Klasifikasi Tumor Payudara Pada Citra Ultrasonografi Menggunakan Multi-fitur Tekstur dan Support Vector Machine
Keywords:
breast tumor, features extraction, texture, classificationAbstract
Breast cancer is a prevalent type of cancer affecting women worldwide. Additionally, Globocan reported nearly 400,000 new cancer cases in Indonesia in 2020, with 16% being breast cancer. The Ministry of Health has prioritised breast cancer treatment due to the high number of cases. Early detection is a crucial factor in increasing patient life expectancy. Stage 1 breast cancer, for instance, has a 5-year life expectancy of 100%. Breast ultrasound or mammary ultrasound is a commonly used method to detect various breast problems, including cysts and tumors. It is a relatively easy procedure, and the necessary equipment is generally available at Health Facility 1. Texture features are extracted from breast ultrasound images using Gabor and Gray Level Co-occurrence Matrix (GLCM) techniques. The resulting feature vector is then selected and its dimensions reduced to simplify the computing process. This vector is then used to train an SVM classifier to distinguish between benign and malignant cases. The accuracy of the classifier is 0.67 (training) and 0.66 (validation). Meanwhile, the loss obtained during training was 0.77 and during validation was 0.84. Further improvement is required for the accuracy of the model to be applicable.