Segmentasi Sel Bertumpuk pada Citra Mikroskopis Sel Kanker Payudara menggunakan Spatial Fuzzy C-Means Clustering dan Rapid Region Merging

Tuwohingide, Desmin (2017) Segmentasi Sel Bertumpuk pada Citra Mikroskopis Sel Kanker Payudara menggunakan Spatial Fuzzy C-Means Clustering dan Rapid Region Merging. Masters thesis, Insitut Teknologi Sepuluh Nopember.

[img] Text
5114201049-Master-theses.pdf
Restricted to Repository staff only

Download (3MB) | Request a copy

Abstract

Penerapan teknik pengolahan citra untuk menganalisis citra mikroskopis sel kanker payudara dilakukan untuk mempermudah diagnosis penyakit kanker payudara. Proses pemisahan sel kanker payudara bertumpuk dianggap penting karena hasil pemisahan sel kanker bertumpuk akan mempengaruhi akurasi perhitungan jumlah sel. Keberhasilan proses pemisahan sel bertumpuk juga dipengaruhi oleh proses identifikasi sel, proses deteksi sel bertumpuk dan penanganan masalah over-segmentation. Pemisahan sel kanker menggunakan algoritma clustering pada citra mikroskopis sel darah putih menghasilkan nilai akurasi yang cukup baik. Kombinasi metode Spatial Fuzzy C-means Clustering (SFCM) dan Rapid Region Merging (RRM) untuk pemisahan sel kanker bertumpuk dan penanganan masalah over-segmentation dipaparkan pada penelitian ini. Citra masukan yang digunakan pada tahapan pemisahan sel bertumpuk adalah citra hasil identifikasi sel kanker payudara berdasarkan metode Gram-Schmidt, sedangkan sel kanker yang diproses pada tahapan pemisahan sel kanker bertumpuk adalah sel kanker yang dideteksi bertumpuk berdasarkan informasi fitur geometri area. Berdasarkan hasil pengujian dilakukan terhadap 40 citra mikroskopis jenis benign dan malignant, kombinasi metode SFCM dan RRM memberikan hasil paling baik berdasarkan perolehan nilai rata-rata Mean Square Error (MSE) sebesar 0,07 pada tahapan identifikasi sel dan nilai akurasi pemisahan sel bertumpuk sebesar 78.41% ================================================================= The application of image processing techniques to analyze the microscopic image of the breast cancer cells was done to make the diagnosis of breast cancer easier. The separation process of overlapped breast cancer cells is important because the separation result of overlapped cancer cells will affect the accuracy of cell counting. The success of overlapped cells separation are also affected by cell identification process, overlapped cell detection process and the handling of over-segmentation problems. The separation of cancer cells using clustering algorithm on white blood cells microscopic image produce a fairly good accuracy. The combination of Spatial Fuzzy C-Means (SFCM) and Rapid Region Merging (RRM) method for separating the overlapped cancer cells and handling the over-segmentation problems are presented in this study. The input image used in overlapped cell separation phases is the image from identification result of breast cancer cell by Gram-Schmidt method, where as the overlapped cancer cells that are processed at separation phase is detected by the area information from geomatric features. Based on the evaluation on 40 microscopic image of benign and malignant types, the combinations between SFCM and RRM method provides superior results with average value of Mean Square Error (MSE) is 0,07 on cell identification phase and the accuracy value of overlapped cells separation is 78,41%.

Item Type: Thesis (Masters)
Uncontrolled Keywords: Segmentasi sel kanker; Segmentasi Sel Bertumpuk; Kanker Payudara; Spatial Fuzzy C-Means; Rapid Region Merging; Segmentasi sel kanker; Segmentasi Sel Bertumpuk; Kanker Payudara.
Subjects: R Medicine > RB Pathology
Divisions: Faculty of Information Technology > Informatics Engineering > (S2) Master Theses
Depositing User: DESMIN TUWOHINGIDE -
Date Deposited: 17 Mar 2017 03:53
Last Modified: 19 Dec 2017 07:38
URI: http://repository.its.ac.id/id/eprint/2142

Actions (login required)

View Item View Item