Institusion
Institut Teknologi Perusahaan Listrik Negara
Author
Hepitia, Dila
Agtriadi, Herman Bedi
Kuswardani, Dwina
Subject
Teknik Informatika
Datestamp
2023-06-21 01:56:09
Abstract :
Breast cancer is a condition in which abnormal cells grow in the breast tissue. Breast cancer
usually grows in several parts, one of which is in the glands that produce milk (lobules). One way to
check for breast cancer is by using ultrasound which is useful for seeing the inside of the breast
whether there are signs of breast cancer that cannot be seen by the eye directly, but the results of the
ultrasound do not always produce good results so that it can influence doctors to This is the goal of
the researcher in conducting this research which implements the Convolutional Neural Network
method on the system created, which aims to be a comparison between visual observations and
computational observations in increasing accuracy in diagnosis in classification of ultrasound images
that detect between normal, benign, malignant. In this study also use the Sharpen kernel in the
convolution process because the sharpen kernel (Matrix 3x3) helps in sharpening the image. In
conducting this research, the researcher used a dataset of 520. To get the results of testing the
classification system that was created, the Confusion Matrix was used. And the results based on the
training model process obtained an accuracy of 76% for the 20th epoch plan. By taking 30 samples of
the data tested on the system, the accuracy of the system was 93%.