Skip to main content
. 2022 May 25;16:832276. doi: 10.3389/fnins.2022.832276

Algorithm 2.

DCL

Input: X1p×n, X2s×n //Input of sMRI and fMRI, respectively.
Output: X^1,X^2
1:  X~1,Lambda1,U1S-POET(X1) //processed by S-POET method
2:  X~2,Lambda2,U2S-POET(X2) //processed by S-POET method
3:  Lambda11 ← Construct diag(Lambda1)
4:  Lambda22 ← Construct diag(Lambda2)
5:  Theta(Lambda11@U1.T@X~1)@(X~2.T@U2@Lambda22)/n
6:  Vtheta, Dtheta ← SVD(Theta, fullmatrices = True) //Singular Value Decomposition
7:  Gamma1U1@Lambda11@Vtheta
8:  Gamma2U2@Lambda22@Vtheta
9:  Amat ← diag(Dtheta) //Diagonal matrix
10:  Cbase ← Common variables corr(X~1,X~2)
11:  C~1 Common matrix (X~1,Cbase,Amat)
12:  C~2 Common matrix (X~2,Cbase,Amat)
13:  D~1 Distinctive matrix (X~1,C~1)
14:  D~2 Distinctive matrix (X~2,C~2)
15:  X~1 Combination of common and distinctive matrices
16:  X~2 Combination of common and distinctive matrices
17:  return X^1, X^2