Article Per Year (5 Year)
p-Index From 2020 - 2025
0.23
P-Index
This Author published in this journals
All Journal Jurnal Teknimedia: Teknologi Informasi dan Multimedia
Banna, Dion Al
Unknown Affiliation
Computer Science & IT Control & Systems Engineering Engineering

Published : 1 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 1 Documents
Search

 1 
PEMANFAATAN SAM DAN YOLOV8 UNTUK DETEKSI DAN SEGMENTATION MRI TUMOR OTAK Ardiansyah, Ardiansyah; Qodri, Krisna Nuresa; Banna, Dion Al; Al-Baihaqi, Muhammad Zulfikhar
TEKNIMEDIA: Teknologi Informasi dan Multimedia Vol. 5 No. 1 (2024): Juni 2024
Publisher : Badan Penelitian dan Pengabdian Masyarakat (BP2M) STMIK Syaikh Zainuddin NW Anjani

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.46764/teknimedia.v5i1.192

Abstract

The development of artificial intelligence (AI) is specific to the field of Computer Vision (CV) to obtain information based on data contained in visual media. AI in the healthcare field such as image recognition and Deep Learning (DL) is a discussion that is often used as an object of research and development. The health sector Limitation is the emergence of AI utilization in the health sector, which encourages DL research. Segmentation Anything Model (SAM) and YOLOv8 are new algorithms introduced. Thus, this research aims to measure the utilization of SAM and YOLOv8 for making the detection and segmentation of Brain Tumor MRI data. Before the training process, researchers first compared roboflow segmentation and the SAM model. The dataset was labeled with a Bounding Box by experts. The dataset contains 455 gliomas, 550 meningiomas, and 620 pituitaries. The research concluded that the utilization of SAM greatly simplified the annotation process. The segmentation YOLOv8 obtained Box accuracy results for all classes of 86% precision, 87% Recall, 89% mAP50, and 71% mAP 50-95. The mask performance evaluation gets the results of 86% precision, 87% Recall, 89% mAP50, and 70% mAP50-95. The research obtained the YOLOv8n-seg model to get excellent results even though it is a tiny model of YOLOv8. This study found the glioma tumor class to be the class with the lowest results because the dataset used was not much. The researcher encourages other researchers to use data augmentation to increase the use of datasets for each class to provide better results.
Co-Authors Al-Baihaqi, Muhammad Zulfikhar Ardiansyah Ardiansyah Krisna Nuresa Qodri