Classification of opacities in pneumoconiosis was accomplished by textural analysis in digitised chest x-rays. A good discrimination was achieved by a set of 10 parameters. These texture measures were computed by algorithms for edge detection, local extremes, difference statistics, the co-occurrence matrix and the power spectrum. The classes of the training set were classified correctly at 99%. A test set comprising additional classes that were not contained in the training set, was classified at 82%.
Download full-text PDF |
Source |
---|---|
http://dx.doi.org/10.1055/s-2008-1048714 | DOI Listing |
Enter search terms and have AI summaries delivered each week - change queries or unsubscribe any time!