Penerapan Metode CNN-LSTM dalam Klasifikasi Penyakit Paru Berdasarkan Citra X-Ray

Abstract

Tuberculosis, pneumonia and lung cancer are lung diseases that are often suffered by people in Indonesia. Then in early 2020, a new disease that entered Indonesia, namely COVID-19 emerged. The symptoms between pneumonia and COVID-19 are similar so it is sometimes difficult to distinguish without a medical examination. Tuberculosis and lung cancer share a common feature, namely the presence of nodules in the lungs. A lung nodule is an opacity that is less than three centimeters in diameter and if it is more than three centimeters then it should be considered whether it has signs of lung cancer. Diagnosis of lung disease can be done through X-Ray examination by a doctor but it cannot be denied that misdiagnosis may occur for some reason. To reduce these misdiagnoses, a system that can classify diseases from X-Ray images is a necessity. The method applied in this research is CNN-LSTM. The stages conducted before the classification process are gray scaling, pre-processing and image augmentation. After those process, the features in the image are extracted using CNN and then classified using LSTM. The data used amounted to 1250 consisting of 875 training data, 250 validation data and 125 testing data. Based on the tests conducted, the system is able to classify those four lung diseases with 93.6% accuracy.