Augmentasi Morfologi Sperma Menggunakan Wasserstein Generative Adversarial Network (WGAN) Dan Optimalisasi Citra Menggunakan Fast Super Resolution Convolutional Neural Network (FSRCNN)

Kuswardhani, Hajjar Ayu Cahyani and Halim, Christina (2024) Augmentasi Morfologi Sperma Menggunakan Wasserstein Generative Adversarial Network (WGAN) Dan Optimalisasi Citra Menggunakan Fast Super Resolution Convolutional Neural Network (FSRCNN). Project Report (Praktek Kerja Lapang dan Magang). UPN Veteran Jawa Timur.

	Text (Cover) cover-full.pdf Download (448kB)
	Text (BAB I) bab1.pdf Download (82kB)
	Text (BAB II) bab2.pdf Restricted to Repository staff only until 21 October 2027. Download (89kB)
	Text (BAB III) bab3.pdf Restricted to Repository staff only until 21 October 2027. Download (1MB)
	Text (BAB IV) bab4.pdf Restricted to Repository staff only until 21 October 2027. Download (9kB)
	Text (Daftar Pustaka) daftar pustaka.pdf Download (140kB)
	Text (Lampiran) lampiran.pdf Restricted to Repository staff only until 21 October 2027. Download (387kB)

Abstract

PT. IGS Indonesia is a technology company specializing in higher education and IT consultancy. In collaboration with Rumah Sakit Umum Daerah Dr. Mohamad Saleh Probolinggo, they are developing Medical AI: Sperm Quants to address the inefficiencies of manual sperm analysis, which is time-consuming and limited by available human resources. This research focuses on analyzing sperm quality, particularly morphology, with the goal of developing an automated tool that can improve efficiency in terms of time and labor, and assist couples facing fertility issues. The primary challenge encountered is the lack of data to build an accurate model, thus this research adopts the Wasserstein Generative Adversarial Networks (WGAN) algorithm to increase the sperm dataset. The research methodology begins with the collection of primary data from RSUD Probolinggo, which is then processed using WGAN to generate more varied and realistic images. However, the images often have poor quality, so the Fast Super-Resolution Convolutional Neural Network (FSRCNN) algorithm is used for image optimization by enhancing resolution and reducing noise. The combination of WGAN augmentation techniques and FSRCNN image optimization is expected to produce more varied and high-quality sperm morphology images, improve the accuracy of the sperm analysis model, and make a significant contribution to infertility diagnosis and research.

Item Type:

Monograph (Project Report (Praktek Kerja Lapang dan Magang))

Contributors: