Perbandingan Kinerja Identifikasi Model VGG-19 Dengan Inception V3 Dalam Klasifikasi Penyakit Appendicitis

Abstract

Appendicitis is a surgical emergency that requires rapid and accurate diagnosis. However, limitations in ultrasound (USG) image interpretation often pose a risk of misdiagnosis, particularly in scenarios with limited medical data. This study aims to determine the most effective classification model for a clinical decision support system by comparing two transfer learning-based Convolutional Neural Network (CNN) architectures: VGG-19 and InceptionV3. Utilizing a dataset of 2,168 images split into 70% training, 10% validation, and 20% testing data, the models were evaluated using metrics such as accuracy, precision, recall, F1-score, and Area Under Curve (AUC). The results demonstrate that InceptionV3 delivered significantly superior performance, achieving an accuracy of 0.9033%, an F1-score of 0.8946% for the appendicitis class, and an AUC of 0.9502%. In contrast, VGG-19 only reached an accuracy of 0.8255%, with a recall for the appendicitis class as low as 0.8019%. The poor recall performance of VGG-19 indicates a high risk of missed diagnosis. This research contributes by recommending a more reliable and effective model to support AI-based appendicitis identification, specifically in limited data scenarios.