Table 2.
Proportion of simulations that had the number of components underestimated, overestimated or correctly estimated by NLSA and SA
| Simulated scenario |
NLSA performance |
SA performance |
||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
|
SNR = 60 |
SNR = 120 |
SNR = 60 |
SNR = 120 |
|||||||||
| UE (%) | OE (%) | CE (%) | UE (%) | OE (%) | CE (%) | UE (%) | OE (%) | CE (%) | UE (%) | OE (%) | CE (%) | |
| Tri-exponential |
|
|
|
|
|
|
|
|
|
|
|
|
| No trapping (α0 = 0) |
85 |
- |
15 |
0 |
- |
100 |
- |
81a |
19a |
- |
100a |
0a |
| Trapping 1 (α0 = 0.006) |
100 |
- |
0 |
46 |
- |
54 |
18 |
- |
82 |
- |
- |
100 |
| Trapping 2 (α0 = 0.06) |
100 |
- |
0 |
60 |
- |
40 |
18 |
- |
82 |
- |
- |
100 |
| Trapping 3 (α0 = 0.2) |
100 |
- |
0 |
95 |
- |
5 |
46 |
- |
54 |
- |
- |
100 |
| Trapping 4 (α0 = 0.6) |
100 |
- |
0 |
100 |
- |
0 |
53 |
- |
47 |
1 |
- |
99 |
| Bi-exponential |
|
|
|
|
|
|
|
|
|
|
|
|
| No trapping (α0 = 0) |
- |
- |
100 |
- |
- |
100 |
- |
86a |
14a |
- |
- |
100a |
| Trapping 1 (α0 = 0.006) |
1 |
- |
99 |
- |
- |
100 |
17 |
18 |
65 |
- |
- |
100 |
| Trapping 2 (α0 = 0.06) |
1 |
- |
99 |
- |
- |
100 |
- |
- |
100 |
- |
- |
100 |
| Trapping 3 (α0 = 0.2) |
11 |
- |
89 |
5 |
- |
95 |
- |
- |
100 |
- |
- |
100 |
| Trapping 4 (α0 = 0.6) | 34 | - | 66 | 32 | - | 68 | - | - | 100 | - | - | 100 |
α0 indicates the tracer trapping. Units are in min-1; UE, underestimated number of components; OE, overestimated number of components; CE, correctly estimated. Results are shown only for the percentage of datasets where the presence of baseline was correctly estimated (Table 1). aAn accuracy in baseline identification was <25%.