Segmentasi Citra untuk Menentukan Skor Kerusakan Hepar Secara Histologis
Abstract
Hepatic disease is the eighth leading cause of death in Indonesia. So far, histologists still use manual way to calculate hepar damage score. With digital image technique is expected to be found the pattern and special characteristics that will form a system that can calculate the score of hepatic damage automatically. Image processing method used is image segmentation, whereas to compare the results of the system and expert manual calculations used kappa test. The system can recognize four classes of abnormalities in a single field of view. Of the 16 attributes in the can, it turns out only the 5 most influential attributes in the data obtained are min, max, mode, perimeter and skew. In comparison with the three algorithms (Ibk, Naive Bayes, and J48), calculations with the J48 algorithm have the highest accuracy of 86%. From the kappa test conducted in the average know value of 0.61-0.80 so that the tightness of kappa agreement is said to be strong (good).