric; and

register one or more of the first image or the second image based on the updated fully convolutional network.

18. The device of claim 17 , wherein the fully convolutional network is not trained with training data before inputting the first image and the second image into the fully convolutional network.

19. The device of claim 17 , wherein the instructions cause the device to determine the updated fully convolutional network by optimizing the similarity metric associated with spatially transforming the first image to match the second image comprises using a self-supervision process in which a processor uses an algorithm to adjust the one or more values of the fully convolutional network until a value of the similarity metric reaches a threshold value.