Faiz, Muhammad Hilman (2018) Segmentasi dan Klasifikasi Citra Sel Darah Putih Menggunakan Metode Pre-Processing Self-Dual Multiscale Morphological Toggle. Undergraduate thesis, Institut Teknologi Sepuluh Nopember.
Preview |
Text
5113100129-Undergraduate_Theses.pdf - Accepted Version Download (2MB) | Preview |
Abstract
Kemajuan teknologi memberikan kemudahan untuk menyelesaikan permasalahan. Dengan adanya pengolahan citra digital, penghitungan dan identifikasi jenis sel darah putih pada blood smear image yang sebelumnya dilakukan secara konvensional kini dapat dilakukan secara otomatis. Meskipun demikian, hingga saat ini belum ada standar metode yang diakui dunia.
Tugas akhir ini mengusulkan sebuah metode self-dual multiscale morphological toggle (SMMT) sebagai metode preprocessing dari blood smear image. Nukleus dan sitoplasma sel darah putih dideteksi dan disegmentasi secara terpisah. Segmentasi nukleus dilakukan menggunakan metode level set. Segmentasi sitoplasma dilakukan menggunakan metode-metode morfologi matematika, yaitu bottom hat, flood fill, dan watershed. Kemudian citra hasil segmentasi diekstrak fitur-fiturnya dan diklasifikasi menggunakan metode decision tree. Data akhir yang didapat adalah jumlah sel darah putih dalam blood smear image dan jenis dari masing-masing sel darah putih yang telah tersegmentasi.
Uji coba yang dilakukan terhadap 247 ctra training dan 30 citra testing menunjukkan bahwa metode ini dapat memberikan hasil segmentasi hapusan darah dan klasifikasi sel darah putih yang akurat dengan rata-rata akurasi, specificity dan sensitivity sebesar 87,78%, 71,94%, dan 88,64%.
========================================================================================================
Using digital image processing, counting and identification of white blood cell types, which previously performed conventionally, now can be done automatically. However, there is no standard method that is recognized worldwide yet.
In this undergraduate thesis, we propose self-dual multiscale morphological toggle (SMMT) as pre-processing method. Nucleus and cytoplasm of white blood cell will be detected and segmented separately. Nucleus segmentation will be performed using level-set method. Cytoplasm segmentation will be done using mathematical morphology methods, namely bottom hat, flood fill, and watershed. Then, segmented image’s features will be extracted and classified using decision tree method. Data produced through these processes will be white blood cell counting and types of each segmented white blood cell.
The experiment was performed for 247 training images and 30 testing images. The result shows that these methods give accurate white blood cell classification with the average accuracy, specificity, and sensitivity of 87,78%, 71,94%, and 88,64%.
| Item Type: | Thesis (Undergraduate) |
|---|---|
| Additional Information: | RSIf 005.1 Fai s-1 3100018074473 |
| Uncontrolled Keywords: | SMMT, morfologi matematika, segmentasi, level-set, sel darah putih, mathematical morphology, segmentation, white blood cell |
| Subjects: | Q Science > QA Mathematics > QA76 Computer software T Technology > TA Engineering (General). Civil engineering (General) > TA1637 Image processing--Digital techniques |
| Divisions: | Faculty of Information Technology > Informatics Engineering > 55201-(S1) Undergraduate Thesis |
| Depositing User: | Muhammad Hilman Faiz |
| Date Deposited: | 22 Feb 2018 03:20 |
| Last Modified: | 13 Apr 2020 00:02 |
| URI: | http://repository.its.ac.id/id/eprint/50698 |
Actions (login required)
 |
View Item |