Deformable registration of medical images
Systems and computer-implemented methods of performing image registration. One method includes receiving a first image and a second image acquired from a patient at different times and, in each of the first image and the second image, detecting an upper boundary of an imaged object in an image coordinate system and detecting a lower boundary of the imaged object in the image coordinate system. The method further includes, based on the upper boundary and the lower boundary of each of the first image and the second image, cropping and padding at least one of the first image and the second image to create an aligned first image and an aligned second image and executing a registration model on the aligned first image and the aligned second image to compute a deformation field between the aligned first image and the aligned second image.
1. A computer-implemented method of performing image registration, the method comprising: receiving a first image and a second image, the first image and the second image acquired from a patient at different times; in each of the first image and the second image: detecting an upper boundary of an imaged object in an image coordinate system, and detecting a lower boundary of the imaged object in the image coordinate system; based on the upper boundary and the lower boundary of each of the first image and the second image, cropping and padding at least one of the first image and the second image to create an aligned first image and an aligned second image; executing a registration model on the aligned first image and the aligned second image to compute a deformation field between the aligned first image and the aligned second image; wherein detecting the upper boundary of the imaged object in each of the first image and the second image includes detecting the upper boundary in each of the first image and the second image using a deep learning model; and wherein detecting the lower boundary of the imaged object in each of the first image and the second image includes detecting the lower boundary in each of the first image and the second image using a deep learning model.
2. The method of claim 1 , wherein receiving the first image and the second image includes receiving a first three-dimensional (3D) image of a liver of the patient and a second 3D image of the liver of the patient.
3. The method of claim 2 , wherein the first 3D image is a computed tomography (CT) image and the second 3D image is a CT image.
4. The method of claim 2 , wherein the first 3D image is a magnetic resonance (MR) image and the second 3D image is a MR image.
5. The method of claim 1 , wherein executing the registration model includes executing a deep learning model trained using a loss function, wherein a first term of the loss function includes a common portion of the imaged object within the aligned first image and the aligned second image.
6. The method of claim 5 , wherein a second term of the loss function includes on one or more lesions visible in both the aligned first image and the aligned second image.
7. A system for performing image registration, the system comprising: an electronic processor configured to: receive a first image and a second image, the first image and the second image acquired from a patient at different times, in each of the first image and the second image: detect an upper boundary of an imaged object in an image coordinate system, and detect a lower boundary of the imaged object in the image coordinate system, based on the upper boundary and the lower boundary of each of the first image and the second image, crop and pad at least one of the first image and the second image to create an aligned first image and an aligned second image, execute a registration model on the aligned first image and the aligned second image to compute a deformation field between the aligned first image and the aligned second image; wherein detecting the upper boundary of the imaged object in each of the first image and the second image includes detecting the upper boundary in each of the first image and the second image using a deep learning model; and
wherein detecting the lower boundary of the imaged object in each of the first image and the second image includes detecting the lower boundary in each of the first image and the second image using a deep learning model.
8. The system of claim 7 , wherein the first image is a first three-dimensional (3D) image of a liver of the patient and the second image is a second 3D image of the liver of the patient.
9. The system of claim 7 , wherein the electronic processor is configured to detect the upper boundary and the lower boundary of the imaged object in each of the first image and the second image via a deep learning model.
10. The system of claim 7 , wherein the electronic processor is configured to execute the registration model by executing a deep learning model trained using a loss function, a first term of the loss function including a common portion of the imaged object within the aligned first image and the aligned second image.
11. The system of claim 10 , wherein a second term of the loss function includes one or more lesions visible in both the aligned first image and the aligned second image.
12. Non-transitory computer readable medium storing instructions that, when executed by an electronic processor, perform a set of functions, the set of functions comprising: receiving a first image and a second image, the first image and the second image acquired from a patient at different times; in each of the first image and the second image: detecting an upper boundary of an imaged object in an image coordinate system, and detecting a lower boundary of the imaged object in the image coordinate system; based on the upper boundary and the lower boundary of each of the first image and the second image, cropping and padding at least one of the first image and the second image to create an aligned first image and an aligned second image; and executing a registration model on the aligned first image and the aligned second image to compute a deformation field between the aligned first image and the aligned second image;
wherein detecting the upper boundary and the lower boundary of the imaged object in each of the first image and the second image includes detecting the upper boundary in each of the first image and the second image using a deep learning model.
13. The non-transitory computer readable medium of claim 12 , wherein receiving the first image and the second image includes receiving a first three-dimensional (3D) image of a liver of the patient and a second 3D image of the liver of the patient.
14. The non-transitory computer readable medium of claim 13 , wherein the first 3D image is one selected from a group consisting of a computed tomography (CT) image and a magnetic resonance (MR) image and the second 3D image is one selected from a group consisting of a CT image and a MR image.
15. The non-transitory computer readable medium of claim 12 , wherein executing the registration model includes executing a deep learning model trained using a loss function, wherein a first term of the loss function includes a common portion of the imaged object within the aligned first image and the aligned second image.
16. The non-transitory computer readable medium of claim 15 , wherein the loss function includes one or more lesions visible in both the aligned first image and the aligned second image.