Deteksi Potensi Aterosklerosis Berdasarkan Tekstur Dinding Arteri pada Citra Ultrasound

Putri, Ayuning Sekar Agriensyah (2025) Deteksi Potensi Aterosklerosis Berdasarkan Tekstur Dinding Arteri pada Citra Ultrasound. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5023211028-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (3MB) | Request a copy

Abstract

Penyakit kardiovaskular (Cardiovascular Disease/CVD) merupakan penyebab utama kematian global, dengan aterosklerosis sebagai bentuk yang paling umum. Kondisi ini ditandai dengan penebalan dan kekakuan dinding arteri akibat akumulasi plak, yang dapat meningkatkan risiko komplikasi seperti stroke dan serangan jantung. Deteksi dini aterosklerosis sangat penting untuk mencegah perkembangan penyakit lebih lanjut, namun metode konvensional seperti pengukuran Intima-Media Thickness (IMT) masih memiliki keterbatasan dalam mendeteksi perubahan mikrostruktural awal. Penelitian ini mengembangkan sistem deteksi potensi aterosklerosis berbasis analisis tekstur citra ultrasound (USG) dinding arteri. Proses dimulai dengan pemrosesan awal citra untuk menghilangkan elemen non-anatomi dan mengurangi speckle noise menggunakan gabungan metode CLAHE, median filter, dan DsFlsmv. Segmentasi dinding arteri dilakukan secara semi-otomatis dengan pendekatan region-based berdasarkan profil intensitas. Hasil segmentasi divalidasi dengan ground truth dan menghasilkan tingkat kecocokan sebesar 98,24%. Dari ROI yang diperoleh, fitur tekstur diekstraksi menggunakan metode Gray Level Difference Statistic (GLDS), Statistical Features (SF), Spatial Gray Level Difference Matrix (SGLDM), Neighborhood Gray Tone Difference Matrix (NGTDM), dan Fractal Dimension Texture Analysis (FDTA) menghasilkan parameter GLDS-kontras, median intensitas (GSM), Angular Second Moment (ASM), Coarseness, H2FDTA, dan H4FDTA. Fitur-fitur tersebut digunakan sebagai input dalam klasifikasi menggunakan Support Vector Machine (SVM). Evaluasi dilakukan terhadap tiga kelompok fitur: klinis, tekstur, dan gabungan. Hasil akurasi terbaik diperoleh dari fitur gabungan menggunakan kernel RBF, dengan akurasi 75,76%, recall 88%, dan AUC 0,81. Hasil ini menunjukkan bahwa integrasi informasi klinis dan tekstur citra dapat meningkatkan akurasi deteksi dini risiko aterosklerosis. Metode ini menunjukkan potensi sebagai pendekatan non-invasif dan terjangkau untuk skrining rutin serta dapat dikembangkan lebih lanjut untuk integrasi ke dalam sistem klinis.

===========================================================================================================================================

Cardiovascular disease (CVD) is the leading cause of death worldwide, with atherosclerosis being its most common form. This condition is characterized by thickening and stiffening of the arterial wall due to plaque buildup, increasing the risk of complications such as stroke and heart attack. Early detection of atherosclerosis is crucial to prevent disease progression, yet conventional methods like Intima-Media Thickness (IMT) measurements have limitations in identifying early microstructural changes.This study developed a system for detecting potential atherosclerosis based on texture analysis of arterial wall ultrasound images. The process began with image preprocessing to remove non-anatomical elements and reduce speckle noise using a combination of CLAHE, median filtering, and DsFlsmv. The arterial wall was segmented semi-automatically using a region-based approach based on intensity profile analysis. The segmentation was validated against ground truth and achieved a matching accuracy of 98.24%. Texture features were extracted from the resulting ROI using the Gray Level Difference Statistic (GLDS), Statistical Features (SF), Spatial Gray Level Difference Matrix (SGLDM), Neighborhood Gray Tone Difference Matrix (NGTDM), and Fractal Dimension Texture Analysis (FDTA) methods, resulting in GLDS-contrast, intensity median (GSM), Angular Second Moment (ASM), Coarseness, H2FDTA, and H4FDTA features. These features were used as input for classification using Support Vector Machine (SVM). Three feature sets were evaluated: clinical, texture-based, and a combination of both. The best accuracy performance was achieved using the combined feature set with a RBF kernel, yielding 75.76% accuracy, 88% recall, and an AUC of 0.81. These results demonstrate that integrating clinical information with image texture can improve the accuracy of early atherosclerosis detection. This method shows potential as a non-invasive and affordable screening approach and may be further developed for integration into clinical systems.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Aterosklerosis, Analisis Tekstur, Region-based, Citra USG, SVM, Atherosclerosis, Texture Analysis, Ultrasound Image, Region-Based Segmentation, SVM
Subjects:	R Medicine > RC Internal medicine > RC691 Blood-vessels--Diseases. R Medicine > RC Internal medicine > RC78.7.U4 Ultrasonic imaging. T Technology > T Technology (General) > T11 Technical writing. Scientific Writing T Technology > T Technology (General) > T57.5 Data Processing T Technology > T Technology (General) > T58.62 Decision support systems T Technology > TA Engineering (General). Civil engineering (General) > TA1637 Image processing--Digital techniques. Image analysis--Data processing.
Divisions:	Faculty of Electrical Technology > Biomedical Engineering > 11410-(S1) Undergraduate Thesis
Depositing User:	Ayuning Sekar Agriensyah Putri
Date Deposited:	04 Aug 2025 06:41
Last Modified:	04 Aug 2025 06:41
URI:	http://repository.its.ac.id/id/eprint/126063

Actions (login required)

View Item