After annihilation, two 511 keV photons travel in opposite directions toward the detector ring. As they pass through tissue, some photons are absorbed or scattered. The probability of attenuation depends on tissue density and path length.

In PET, both photons in a coincidence pair must reach the detectors to be registered as a true event. If one photon is attenuated, the event is lost entirely. This produces position-dependent count loss, particularly in deeper or denser regions of the body.

To correct for this, an attenuation map is required. In modern systems, a CT scan is acquired immediately before or after the PET acquisition. The CT image provides tissue density information, which is converted into linear attenuation coefficients at 511 keV.

During reconstruction, each line of response is weighted according to the estimated attenuation along its path. This compensates for photon losses and restores more accurate activity distribution.

Attenuation correction improves uniformity and quantitative accuracy but requires precise spatial alignment between PET and CT data. Misregistration due to breathing or patient motion can introduce artefacts.