Here, each output neuron i corresponds to one of the m tissue classes X. A voxel represented by its feature vector x is assigned to the class X(x) = i with maximal activation y (x) of the corresponding output neuron i:

Assignment of all the voxels to a tissue class according to Eq. (49) finally yields the segmentation of the data set.

This procedure requires the supervised training of the output weights Sj. For this purpose, the global training procedure explained in Section 3.3 is employed. The training

TABLE 7 Explanation of the contingency tables used for the comparison of two segmentation procedures: The tables contain a contingency matrix

Manual assignment

TABLE 7 Explanation of the contingency tables used for the comparison of two segmentation procedures: The tables contain a contingency matrix

Manual assignment

Was this article helpful?

## Post a comment