Biomarker Extraction Based on Subspace Learning for the Prediction of Mild Cognitive Impairment Conversion

Algorithm 1

Projection matrices learning algorithm based on auxiliary data.

Input: The cortical thickness matrix of AD and NC subjects ;

  The cortical volume matrix of AD and NC subjects ;

  The corresponding label matrix ;

  The balancing parameters ;

Output: The two projection vectors and for thickness and volume data.

1. Compute the data affinity matrices and ;

2. Compute the diagonal matrices and ;

3. Compute the Laplacian matrices and ;

4. Initialize and with zero matrix;

5. Repeat

6. Update according to (11);

7. Update according to (12);

8. Until convergence.