Table 3.
Open in new tab

Multivariable analyses

Multivariable modelSignificant parameters for multivariable logistic regressionMultivariable logistic regression, P-valueAUC with cross-validationAIC
1. Step AIC modelAREXAmide,0.625µT0.013 (*)0.75178.95
AREXNOE,0.625µT0.008 (**)
1/(RA·T2A)0.004 (**)
T10.004 (**)
2. Step AIC model with PCAMTRAsym,0.625µT0.014 (*)0.71178.94
1/(RA·T2A)0.030 (*)
Multivariable analysis including perfusion rCBV with multiple imputation
3. Step AIC model1/(RA·T2A)0.012 (*)0.67
[0.67, 0.67]		N/A
T10.028 (*)
AREXAmide,0.625µT0.047 (*)
AREXNOE,0.625µT0.026 (*)
4. Step AIC model with PCA1/(RA·T2A)0.017 (*)0.67
[0.65, 0.69]		N/A
Multivariable modelSignificant parameters for multivariable logistic regressionMultivariable logistic regression, P-valueAUC with cross-validationAIC
1. Step AIC modelAREXAmide,0.625µT0.013 (*)0.75178.95
AREXNOE,0.625µT0.008 (**)
1/(RA·T2A)0.004 (**)
T10.004 (**)
2. Step AIC model with PCAMTRAsym,0.625µT0.014 (*)0.71178.94
1/(RA·T2A)0.030 (*)
Multivariable analysis including perfusion rCBV with multiple imputation
3. Step AIC model1/(RA·T2A)0.012 (*)0.67
[0.67, 0.67]		N/A
T10.028 (*)
AREXAmide,0.625µT0.047 (*)
AREXNOE,0.625µT0.026 (*)
4. Step AIC model with PCA1/(RA·T2A)0.017 (*)0.67
[0.65, 0.69]		N/A

Multivariable logistic regression was performed with and without perfusion rCBV values included. Principal component analysis (PCA) was used to reduce the AREX and MTR parameters in Model 2, whereas Model 1 used all the individual MT/CEST variables as input. Where rCBV was included (ie, Models 3 and 4), multiple imputation was used to deal with missing data. Significant parameters and values are bolded, and asterisks (*) and (**) indicate P-values below .05 and .01, respectively. For AUCs with multiple imputation and cross-validation, the 1st and 3rd quartile are shown in brackets.

Table 3.
Open in new tab

Multivariable analyses

Multivariable modelSignificant parameters for multivariable logistic regressionMultivariable logistic regression, P-valueAUC with cross-validationAIC
1. Step AIC modelAREXAmide,0.625µT0.013 (*)0.75178.95
AREXNOE,0.625µT0.008 (**)
1/(RA·T2A)0.004 (**)
T10.004 (**)
2. Step AIC model with PCAMTRAsym,0.625µT0.014 (*)0.71178.94
1/(RA·T2A)0.030 (*)
Multivariable analysis including perfusion rCBV with multiple imputation
3. Step AIC model1/(RA·T2A)0.012 (*)0.67
[0.67, 0.67]		N/A
T10.028 (*)
AREXAmide,0.625µT0.047 (*)
AREXNOE,0.625µT0.026 (*)
4. Step AIC model with PCA1/(RA·T2A)0.017 (*)0.67
[0.65, 0.69]		N/A
Multivariable modelSignificant parameters for multivariable logistic regressionMultivariable logistic regression, P-valueAUC with cross-validationAIC
1. Step AIC modelAREXAmide,0.625µT0.013 (*)0.75178.95
AREXNOE,0.625µT0.008 (**)
1/(RA·T2A)0.004 (**)
T10.004 (**)
2. Step AIC model with PCAMTRAsym,0.625µT0.014 (*)0.71178.94
1/(RA·T2A)0.030 (*)
Multivariable analysis including perfusion rCBV with multiple imputation
3. Step AIC model1/(RA·T2A)0.012 (*)0.67
[0.67, 0.67]		N/A
T10.028 (*)
AREXAmide,0.625µT0.047 (*)
AREXNOE,0.625µT0.026 (*)
4. Step AIC model with PCA1/(RA·T2A)0.017 (*)0.67
[0.65, 0.69]		N/A

Multivariable logistic regression was performed with and without perfusion rCBV values included. Principal component analysis (PCA) was used to reduce the AREX and MTR parameters in Model 2, whereas Model 1 used all the individual MT/CEST variables as input. Where rCBV was included (ie, Models 3 and 4), multiple imputation was used to deal with missing data. Significant parameters and values are bolded, and asterisks (*) and (**) indicate P-values below .05 and .01, respectively. For AUCs with multiple imputation and cross-validation, the 1st and 3rd quartile are shown in brackets.

Close
This Feature Is Available To Subscribers Only

Sign In or Create an Account

Close

This PDF is available to Subscribers Only

View Article Abstract & Purchase Options

For full access to this pdf, sign in to an existing account, or purchase an annual subscription.

Close