Estimation of partially missing attenuation in time-of-flight positron emission tomography

A method for reconstructing in positron emission tomography with partially known attenuation, the method comprising:

• acquiring, with a computed tomography (CT) scanner, the partially known attenuation for a CT volume of a patient;
• estimating attenuation outside of the CT volume with a machine-learned estimator;
• generating a positron emission tomography image based on the attenuation outside the CT volume and the partially known attenuation for the CT volume;
• and
• displaying the positron emission tomography image.

A system for reconstructing in positron emission tomography (PET) with partially known attenuation, the system comprising:

• an x-ray scanner configured to obtain attenuation data for a patient volume;
• a bed configured to move to scan different parts of a patient in a positron emission tomography (PET) scan;
• rings of detectors spaced axially operable to perform the PET scan along lines of response between the detectors, a first set of lines of response oblique to the axial spacing of the detectors extending out of the patient volume;
• and
• a processor connected to the detectors, the processor configured to reconstruct activity distribution using time-of-flight for detected emissions along the lines of response including the first set and others, the activity distribution including corrections for the first set of lines of response where extending out of the patient volume, the corrections being based on output of a neural network;
• and
• a display configured to display a PET image of the activity distribution.