Skip to main content

Table 2 The cortical thickness data underwent cross-validation

From: Cortical thickness analysis in temporal lobe epilepsy using fully Bayesian spectral method in magnetic resonance imaging

Residuals

FE

Nonstationary

J

MSE

Independent

Gaussian

No

4

4.85E−02

Independent

Gaussian

No

9

4.78E−02

Independent

Gaussian

Yes

4

4.15E−02

Independent

Gaussian

Yes

9

4.05E−02

Independent

Horseshoe

No

4

4.79E−02

Independent

Horseshoe

No

9

4.77E−02

Independent

Horseshoe

Yes

4

4.11E−02

Independent

Horseshoe

Yes

9

4.09E−02

Matérn

Gaussian

No

4

4.74E−02

Matérn

Gaussian

No

9

4.78E−02

Matérn

Gaussian

Yes

4

4.59E−02

Matérn

Gaussian

Yes

9

4.01E−02

Matérn

Horseshoe

No

4

4.83E−02

Matérn

Horseshoe

No

9

4.90E−02

Matérn

Horseshoe

Yes

4

4.16E−02

Matérn

Horseshoe

Yes

9

4.22E−02

  1. The residual correlation of \(\tilde{E}\) affects the strategies used to fit the data; the prior for the fixed effects (“FE”) \(\tilde{B}\), the inclusion of the nonstationary subject random effects (“nonstationary”) \(\tilde{Z}\), and the number of basis functions J. Methods are compared by the mean squared prediction error (MSE).