IMPLEMENTASI ARSITEKTUR CONVLSTM-U DALAM SEGMENTASI SEMANTIK SEL TUNGGAL KANKER SERVIKS PADA CITRA PAP SMEAR

NUGROHOPUTRI, RIFA FADHILA and Zayanti, Des Alwine and Desiani, Anita (2022) IMPLEMENTASI ARSITEKTUR CONVLSTM-U DALAM SEGMENTASI SEMANTIK SEL TUNGGAL KANKER SERVIKS PADA CITRA PAP SMEAR. Undergraduate thesis, Sriwijaya University.

	Text RAMA_44201_08011281823039.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (2MB) | Request a copy
Preview	Text RAMA_44201_08011281823039_0004127001_0011127702_01_front_ref.pdf - Accepted Version Available under License Creative Commons Public Domain Dedication. Download (1MB) | Preview
	Text RAMA_44201_08011281823039_0004127001_0011127702_02.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (783kB) | Request a copy
	Text RAMA_44201_08011281823039_0004127001_0011127702_03.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (302kB) | Request a copy
	Text RAMA_44201_08011281823039_0004127001_0011127702_04.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (843kB) | Request a copy
	Text RAMA_44201_08011281823039_0004127001_0011127702_05.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (167kB) | Request a copy
	Text RAMA_44201_08011281823039_0004127001_0011127702_06_ref.pdf - Bibliography Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (306kB) | Request a copy
	Text RAMA_44201_08011281823039_TURNITIN.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (10MB) | Request a copy

Abstract

Attempt to detect cervical cancer with pap smear examination are prone to errors because it is done manually. In this case, a medical image processing system is needed to reduce cervical cancer misdiagnosis by applying semantic image segmentation using Convolutional Neural Network (CNN) method. One of the CNN architectures that has good performance in biomedical image segmentation is U-Net. Some of the weaknesses of the U-Net architecture can be solved by ConvLSTM which is a modification of the LSTM architecture to adapt to image data. In this study, the implementation of the ConvLSTM-U architecture which is a modification of the U-Net architecture with ConvLSTM for segmentation of 4 labels on pap smear images. The stages of research are data preprocessing, training, testing, evaluation, analysis, interpretation of results, and conclusion of research results. The results of segmentation research using the Herlev dataset generate an accuracy value of 90.99%, Intersection over Union (IoU) of 63.20%, F1-Score of 75.22%, precision of 74.65%, sensitivity of 76.36%, and specificity of 93.48%. Based on the results obtained, it can be said that the ConvLSTM-U architecture is very good in performing semantic segmentation of cervical cancer single cells, the prediction results given are very close to ground truth, architecture do not cause over-segmentation, architecture are sufficient in detecting the features you want to segment however, it is still not good at detecting pixels in each class, especially in the thin cytoplasmic class.

Item Type:	Thesis (Undergraduate)
Uncontrolled Keywords:	Segmentasi semantik, sel tunggal, kanker serviks, U-Net, ConvLSTM
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:	Rifa Fadhila Nugrohoptri
Date Deposited:	10 Aug 2022 06:35
Last Modified:	10 Aug 2022 06:35
URI:	http://repository.unsri.ac.id/id/eprint/76966

Actions (login required)

View Item