Cancer’s Probability Assessment from Breast Ultrasonograhys by Using Image Processing
Agwin Fahmi Fahanani, Habiba Aurora, Maskuril Barkah Barkah, Irma Darinafitri, Yuyun Yueniwati · Advances in health sciences research/Advances in Health Sciences Research · 2023
In Indonesia, breast cancer is the first leading cause of death in women.Even data from the WHO, shows that in 2020, from total of 213,000 cancer cases, there are 65,000 cases of breast cancer or about 30% of all cancer cases in women.Breast ultrasonograhy (BUS) is an effective way to detect cancer.In addition, BUS can distinguish benign and malignant, so it can reduce unnecessary biopsy procedure.Cancer's probability assessment is necessary to examine by radiologist.One of the standard assessment is breast imaging-reporting and data system (BI-RADS).Radiologist use this standard from BUS to grade the cancer's probability manually, so the result might be subjective.This research offered a new approach to standardize the system of cancer's probability assessment by image processing.The proposed methods in this research begin with the pre-processing.The BUS images have been improved by image filtering and enhancing.After preprocessing, the cancer's probability part has been segmented by semi-automatic methods.The manual method uses for locate cancer's probability part roughly.The automatic method uses for segment the area of cancer.After that, the feature area of cancer has been extracted.The K-nearest neighbors (KNN) algorithm then used to classify the cancer's probability according the BI-RADS.The proposed method produces 0.79 in Cohen's kappa coefficient.This number indicates the system has substantial agreement with radiologist assessment.