Table 4.

Summary of $k_2^r$ estimates obtained from 100 sets of time-activity curves (TACs) simulated at three different noise levels. All values are percentage units and include the coefficient of variation (COV), percent bias (BIAS%), and percent root mean square error (RMSE%) of the $k_2^r$ estimates. Simulated data set included five TACs of different binding level (BP_ND = 0.25, 0.5, 1.0, 2.0, 4.0) that were generated using a 2-tissue compartmental model including a blood volume term. The BP_ND values were chosen to cover the range of values obtained in clinical studies (see table 1). The $k_2^r$ estimates obtained using SRTM and Zhou were subsequently held fixed for application of SRTM2 and Zhou2, respectively. Note that SRTMC provides a single $k_2^r$ estimate for all binding levels.

BPND	SRTM			Zhou			SRTMC
	COV	BIAS%	RMSE%	COV	BIAS%	RMSE%	COV	BIAS%	RMSE%
Low noise (α = 0.02)
0.25	19.0	4.53	20.4	16.7	202	210	3.67	10.2	11.1
0.5	11.8	1.16	12.0	10.4	142	145
1	9.1	−0.49	9.1	10.2	120	122
2	8.3	0.94	8.5	8.7	119	121
4	8.5	1.00	8.7	8.8	122	124
Medium noise (α = 0.03)
0.25	28.2	3.27	29.3	29.2	333	357	5.48	11.7	13.2
0.5	17.5	4.73	18.9	16.0	188	194
1	14.4	3.98	15.5	13.5	156	160
2	12.5	2.89	13.2	11.7	142	146
4	12.3	1.97	12.7	16.3	159	165
High noise (α = 0.04)
0.25	35.8	8.31	39.6	47.7	569	654	6.02	10.1	12.1
0.5	23.2	2.04	23.7	21.3	249	261
1	17.8	4.85	19.3	15.8	189	196
2	16.3	−0.56	16.2	17.3	173	180
4	16.0	−0.22	15.9	67.3	205	291