MODIFIKASI ARSITEKTUR U-NET MENGGUNAKAN EFFICIENTNET DAN ATTENTION GATE UNTUK SEGMENTASI SEMANTIK KANKER SERVIKS PADA CITRA PAP SMEAR

RANTI, DWI and Desiani, Anita and Resti, Yulia (2025) MODIFIKASI ARSITEKTUR U-NET MENGGUNAKAN EFFICIENTNET DAN ATTENTION GATE UNTUK SEGMENTASI SEMANTIK KANKER SERVIKS PADA CITRA PAP SMEAR. Undergraduate thesis, Sriwijaya University.

Preview	Image RAMA_44201_08011182126002_cover.jpg - Cover Image Available under License Creative Commons Public Domain Dedication. Download (108kB) | Preview
	Text RAMA_44201_08011182126002.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (19MB) | Request a copy
	Text RAMA_44201_08011182126002_TURNITIN.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (11MB) | Request a copy
	Text RAMA_44201_08011182126002_0011127702_0019077302_01_front_ref.pdf - Accepted Version Available under License Creative Commons Public Domain Dedication. Download (5MB)
	Text RAMA_44201_08011182126002_0011127702_0019077302_02.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (4MB) | Request a copy
	Text RAMA_44201_08011182126002_0011127702_0019077302_03.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (1MB) | Request a copy
	Text RAMA_44201_08011182126002_0011127702_0019077302_04.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (7MB) | Request a copy
	Text RAMA_44201_08011182126002_0011127702_0019077302_05.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (189kB) | Request a copy
	Text RAMA_44201_08011182126002_0011127702_0019077302_06_ref.pdf - Bibliography Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (2MB) | Request a copy

Abstract

Cervical cancer is a type of cancer that develops in the cervix area due to infection by the Human Papillomavirus (HPV). Early diagnosis of cervical cancer can be performed through several methods, one of which is the Pap smear test. In Pap smear images, abnormal changes can be observed in the nucleus, cytoplasm, and folded cytoplasm of the cells. To accurately separate these cells, an image segmentation process is required. Manual image segmentation is time-consuming and prone to errors, hence the need for automatic image segmentation using deep learning methods, specifically Convolutional Neural Networks (CNNs). One CNN architecture commonly used for image segmentation is U-Net. This study proposes a modified U-Net architecture by incorporating EfficientNet in the encoder and Attention Gate in the decoder for semantic segmentation of Pap smear images. EfficientNet addresses the issue of shallow learning layers in the encoder, which limit the model’s ability to learn from complex images. The Attention Gate in the decoder’s skip connections helps reduce the inclusion of irrelevant information. The proposed architecture achieved an accuracy of 93.82%, indicating that the model can accurately segment almost all labels in the image. The sensitivity of 83.14% shows that the model performs well in predicting relevant labels in Pap smear images. A specificity of 95.85% indicates strong performance in identifying non-cancerous labels. The F1-Score of 83.03% reflects a good balance between sensitivity and specificity across all labels. The Intersection over Union (IoU) score of 72% suggests a fairly good match between the prediction and the ground truth. Overall, the results show that the model performs well in segmenting cervical cancer in Pap smear images. However, improvements in sensitivity, F1-Score, and IoU are needed to further optimize performance.

Item Type:	Thesis (Undergraduate)
Uncontrolled Keywords:	Segmentasi Semantik, Kanker Serviks, Citra Pap Smear, U-Net, EfficientNet, Attention Gate
Subjects:	Q Science > QA Mathematics > QA8.9-QA10.3 Computer science. Artificial intelligence. Computational complexity. Data structures (Computer scienc. Mathematical Logic and Formal Languages
Divisions:	08-Faculty of Mathematics and Natural Science > 44201-Mathematics (S1)
Depositing User:	Dwi Ranti
Date Deposited:	26 May 2025 02:20
Last Modified:	26 May 2025 02:20
URI:	http://repository.unsri.ac.id/id/eprint/174028

Actions (login required)

View Item