PREPROCESSING PADA SEGMENTASI CITRA PARU-PARU DAN JANTUNG MENGGUNAKAN ANISOTROPIC DIFFUSION FILTER

Abstract

This paper propose a preprocessing techniques in

lung segmentation scheme using Anisotropic Diffusion filters.

The aim is to improve the accuracy, sensitivity and specificity

results of segmentation. This method was chosen because it

has the ability to detect the edge, namely in doing smoothing,

this method can obscure noise, while maintaining the edges of

objects in the image. Characteristics such as this is needed to

process medical image filter, where the boundary between the

organ and the background is not so clear. The segmentation

process is done by K-means Clustering and Active Contour to

segment the lungs. Segmentation results were validated using

the Receiver Operating Characteristic (ROC) showed an

increased accuracy, sensitivity and specificity, when

compared with the results of segmentation in the previous

paper, in which the preprocessing method used is Gaussian

Lowpass filter.