Kernel Spherical K-Means and Support Vector Machine for Acute Sinusitis Classification

Abstract

Acute sinusitis is an inflammation of the sinus which causes the cavity around the sinus to swells due to accumulated mucus. It makes the patient experience difficulty in breathing through the nose. Generally, it is caused by the common cold, and in most cases, the patient recovers within seven to ten days. However, persistent acute sinusitis can cause severe infections and other complications. Therefore, it requires timely detection and more accurate method of classification. Many techniques have been used to classify acute sinusitis but, in this study, the machine learning methods which includes Kernel Spherical K-Means (KSPKM) and Support Vector Machine (SVM) was applied. SPKM is the application of K-Means, in this research, it was modified by changing the inner product with kernel function to ensure linear data separation on higher dimensions for the maximization of SPKM performance. The SVM is a binary classification method that helps to create a model with good generalization ability. We used CT scan result data from RSCM, Central Jakarta. Simulations were performed with different percentage of training data. The results were compared in terms of Accuracy and Running Time. The score showed that the performance of KSPKM attained an accuracy rate of 97%, while SVM reached 90%.

The comparison study of kernel KC-means and support vector machines for classifying schizophrenia

“…Moreover, the kernel approach is used to resolve a simple hyperplane that is useful in a classification problem with only two classes. In other words, this method brings the form of mapping input into a space that has higher dimension to support nonlinear classification problems where the hyperplane causes the maximum separation between each class [24]. Therefore, the optimization problem observed is defined to minimize the following function [25] while determining the value of weight w ∈ ℝ n and the bias b ∈ ℝ n .…”

Acute sinusitis data classification using grey wolf optimization-based support vector machine

“…There are several methods in diagnosing sinusitis i.e. computed tomography scanning (CT scan), nasal endoscopy, nasal and sinus culture, allergy test and magnetic resonance imaging (MRI) [4]. In this research, the CT scan data attributes will be used for classification methods.…”

“…The previous research on support vector machine was proposed to process this data [4]. Also, many researches before using other methods of classification for classifying disease data, such as binary logistic regression for ovarian cancer classification [5], fast fuzzy clustering for breast cancer data [6], naive bayes classifier for predicting colon cancer [7], random forest classifier for predicting prostate cancer [8], and many more.…”

Accelerating Spherical k-Means

“…Spherical k-means [8] is a widely used adaptation of the k-means clustering algorithm to high-dimensional sparse data, such as document vectors where Cosine similarity is a popular choice. While it is generally used for clustering documents, it has also been applied to medical images [2,20], multivariate species occurrence data [14], and plant leaf images [1]. Because of its importance, several improvements and extensions have been suggested.…”