CLASSIFICATION OF CHRONIC LYMPHOCYTIC LEUKEMIA USING THE SWIN TRANSFORMER

Dharmawan, Muh.Yassar Nurfajri (2026) CLASSIFICATION OF CHRONIC LYMPHOCYTIC LEUKEMIA USING THE SWIN TRANSFORMER. Undergraduate thesis, UPN Veteran Jawa Timur.

	Text (Cover) 22081010088.-Cover.pdf Download (2MB)
	Text (Bab 1) 22081010088,-Bab 1.pdf Download (327kB)
	Text (Bab 2) 22081010088,-Bab 2.pdf Restricted to Repository staff only until 14 July 2029. Download (585kB) \| Request a copy
	Text (Bab 3) 22081010088,-Bab 3.pdf Restricted to Repository staff only until 14 July 2029. Download (835kB) \| Request a copy
	Text (Bab 4) 22081010088,-Bab 4.pdf Restricted to Repository staff only until 14 July 2029. Download (799kB) \| Request a copy
	Text (Bab 5) 22081010088,-Bab 5.pdf Download (255kB)
	Text (Daftar Pustaka) 22081010088,-Daftar Pustaka.pdf Download (282kB)
	Text (Lampiran) 22081010088,-Lampiran.pdf Restricted to Repository staff only Download (258kB) \| Request a copy

Abstract

Delayed diagnosis of Chronic Lymphocytic Leukemia (CLL) in Indonesia occurs due to the limitations of standard medical facilities such as flow cytometry. This study applies the Swin Transformer algorithm (Swin Tiny) for binary classification of normal and CLL white blood cells using 10,216 microscopic images from Mendeley Data. The preprocessing stage includes image size standardization, Z-score normalization, and the application of class weighting. Hyperparameter optimization is conducted via Grid Search, and the best model is integrated into a Streamlit web application featuring Out-of-Distribution (OOD) detection based on Mahalanobis Distance. The results show that the shifted window attention mechanism effectively extracts local and global features, while class weighting successfully prevents data bias. Evaluation indicates accurate and balanced model performance, with accuracy, precision, recall, and F1-score consistently around 91% (specifically an F1-score of 0.9129), supported by a low error rate on the ROC-AUC curve. The system is also successfully implemented in the web application for real-time prediction. These characteristics prove that the developed model is reliable as a digital tool to support clinical decisions in hematology.

Item Type:

Thesis (Undergraduate)

Contributors:

Contribution	Contributors	NIDN/NIDK	Email
Thesis advisor	Muttaqin, Faisal	NIDN0030058602	faisalmuttaqin.if@upnjatim.ac.id
Thesis advisor	Mulyo, Budi Mukhamad	NIDN0718118904	budi.m.mulyo.fasilkom@upnjatim.ac.id

Subjects:

T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK5105 Computer Network

Divisions:

Faculty of Computer Science > Departemen of Informatics

Depositing User:

Muh Yassar Nurfajri Dharmawan

Date Deposited:

21 Jul 2026 02:07

Last Modified:

21 Jul 2026 03:29

URI:

https://repository.upnjatim.ac.id/id/eprint/55495