Classification of sinusitis using kernel three-way c-means

Sinusitis can be defined as acute and chronic sinusitis, according to the duration of symptoms. In this study, kernel three-way c-means, as the modification of the three-way c-means method that used kernel distance instead of Euclidean distance, was used. Three-way c-means itself is the upgrade version of the rough k-means algorithm that integrates three-way weight and three-way assignments to assign data points into clusters with the appropriate weight. The performance was later compared using the sinusitis dataset taken from Cipto Mangunkusumo Hospital, Indonesia, which was consists of 102 acute and 98 chronic sinusitis samples. From the experiments, three-way c-means was obtained 62.09% accuracy, 55.21% sensitivity, 62.76% precision, 68.77% specificity, and 58.59% F1-Score in 1.82 seconds. Meanwhile, kernel three-way c-means with the 8th polynomial kernel was provided 67.48% accuracy, 74.82% sensitivity, 64.52% precision, 60.77% specificity, and 69.12% F1-Score in 2.24 seconds. Therefore, it was concluded that kernel three-ways c-means performs better with the slower running time than the three-way c-means.