Table 1. User-specified parameters.

Manuscript Subsection	Parameter	Description	Value Used
Hybrid data fidelity
	η	weighted least-squares (EID & PCD)	3
	gi	for all i (EID & PCD)	1
Penalized algebraic reconstruction
	λC,e	hybrid data fidelity; for all e	1
	α	regularization scaling factor	0.01
Joint bilateral filtration
	b	BF domain radius	6
RSKR
	h0	noise multiplier	1.5
	γ	regularization scaling exponent	0.5
	s	voxel stride used for tiling	3
Dictionary learning
	b	3D patch radius	4
	s	voxel stride for noise measurement	3
	na	dictionary atom count	1024
K-SVD algorithm
	n0	max non-zero coefficients	48
		number of patches used	500,000
		K-SVD iterations	25
regularization with OMP
	n0	max non-zero coefficients	5
Computational considerations
initialization (Fig 1, step 5)		BiCGSTAB iterations, per energy	30
each Bregman iteration (Fig 1, step 18)		BiCGSTAB iterations, per energy	25
(Fig 1, steps 13–18)		total number of Bregman iterations	6