CLASSIFICATION OF CERVICAL INTRAEPITHELIAL NEOPLASIA BASED ON MACHINE LEARNING AND DEEP LEARNING USING COLPOSCOPY IMAGES

• 主要作者: Zakaria Raga Permana, Zendi
• 格式: Theses
• 語言: Indonesia
• 在線閱讀: https://digilib.itb.ac.id/gdl/view/86901

Cervical lesions are one of the early signs for diagnosing cervical cancer, generally divided into three categories: Cervical Intraepithelial Neoplasia (CIN)1, CIN2, and CIN3. This study aims to develop a CIN classification model based on machine learning and deep learning. The machine learning model is built by utilizing the extraction values of texture and color features in cervical lesion images, namely Gray Level Co-occurence Matrix (GLCM), L*a*b* color space, and Local Binary Pattern (LBP). In contrast, the deep learning model is built by utilizing the basic model Convolutional Neural Network (CNN). CIN images in training and testing data use Intel Mobile Optical Detection Technologies, which is a collection of CIN images tha have been grouped and verified by experts. The experiment in this study was conducted to analyze the performance of the Extra Tree model of 0.98, 0.97, and 0.98. In contrast, the performance of the basic CNN model obtained accuracy, sensitivity, and specificity values of 0.94, 0.96, and 0.89. The findings in this study indicate that the dataset and conduct further testing with more complex classification methods in order to improve the performance of the classification model.