Abstract:
The possibility of using cluster analysis to differentiate pathologies (cancer and tuberculosis), leading to the appearance of spherical formations in the lungs. As the diagnostic features used parameters are defined on the basis of imaging X-ray computed tomography. We compared the results of two methods of cluster analysis: k-means and hierarchical clustering. The criterion of the quality of the method of cluster analysis is to compare the clustering results to verify the diagnosis. It was established that the complete-linkage method (hierarchical clustering) is more reliable than the k-means method allocates the correct diagnosis.
Keywords:spherical formation in the lungs, fractal dimension, cluster analysis.
Fulltext:
PDF file (286 kB)