PET quantification assumes that detected coincidence events can be converted into activity concentration within a given voxel. For this to be valid, several conditions must be satisfied.

First, the PET scanner must be correctly calibrated against a dose calibrator. The system must accurately relate detected count rate to absolute activity concentration. Errors in cross-calibration introduce systematic bias.

Second, photon attenuation must be corrected. If attenuation correction is inaccurate or misregistered, deeper structures will appear falsely reduced in activity.

Third, scatter correction is required. Scattered photons that are incorrectly assigned to lines of response artificially increase background counts and reduce contrast, leading to over- or underestimation of activity.

Fourth, partial volume effects reduce apparent activity in small structures due to spatial resolution limitations. This causes underestimation of true uptake in small lesions.

Reconstruction parameters also influence quantitative accuracy. Excessive smoothing reduces peak values, while insufficient modelling of detector response may distort activity distribution.

Count statistics play an additional role. Low-count acquisitions increase noise and variability in quantitative measurements.

Quantitative accuracy in PET is not determined by a single factor but by the cumulative accuracy of system calibration, physical corrections, and reconstruction modelling.