Adam, Safri (2020) Inisialisasi Otomatis Metode Level Set untuk Segmentasi Objek Overlapping pada Citra Panorama Gigi. Masters thesis, Institut Teknologi Sepuluh Nopember.
Preview |
Text
05111850010034-Master_Thesis.pdf Download (3MB) | Preview |
Abstract
Ekstraksi fitur dari gigi yang sudah disegmentasi mempunyai tantangan ketika dihadapkan pada gigi yang overlap. Segmentasi pada gigi yang overlap menggunakan algoritma terkini masih menghasilkan satu objek sehingga perlu dilakukan pemisahan yang diawali mendapatkan objek overlap terlebih dahulu. Metode level set digunakan untuk melakukan segmentasi objek overlap, namun memiliki kelemahan yaitu perlu didefinisikan inisial awal secara manual oleh pengguna yang cukup melelahkan jika diterapkan pada jumlah data yang banyak. Dalam penelitian ini diusulkan strategi inisialisasi otomatis pada metode level set untuk melakukan segmentasi gigi yang overlap pada citra panorama gigi. Proses melakukan segmentasi gigi yang overlap diawali dengan pendefinisian region of interest (ROI) kemudian dibagi menjadi 4 tahap yaitu: preprocessing, segmentasi objek gigi, segmentasi objek overlap menggunakan metode level set dengan inisialisasi awal otomatis, pemotongan dua gigi overlap. Metode yang digunakan dalam melakukan inisialisasi otomatis pada metode level set memanfaatkan fitur intensitas dan geometri. Inisialisasi objek overlap dengan fitur intensitas menggunakan metode multi-thresholding HCA (hierarchical cluster analysis) dengan menggunakan threshold ketiga. Sedangkan pada fitur geometri dapat memanfaatkan orientasi dari objek overlap yaitu berorientasi vertikal. Dari hasil evaluasi yang diperoleh dapat ditarik beberapa kesimpulan diantaranya: strategi gabungan menggunakan metode multi-thresholding HCA, operasi morfologi erosi, dan seleksi objek berdasarkan fitur geometri, berhasil dalam menentukan lokasi objek overlap dengan baik dengan akurasi sebesar 87%. Segmentasi objek overlap menggunakan metode level set mampu mencapai hasil yang baik dengan nilai ME dan RAE sebesar 0,9342% dan 24,283%. Strategi gabungan yang terdiri dari penentuan titik potong dan operasi aljabar persamaan garis yang melewati dua titik dapat memisahkan dua gigi yang overlap secara interaktif pada proses segmentasi gigi dengan cukup baik terbukti memperoleh nilai ME dan RAE masing-masing sebesar 6,77% dan 13,55%, daripada menggunakan metode otomatis yang memperoleh 16,41% dan 52,14%. Sistem yang telah dibangun diharapkan mampu membantu melakukan segmentasi terhadap citra gigi yang overlap untuk penilaian estimasi usia manusia melalui citra gigi dalam bidang Odontologi Forensik.
=================================================================================================================================
Feature extraction from segmented teeth has challenges when faced with overlapping teeth. Segmentation of overlapping teeth using the latest algorithm still produces one object, so it is necessary to separate that beginning by obtaining the overlapping object. The level set method is used to segment overlap objects. Nevertheless, it has a limitation where the initialization was done manually by the user which is exhausting if applied to the large amount of data. In this study, we proposed an automatic initialization strategy of the level set method to segment the overlapping teeth on dental panoramic radiographs. The process of segmenting overlapping teeth was begun by defining the Region of Interest (ROI), then was divided into 4 stages, i.e. pre-processing, dental objects segmentation, overlapping object segmentation using a level set method with automatic initialization, and separating two overlapping teeth. The method used in automatic initialization of the level set method is facilitated by intensity and geometry features. The initialization of overlap objects with intensity features was conducted by using the multi-thresholding HCA (Hierarchical Cluster Analysis) method with the third threshold. Whereas the geometry features can utilize the orientation of overlap objects, namely vertically oriented. The proposed automatic initialization strategy that consisted of the HCA multi-thresholding method, erosion morphological operations, and object selection based on geometry features were succeed in determining the location of overlapping objects. The level set method was able to initialize overlapping objects properly with accuracy of 87%. The overlap object segmentation using the level set method is able to achieve good results with ME and RAE values of 0.9342% and 24.283%. The combined strategy which consists in determining the intersection point and algebraic operation of the equation of a line passing through two points can separate two teeth that overlap interactively in the process of tooth segmentation. This is evidenced by the ME and RAE values of 6.77% and 13.55%, instead of using the automatic method which obtained 16.41% and 52.14%. The system that has been built is expected to be able to help the segmentation of the overlapping dental images for the assessment of human age estimation through dental images in the field of Forensic Odontology.
| Item Type: | Thesis (Masters) |
|---|---|
| Additional Information: | RTIf 006.4 Ada i-1 2020 |
| Uncontrolled Keywords: | Overlapping, Radiografi Panorama Gigi, Segmentasi |
| Subjects: | T Technology > T Technology (General) > T58.5 Information technology. IT--Auditing T Technology > T Technology (General) > T58.62 Decision support systems T Technology > TA Engineering (General). Civil engineering (General) > TA1637 Image processing--Digital techniques |
| Divisions: | Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Informatics Engineering > 55101-(S2) Master Thesis |
| Depositing User: | Safri Adam |
| Date Deposited: | 18 Dec 2023 01:40 |
| Last Modified: | 18 Dec 2023 01:40 |
| URI: | http://repository.its.ac.id/id/eprint/73092 |
Actions (login required)
 |
View Item |