Systems and methods for signal processing in molecular imaging

A method for signal processing, comprising:

• obtaining, by at least one processor, a first signal that is acquired by sampling, according to a first sampling frequency, an electrical signal of a detector of an emission computed tomography (ECT) device or a gamma camera, wherein the electrical signal is an analog signal and the first signal is a digital signal converted from the electrical signal, the electrical signal is generated by detector units of the detector by detecting radiation rays emitted from annihilation events in a subject after the subject is injected with a radiopharmaceutical;
• generating, based on the first signal and a target machine learning model, a second signal by the at least one processor, the second signal being a digital signal corresponding to a second sampling frequency that is different from the first sampling frequency, the target machine learning model specifying a target mapping between the first signal and the second signal;
• obtaining, by the at least one processor, photon information of the radiation rays based on the second signal;
• and
• reconstructing, by the at least one processor, an image based on the photon information of the radiation rays.

A detector of an emission computed tomography (ECT) device or a gamma camera, comprising:

• one or more detection units configured to generate an electrical signal by detecting radiation rays emitted from annihilation events in a subject after the subject is injected with a radiopharmaceutical;
• and
• one or more processing circuits configured to:
• generate a first signal by sampling, according to a first sampling frequency, the electrical signal, wherein the electrical signal is an analog signal and the first signal is a digital signal converted from the electrical signal;
• and
• generate a second signal based on the first signal and a target machine learning model, the second signal being a digital signal corresponding to a second sampling frequency that is different from the first sampling frequency, the target machine learning model specifying a target mapping between the first signal and the second signal.