|
DSpace@UM >
Faculty of Computer Science and Information Technology >
Masters Dissertations: Computer Science >
Please use this identifier to cite or link to this item:
http://hdl.handle.net/1812/977
|
|
| Title: | Feature extraction and classification of Cardiotocogram signal to detect fetal condition |
| Authors: | Raseeda Hamzah |
| Keywords: | Cardiotocogram CTG Hilbert Huang Transform HHT Intrinsic Mode function IMF Fetal condition detection |
| Issue Date: | 2010 |
| Publisher: | University Malaya |
| Abstract: | ABSTRACT
Cardiotocogram (CTG) is a machine to detect fetal condition in mother’s abdomen by attaching electrode to mother’s abdomen surface. The fetal condition would be diagnosed by an expert based on the CTG signal. However there is a need for a computer based analysis especially when the doctors are not available or in condition when the monitoring and analysis need to be done continuously. Therefore CTG software may help in assisting the initial diagnosis for fetal’s condition (normal or abnormal). Hilbert Huang Transform (HHT) is used in this project because of the capability in analysing nonlinear and nonstationary signal. It is also able to generate Intrinsic Mode function (IMF) which can provide meaningful information of the signal. The classification is done by using Linear Discriminant Analysis (LDA) and Artificial Neural Network (ANN). The aim of the classification is to differentiate between two groups of CTG signal. In this project, engine development and simulation is using Matlab software. A total of 110 datasets (CTG signal)gathered from Hospital Universiti Kebangsaan Malaysia are used in this project to test the efficiency of the engine. In comparison between LDA and ANN, the classification performance shows that LDA is better. LDA had achieved up to 96% correct classification while ANN has achieved 75% correct classification.
ABSTRAK
Kardiotokogram adalah sebuah mesin yang mengesan keadaan bayi dalam kandungan ibu
dengan melekatkan wayar pengesan kepada permukaan perut. Keadaan bayi akan di analisa
oleh pakar melalui isyarat mesin KTG. Walaubagaimanapun, analisa berasaskan computer diperlukan apabila ketiadaan pakar atau dalam keadaan pemantauan dan analisa perlu dilakukan berterusan. Oleh itu perisian KTG boleh membantu dalam analisa awal keadaan bayi. Penukar Huang Hilbert digunakan dalam projek ini kerana kebolehannya dalam menganalisa data tidak linear dan tidak static. Ia juga berkebolehan menghasilkan Fungsi Keadaan Intrinsik yang memberi informasi berguna Pengkelasan dilakukan menggunakan Analisis Pembeza Layan Linear dan Jaringan Saraf uatan. Tujuan pengelasan adalah bagi
membezakan antara 2 kelas bagi isyarat KTG. Dalam projek ini, pembangunan enjin dan simulasi adalah menggunakan perisian Matlab. Sejumlah 110 set data (Isyarat KTG)
dikumpulkan dari Hospital Universiti Kebangsaan Malaysia digunakan dalam projek ini untuk menguji kecekapan enjin. Dalam perbandingan antara LDA dan ANN, prestasi
pengelasan menunjukkan LDA lebih baik. LDA telah mencapai sehingga 96% klasifikasi
betul manakala ANN telah mencapai 75% klasifikasi betul. |
| Description: | Dissertation (M.C.S.) -- Faculty of Computer Science & Information Technology, University of Malaya, 2010. |
| URI: | http://dspace.fsktm.um.edu.my/handle/1812/977 |
| Appears in Collections: | Masters Dissertations: Computer Science
|
This item is protected by original copyright
|
|