Systems and methods for super time-of-flight resolution
AssigneeSiemens Medical Solutions USA, Inc.
InventorsJorge Cabello, Maurizio Conti
Life of this patent
Estimating the term…
Abstract
Systems and methods of generating improved resolution histo-images are disclosed. A system includes a positron emission tomography (PET) imaging modality configured to execute a first scan to acquire a first PET dataset and a processor configured to back-project the first PET dataset to generate a first histo-image having a first resolution, input the first histo-image to a trained neural network, receive a second histo-image from the trained neural network, and input the second histo-image to a reconstruction process configured to generate a reconstructed PET image. The second histo-image has a second resolution higher than the first resolution. The second histo-image represents the first PET dataset.
Claims20
A system, comprising:
- a positron emission tomography (PET) imaging modality configured to execute a first scan to acquire a first PET dataset at a first response time resolution;
- and
- a processor configured to:
- back-project the first PET dataset to generate a first histo-image that represents the first PET dataset and having the first response time resolution;
- input the first histo-image to a trained neural network;
- receive a second histo-image from the trained neural network, wherein the second histo-image has an effective second response time resolution, wherein the effective second response time resolution is higher than the first response time resolution and higher than otherwise possible for the PET imaging modality, and wherein the second histo-image represents a higher response time resolution version of the first PET dataset;
- and
- input the second histo-image to a reconstruction process configured to generate a reconstructed PET image.
A method of generating reconstructed positron emission tomography (PET) images, comprising:
- executing a first scan to acquire a first PET dataset;
- back-projecting the first PET dataset to generate a first histo-image that represents the first PET data and having a first response time resolution;
- inputting the first histo-image to a trained neural network;
- receiving a second histo-image from the trained neural network, wherein the second histo-image has an effective second response time resolution, wherein the effective second response time resolution is higher than the first response time resolution, and wherein the second histo-image represents a higher response time resolution version of the first PET dataset;
- and
- inputting the second histo-image to a reconstruction process configured to generate a reconstructed PET image.
A method of training a neural network to generate higher-resolution histo-images, comprising:
- receiving a training dataset comprising a first plurality of histo-images having a first response time resolution and a second plurality of histo-images having a second response time resolution, wherein each of the histo-images in the first plurality of histo-images has at least one corresponding histo-image in the second plurality of histo-images, and wherein the second response time resolution of each histo-image in the second plurality of histo-images is higher than the first response time resolution of the corresponding histo-image in the first plurality of histo-images;
- inputting each histo-image in the first plurality of histo-images to a neural network configured to generate a corresponding higher-resolution histo-image having an effective second response time resolution;
- comparing each of the higher-resolution histo-images to the at least one corresponding histo-image in the second plurality of histo-images to determine any differences between each higher-resolution histo-image and the at least one corresponding histo-image in the second plurality of histo-images;
- and
- modifying the neural network based on the determined differences between each higher-resolution histo-image and the at least one corresponding histo-image in the second plurality of histo-images.
Dependencies are inferred from the claim text.