Pengembangan Aplikasi Menggunakan Metode Convolutional Neural Network dengan Arsitektur VGG-16 untuk Identifikasi Penyakit Paru

Abstract

The rate of lung problems around the world continues to increase. According to WHO, different lung diseases including pneumonia, tuberculosis, and Covid-19 disease, the characteristics of these two diseases are almost the same [1]. The cause of death of people with lung disease during the Covid- 19 pandemic is due to the lengthy diagnosis process. Other factors, such as X-ray imaging results, often appear fuzzy and lack contracture, so an image seen by multiple observers can make different diagnoses. The study will classify lung images into four categories: normal lungs, tuberculosis, pneumonia, and Covid-19. The method chosen by the researcher is a Constitutive Neural Network using the VGG-16 architecture. The test results show that the Convolutional Neural Network (CNN) model gets the test results on the model having the highest accuracy in the scenario without using a pre-trained model by using a data comparison of 9:1 at epoch 50, with an accuracy of 94%, while the lowest result is in the scenario test data 8:2 epoch 50 at non-pre- trained model with an accuracy of 87%.