Dear Steve,
Thanks for your help. I have two questions:
1.) To scale the eigenvectors based on the eigenvalues distribution, should I scale all the eigenvectors based in the ratio between the biggest obtained eigenvalue across datasets and the corresponding dataset's eigenvalues?
2.) Should I apply PCA to the dataset, and apply MELODIC to the concatenated PCA maps or should I back-project by using the scaled eigenvectors and apply MELODIC to the resulting subjects maps?
Best,
Gabriel
|