Detection of Motility and Morphological Abnormalities of Sperm Using 3D-Convolutional Neural Network

Farid, Muhammad Nashif (2026) Detection of Motility and Morphological Abnormalities of Sperm Using 3D-Convolutional Neural Network. Undergraduate thesis, UPN "Veteran" Jawa Timur.

	Text (COVER) abstrack-notas_merged.pdf Download (5MB)
	Text (BAB I) Bab 1.pdf Download (273kB)
	Text (BAB II) Bab 2.pdf Restricted to Repository staff only until 8 March 2029. Download (1MB)
	Text (BAB III) Bab 3.pdf Restricted to Repository staff only until 8 March 2029. Download (225kB)
	Text (BAB IV) Bab 4.pdf Restricted to Repository staff only until 8 March 2029. Download (1MB)
	Text (BAB V) Bab 5.pdf Download (16kB)
	Text (DAFTAR PUSTAKA) Daftar Pustaka.pdf Download (151kB)
	Text (LAMPIRAN) Lampiran.pdf Restricted to Repository staff only Download (265kB)

Abstract

Infertility is a significant global reproductive health issue, with approximately one in six couples of reproductive age experiencing difficulty in achieving pregnancy. The evaluation of sperm quality is a crucial component in the diagnosis of male infertility. However, conventional manual sperm analysis through microscopic observation still presents limitations in terms of objectivity, consistency, and efficiency, as it heavily depends on the precision and experience of laboratory personnel. This study aims to develop an automated sperm health analysis system based on Computer-Aided Sperm Analysis (CASA) by integrating deep learning models for motility and morphology classification. The dataset consists of microscopic sperm videos acquired using an OptiLab IRIS-5 binocular microscope. The research process begins with video standardization using FFmpeg, followed by image quality enhancement through contrast stretching techniques. For motility analysis, a 3D-Convolutional Neural Network (3D-CNN) architecture is implemented to extract both spatial and temporal information from sperm movement patterns. Meanwhile, morphology analysis is performed using the EfficientNet-V2 model through a transfer learning approach, after sperm objects are isolated using morphological erosion operations. Experimental results demonstrate that the 3D-CNN model achieves an accuracy of 79% in classifying sperm motility, while the EfficientNet-V2 model attains an accuracy of 97% in detecting morphological abnormalities. Both models are integrated into a web-based interactive dashboard developed using Streamlit, which presents motility statistics, the percentage of normal and abnormal morphology, and clinical indications based on WHO standards. The findings indicate that the integration of 3D-CNN and EfficientNet-V2 provides a more objective, consistent, and efficient sperm analysis system, demonstrating its potential as a reliable diagnostic support tool in clinical laboratories.

Item Type:

Thesis (Undergraduate)

Contributors:

Contribution	Contributors	NIDN/NIDK	Email
Thesis advisor	Diyasa, I Gede Susrama Mas	19700619 2021211 009	igsusrama.fasilkom@upnjatim.ac.id
Thesis advisor	Pratama, Alfan Rizaldy	199906062024061001	alfan.fasikom@upnjatim.ac.id

Subjects:

Q Science > QA Mathematics > QA76.6 Computer Programming

Divisions:

Faculty of Computer Science > Departemen of Data Science

Depositing User:

Muhammad Nashif Farid

Date Deposited:

09 Mar 2026 07:33

Last Modified:

09 Mar 2026 07:33

URI:

https://repository.upnjatim.ac.id/id/eprint/50280