Csf

294,677 662 1,951

90,075

After preprocessing, the data sets were segmented by applying two different strategies:

• Manual interactive assignment of codebook vectors to tissue classes according to Section 8.1.

• Semiautomatic classification by a GRBF network according to Section 8.2.

In the following, the results and their evaluation are discussed with respect to the relation between segmentation quality and the amount of human intervention required for the segmentation procedures.

At first, the results for vector quantization of the gray-value feature space with manual interactive assignment of codebook vectors to tissue classes will be discussed. The parameters used for minimal free energy vector quantization are listed in Table 3.

As can be seen from Table 2, the simple manual assignment of codebook vectors already yields good segmentation results. Typical examples are presented in Figs. 9b, 10b, 11b.

The average manual processing time for the assignment of codebook vectors to tissue classes was 6 minutes per data set, i.e., the procedure requires only a little human intervention.

However, this approach sometimes yields suboptimal results, especially in cases where a codebook vector cannot be

TABLE 11 Contingency table for data set 4; class labels corresponded in 609,789 of 659,031 voxels (92.5%)

Manual assignment

TABLE 11 Contingency table for data set 4; class labels corresponded in 609,789 of 659,031 voxels (92.5%)

Manual assignment

0 0

Post a comment