| | 1.5 Tesla | 3 Tesla |
|---|
Algorithm | Fermi | ARMA | Exponential | B-spline | Fermi | ARMA | Exponential | B-spline |
Fermi | p = 1 | p = 0.06 | p = 0.64 |
p = 0.001
| p = 1 |
p < 0.001
|
p < 0.001
|
p = 0.001
|
ARMA | p = 0.06 | p = 1 | p = 0.16 | p = 0.04 |
p < 0.001
| p = 1 | p = 0.36 | p = 0.6 |
Exponential | p = 0.64 | 0.16 | p = 1 |
p = 0.001
|
p < 0.001
| p = 0.36 | p = 1 | p = 0.15 |
B-spline |
p = 0.001
| 0.04 |
p = 0.001
| p = 1 |
p = 0.001
| p = 0.6 | p = 0.15 | p = 1 |
- The table shows the results of the comparison of the correlation coefficients of the individual algorithms. Fermi function deconvolution had the strongest r-value at both field strengths. In comparison with the other methods it performed significantly better at 3 Tesla. At 1.5 Tesla it performed significantly better than B-spline deconvolution but not than ARMA and exponential deconvolution. The comparison of the correlation strength of the individual algorithms with the microsphere reference-standard was based on Fisher z transformation. P values of less than 0.008 remained significant after Bonferroni correction for multiple comparisons.
