Segmentasi Tumor Payudara Pada Citra Breast Ultrasound 2 Dimensi Dengan Metode Active Contour

Helena, Helena (2020) Segmentasi Tumor Payudara Pada Citra Breast Ultrasound 2 Dimensi Dengan Metode Active Contour. Undergraduate thesis, Institut Teknologi Sepuluh Nopember.

[img] Text
07311640000019-Undergraduate_Thesis.pdf - Accepted Version
Restricted to Repository staff only

Download (8MB) | Request a copy

Abstract

Ultrasonografi adalah salah satu modalitas pencitraan yang digunakan untuk mendeteksi kelainan massa nodul payudara. Ultrasonografi payudara tidak memiliki efek radiasi pengion dan memiliki resolusi kontras yang baik. Speckle adalah sumber utama noise pada citra ultrasonografi yang dapat mengurangi kontras dan batas citra. Maka, perlu dilakukan pereduksian speckle noise tanpa merusak fitur citra. Pada bagian pra-pemrosesan dilakukan resizing, grayscalling, peningkatan kontras dengan metode power law transformation, dan filtering dengan menggunakan penggabungan metode wavelet denoising, bilateral filter, dan anisotropic diffusion filter. Pada bagian segmentasi digunakan metode Simple Linear Iterative Clustering (SLIC), otsu thresholding, active contour untuk mendapatkan batas antara sel-sel tumor dengan sel-sel normal pada citra, dan dilakukan estimasi luasan tumor menggunakan metode shoelace. Berdasarkan analisa statistik kuantitas, maka hasil restorasi citra yang paling baik didapatkan dengan nilai gamma sebesar 0.6 pada tahap gamma correction; jumlah iterasi sebanyak 50 kali, nilai kappa 5, dan nilai delta 0.25 pada metode anisotropic diffusion filter, serta dapat diketahui bahwa penggabungan metode wavelet denoising, bilateral filter, dan anisotropic diffusion filter yang digunakan lebih efektif dalam mereduksi speckle noise dibandingan dengan metode lee filter, Gaussian filter, dan median filter. Hasil segmentasi untuk pendeteksian tumor pada citra breast ultrasound cukup baik. Hal ini dilihat dari tingkat akurasi sistem sebesar 89%, tingkat sensitivitas 100%, tingkat spesifisitas 75%, dan tingkat error 11%. Kemudian, untuk rata-rata error area hasil segmentasi terhadap area hand-mark didapatkan sebesar 30%. Hasil penelitian ini dapat menjadi bahan pertimbangan dalam mendiagnosa tumor payudara pada citra ultrasound. ================================================================================================================== Ultrasonography is one of the imaging modalities used to detection breast nodule mass abnormalities. Breast ultrasonography has no ionizing radiation effect and has good contrast resolution. Spots are the main source of noise in ultrasonographic images that can reduce image contrast and boundaries. So, it is necessary to reduce speckle noise without damaging the image features. In the pre-improvement section changes, grayscalling, contrast enhancement with the power law transformation method, and filtering using a combination of denoising wavelet methods, bilateral filters, and anisotropic filter diffusion. In the segmentation section the Simple Linear Iterative Clustering (SLIC) method, otsu thresholding, active contour are used to obtain the boundary between tumor cells and normal cells in the image, and tumor estimation is performed using the shoelace method. Based on statistical assessments, the best image restoration results are obtained with a gamma value of 0.6 in gamma correction; 50 times iteration, kappa value of 5, and delta value of 0.25 in the anisotropic diffusion filter method, and can be in accordance with the merging of the wavelet denoising method, bilateral filter, and anisotropic diffusion filter used more effectively in reducing speckle noise compared to the lee filter method, Gaussian filters, and median filters. The results of segmentation for tumor detection on breast ultrasound are quite good. This can be seen from the level of system accuracy of 89%, a sensitivity level of 100%, a specificity level of 75%, and an error rate of 11%. Then, for the average error area resulting from system segmentation compared to the hand-mark area obtained by 30%. The results of this study can be taken into consideration in diagnosing breast tumors on ultrasound images.

Item Type: Thesis (Undergraduate)
Uncontrolled Keywords: Active contour, Bilateral filter, pendeteksian tumor, Speckle noise, Wavelet denoising, Anisotropic diffusion filter, tumor detection.
Subjects: Q Science > QM Human anatomy
R Medicine > R Medicine (General)
T Technology > T Technology (General)
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.
Divisions: Faculty of Electrical Technology > Biomedical Engineering > 11410-(S1) Undergraduate Thesis
Depositing User: Helena -
Date Deposited: 26 Aug 2020 07:14
Last Modified: 26 Aug 2020 07:14
URI: https://repository.its.ac.id/id/eprint/80981

Actions (login required)

View Item View Item