Abstract: Systems and methods for reconstructing medical images based on the trained deep learning processes, and for training deep learning processes, are disclosed. In some examples, image measurement data is received. A histo-image is generated based on the image measurement data. Further, an attenuation map, such as a ?-map, is received. An attenuation histo-image is generated based on the attenuation map. Further, a trained machine learning process, such as a trained neural network, is applied to features generated from the histo-image and the attenuation histo-image. Based on the application of the machine learning process to the histo-image and the attenuation histo-image, output image data characterizing an image volume is generated. In some examples, a machine learning process is trained based on histo-images and corresponding attenuation histo-images. The trained machine learning process may be employed to reconstruct images, such as positron emission tomography (PET) images.

Abstract: Systems and methods for training end-to-end deep learning reconstruction processes, and for reconstructing medical images based on the trained deep learning processes, are disclosed. In some examples, input projection data is received. An untrained machine learning process is applied to the input projection data and, based on the application of the machine learning process to the projection data, an output image is generated. Further, a forward projection process is applied to the output image and, based on the application of the forward projection process to the output image, forward projected image data is generated. A loss value is then determined based on the forward projected image data and the input projection data. The loss value is then compared to a threshold value to determine whether the machine learning process is trained. The trained machine learning process may be employed to reconstruct images, such as positron emission tomography (PET) images.