Tambunan, Rivaldo Panangian (2025) Klasifikasi Citra Hasil Endoskopi Pada Sistem Gastrointestinal Bagian Bawah Menggunakan Metode Transfer Learning dengan Convolutional Neural Network. Other thesis, Institut Teknologi Sepuluh Nopember.
![]() |
Text
5025201134_Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 April 2027. Download (2MB) | Request a copy |
Abstract
Sistem gastrointestinal sering menjadi perhatian utama dalam penelitian medis karena berbagai gangguan, seperti polip dan kolitis ulseratif, yang jika tidak segera ditangani dapat berkembang menjadi kondisi yang serius. Endoskopi merupakan metode utama yang digunakan untuk mendeteksi penyakit ini, meskipun prosesnya sering memakan waktu dan membutuhkan tenaga ahli yang signifikan. Penelitian ini bertujuan untuk klasifikasi citra hasil endoskopi dengan memanfaatkan metode Transfer Learning berbasis Convolutional Neural Network (CNN). Dataset HyperKvasir dan GastroVision digunakan sebagai dataset citra, terdapat tiga kelas : polip, kolitis ulseratif, dan mukosa normal. Model pre-trained seperti VGG19, ResNet101V2, dan InceptionV3 akan digunakan sebagai ekstraksi fitur dan menambahkan fully connected layer untuk melakukan klasifikasi kelas polip, kolitis ulseratif, dan mukosa normal. Dalam penggunaan model pre-trained tersebut akan menggunakan teknik finetuning pada layer awal setiap model pre-trained. Hasil eksperimen menunjukkan bahwa model pre-trained ResNet101V2 memberikan hasil terbaik dengan tingkat akurasi, recall, dan F1-score yang tinggi. Dengan Hasil akurasi sebesar 0.9881, loss 0.0687, Recall 0.9881, Presicion 0.9882, dan F1-Score 0.9881. Penelitian ini diharapkan dapat berkontribusi dalam mendukung deteksi dini penyakit gastrointestinal secara lebih cepat dan efisien, sekaligus mempermudah proses diagnosis di bidang medis.
==============================================================================================================================
The gastrointestinal system is a major focus in medical research due to various disorders, such as polyps and ulcerative colitis, which, if left untreated, can develop into severe conditions. Endoscopy is the primary method used to detect these diseases, although the process often requires significant time and specialized expertise. This study aims to classify endoscopic images using a Transfer Learning approach based on Convolutional Neural Networks (CNN). The HyperKvasir and GastroVision datasets were utilized, consisting of three main classes: polyps, ulcerative colitis, and normal mucosa. Pre-trained models such as VGG19, ResNet101V2, and InceptionV3 were employed for feature extraction, with the addition of fully connected layers for classifying the three aforementioned categories. Fine-tuning techniques were applied to the initial layers of each pre-trained model to optimize performance. Experimental results demonstrated that the ResNet101V2 pre-trained model achieved the best performance, with high accuracy, recall, and F1-score. The final results include an accuracy of 0.9881, loss of 0.0687, recall of 0.9881, precision of 0.9882, and F1-score of 0.9881.This research is expected to contribute to the early detection of gastrointestinal diseases more efficiently and effectively, thereby facilitating the diagnostic process in the medical field.
Item Type: | Thesis (Other) |
---|---|
Uncontrolled Keywords: | Polip, Kolitis Ulseratif, Mukosa Normal, Transfer Learning, Convolutional Neural Network, HyperKvasir, GastroVision, Polyps, Ulcerative Colitis, Normal Mucosa |
Subjects: | T Technology > T Technology (General) > T57.5 Data Processing T Technology > T Technology (General) > T57.8 Nonlinear programming. Support vector machine. Wavelets. Hidden Markov models. T Technology > T Technology (General) > T59.7 Human-machine systems. |
Divisions: | Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Informatics Engineering > 55201-(S1) Undergraduate Thesis |
Depositing User: | Rivaldo Panangian Tambunan |
Date Deposited: | 02 Feb 2025 08:35 |
Last Modified: | 02 Feb 2025 08:35 |
URI: | http://repository.its.ac.id/id/eprint/117621 |
Actions (login required)
![]() |
View Item |