Methods to induce non-rigid mesh transformation in-variance to CNN based medical imaging classifiers
Embodiments of the present systems and methods may provide techniques that are applicable to a variety of imaging modalities and that utilize prior knowledge of the dynamics of a physiological system to analytically generate augmented samples for machine learning. For example, in an embodiment, a method implemented in a computer comprising a processor, memory accessible by the processor, and computer program instructions stored in the memory and executable by the processor, the method may comprise generating a transform for an image of tissue based on deformation of the tissue under compression, obtaining an image of tissue using an imaging modality, and generating an output image by transforming the image of the tissue using the transform.
1. A method implemented in a computer system comprising a processor, memory accessible by the processor, and computer program instructions stored in the memory and executable by the processor, the method comprising:
generating, at the computer system, a transform for an image of tissue based on deformation of the tissue under compression, wherein when an imaging modality generates the image of the tissue without an associated elastogram, the transform is generated by:
generating an elastogram based on reference images with exemplary elastograms;
modeling the tissue as a rectangular grid of masses connected by springs based on the generated elastogram; and
generating a displacement over the rectangular grid with a constrained bottom, wherein the displacement minimizes a total potential elastic energy with respect to positions of all the masses, except top and bottom masses, which induce compressional pressure;
obtaining, at the computer system, an image of tissue using the imaging modality; and
generating, at the computer system, an output image by transforming the image of the tissue using the transform.
2. The method of claim 1 , wherein when the imaging modality generates the image of the tissue with the associated elastogram the transform is generated by:
modeling the tissue as the rectangular grid of masses connected by the springs based on the associated elastogram; and
generating the displacement over the rectangular grid with the constrained bottom, wherein the displacement minimizes the total potential elastic energy with respect to the positions of all the masses, except the top and bottom masses, which induce the compressional pressure.
3. The method of claim 1 , wherein the imaging modality comprises one of ultrasound or magnetic resonance imaging.
4. The method of claim 1 , wherein the output image is further generated based on information indicating a pressure applied to the tissue during acquisition of the image of the tissue.
5. The method of claim 1 , wherein the imaging modality comprises one of mammography or tomography.
6. A system comprising a processor, memory accessible by the processor, and computer program instructions stored in the memory and executable by the processor to perform:
generating, at the computer system, a transform for an image of tissue based on deformation of the tissue under compression, wherein when an imaging modality generates the image of the tissue without an associated elastogram, the transform is generated by:
generating an elastogram based on reference images with exemplary elastograms;
modeling the tissue as a rectangular grid of masses connected by springs based on the generated elastogram; and
generating a displacement over the rectangular grid with a constrained bottom, wherein the displacement minimizes a total potential elastic energy with respect to positions of all the masses, except top and bottom masses, which induce compressional pressure;
obtaining, at the computer system, an image of tissue using the imaging modality; and
generating, at the computer system, an output image by transforming the image of the tissue using the transform.
7. The system of claim 6 , wherein when the imaging modality generates the image of the tissue with the associated elastogram the transform is generated by:
modeling the tissue as the rectangular grid of masses connected by the springs based on the associated elastogram; and
generating the displacement over the rectangular grid with the constrained bottom, wherein the displacement minimizes the total potential elastic energy with respect to the positions of all the masses, except the top and bottom masses, which induce the compressional pressure.
8. The system of claim 6 , wherein the imaging modality comprises one of ultrasound or magnetic resonance imaging.
9. The system of claim 6 , wherein the output image is further generated based on information indicating a pressure applied to the tissue during acquisition of the image of the tissue.
10. The system of claim 6 , wherein the imaging modality comprises one of mammography or tomography.
11. A computer program product comprising a non-transitory computer readable storage having program instructions embodied therewith, the program instructions executable by a computer, to cause the computer to perform a method comprising:
generating, at the computer system, a transform for an image of tissue based on deformation of the tissue under compression, wherein when an imaging modality generates the image of the tissue without generating an associated elastogram the transform is generated by:
generating an elastogram based on reference images with exemplary elastograms;
modeling the tissue as a rectangular grid of masses connected by springs based on the generated elastogram; and
generating a displacement over the rectangular grid with a constrained bottom, wherein the displacement minimizes a total potential elastic energy with respect to positions of all the masses, except top and bottom masses, which induce compressional pressure;
obtaining, at the computer system, an image of tissue using the imaging modality; and
generating, at the computer system, an output image by transforming the image of the tissue using the transform.
12. The computer program product of claim 11 , wherein when the imaging modality generates the image of the tissue with the associated elastogram the transform is generated by:
modeling the tissue as the rectangular grid of masses connected by the springs based on the associated elastogram; and
generating the displacement over the rectangular grid with the constrained bottom, wherein the displacement minimizes the total potential elastic energy with respect to the positions of all the masses, except the top and bottom masses, which induce the compressional pressure.
13. The computer program product of claim 11 , wherein the imaging modality comprises one of ultrasound or magnetic resonance imaging.
14. The computer program product of claim 11 , wherein the output image is further generated based on information indicating a pressure applied to the tissue during acquisition of the image of the tissue.