Systems and methods for improved PET imaging
A method is provided that includes acquiring initial PET imaging data. The method also includes acquiring CT imaging data. Further, the method includes training a deep learning model for PET image reconstruction using the initial PET imaging data and the CT imaging data.
1. A method including:
acquiring initial positron emission tomography (PET) imaging data;
acquiring computed tomography (CT) imaging data; and
training a deep learning model for PET image reconstruction using the initial PET imaging data and the CT imaging data, wherein training the deep learning model includes a first stage and a second stage, with the first stage using the initial PET imaging data and the CT imaging data as inputs and providing modified PET imaging data as an output, and with the second stage using the initial PET imaging data and the modified PET imaging data as inputs and providing further modified PET imaging data as an output.
2. The method of claim 1 , wherein training the deep learning model includes utilizing at least one convolutional block, with the at least one convolutional block using the initial PET imaging data as an input.
3. The method of claim 2 , wherein the at least one convolutional block includes plural convolutional layers.
4. The method of claim 1 , further comprising converting the CT imaging data from a CT format to a PET format before using the CT imaging data to train the deep learning model.
5. The method of claim 4 , wherein acquiring the CT imaging data comprises acquiring the CT imaging data using X-ray photons, and wherein converting the CT imaging data to the PET format comprises converting the CT imaging data to PET equivalent imaging data that represents how the CT imaging data would appear if the CT imaging data were acquiring using PET gamma photons.
6. A system including:
a PET acquisition unit configured to acquire initial PET imaging data;
a CT acquisition unit configured to acquire CT imaging data; and
a processing unit configured to acquire the initial PET imaging data and the CT imaging data from the PET acquisition unit and CT acquisition unit, respectively, and to reconstruct an image using a deep learning model, with the initial PET imaging data and the CT imaging data used as inputs to the deep learning model, wherein the deep learning model uses a first stage and a second stage to reconstruct the image, with the first stage using the initial PET imaging data and the CT imaging data as inputs and providing modified PET imaging data as an output, and with the second stage using the initial PET imaging data and the modified PET imaging data as inputs and providing further modified PET imaging data as an output.
7. The system of claim 6 , wherein the deep learning model utilizes at least one convolutional block, with the at least one convolutional block using the initial PET imaging data as an input.
8. The system of claim 7 , wherein the at least one convolutional block includes plural convolutional layers.
9. The system of claim 8 , wherein at least two convolutional layers have a common number of filters, and at least one convolutional layer has a different number of filters that is different from the common number.
10. The system of claim 6 , wherein the processing unit is configured to convert the CT imaging data from a CT format to a PET format before using the CT imaging data as an input to the deep learning model.
11. The system of claim 10 , wherein the CT acquisition unit is configured to acquire the CT imaging data using X-ray photons, and wherein the processing unit is configured to convert the CT imaging data to PET equivalent CT imaging data that represents how the CT imaging data would appear if the CT imaging data were acquiring using PET gamma photons.
12. A method including:
acquiring initial PET imaging data with a PET acquisition unit;
acquiring CT imaging data with a CT acquisition unit; and
reconstructing an image using a deep learning model, wherein the initial PET imaging data and the CT imaging data are used as inputs to the deep learning model, wherein the deep learning model uses a first stage and a second stage to reconstruct the image, with the first stage using the initial PET imaging data and the CT imaging data as inputs and providing modified PET imaging data as an output, and with the second stage using the initial PET imaging data and the modified PET imaging data as inputs and providing further modified PET imaging data as an output.
13. The method of claim 12 , wherein the deep learning model utilizes at least one convolutional block, with the at least one convolutional block using the initial PET imaging data as an input.
14. The method of claim 13 , wherein the at least one convolutional block includes plural convolutional layers.
15. The method of claim 14 , wherein at least two convolutional layers have a common number of filters, and at least one convolutional layer has a different number of filters that is different from the common number.
16. The method of claim 12 , further comprising converting the CT imaging data from a CT format to a PET format before using the CT imaging data as an input to the deep learning model.
17. The method of claim 12 , further comprising acquiring the CT imaging data using X-ray photons, and converting the CT imaging data to PET equivalent CT imaging data that represents how the CT imaging data would appear if the CT imaging data were acquiring using PET gamma photons.