Pre Cervical Cancer Detection on Visual Inspection of Acetic Acid (VIA) Test Image Using K-Means Clustering Method

Abstract

We built an approach in early detection of cervical cancer using image recognition of Visual Inspection with Acetic Acid (VIA) image. We used k-Means clustering to segment the expected region of cervical cell in the VIA test image. The positive suspect shown by white lesion which are called acetowhite. We used VIA test image captured from mobile phone as the dataset. From the acetowhite area, we extracted the color moment feature and the Gray Level Co-occurence Matrix (GLCM) feature. The color moment and GLCM feature were then classified as positive or negative using Support Vector Machine (SVM) classifier. The best performance were an accuracy of 72,14%, with sensitivity of 70% and specificity of 74% using k-Means clustering with k=2 and SVM with linear kernel.