TABLE 1.

Summary of loss functions and hyper‐parameters

Functions	Purpose	Hyper‐parameters	Value
losscycle	Cycle consistency loss, GCBCT→CT = GCT→CBCT ’	λcycle	10.0
lossadv	Adversarial loss, synthetizes realistic images	λadv	1.0
lossgrad	Gradient loss, preserves edges and structures	λgrad	1.0
lossidem	Idempotent loss, idempotency preservation	λidem	1.0
losstv	Total variation regularization loss, produce spatially uniform images	λtv	1.0