Development of a Modified UNet-Based Image Segmentation Architecture for Brain Tumor MRI Segmentation

DOI

10.2991/978-94-6463-208-8_7How to use a DOI?

Keywords

Deep Learning; MRI; Image Segmentation; UNet; UBNet

Abstract

Segmentation of the tumor part on the head MRI image is an important thing that must be done by the radiologist in the patient's diagnosis. Therefore, segmentation must be done accurately because this determines the results of the diagnosis and the determination of the next steps taken by the doctor. Segmentation is currently done manually or automatically with a computer system. Several previous studies have developed a brain MRI image segmentation method for tumors based on deep learning imaging. However, the deep learning architecture developed is composed of complex structures and takes a long time to process. So, this paper discusses the study of the development of a lightweight and accurate image segmentation architecture. We propose a study of changes in the size of the MRI image input from 512 × 512 to 16 × 16 to review its effect on the evaluation using the Dice Coefficient method and visual representation of the image. The smaller the input size, the fewer computational processes will occur so that the processing speed will increase. However, the smaller the input size, the less visible the visual representation of the image is. In addition, a study of modifications to the UNet architecture was also carried out combined with the UBNet classification architecture to compare the performance of the two models. The research was carried out computationally and obtained an average accuracy of more than 95% with a quite different visual appearance.

Copyright

© 2023 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Muhammad Masdar Mahasin
AU  - Agus Naba
AU  - Chomsin Sulistya Widodo
AU  - Yuyun Yueniwati
PY  - 2023
DA  - 2023/06/26
TI  - Development of a Modified UNet-Based Image Segmentation Architecture for Brain Tumor MRI Segmentation
BT  - Proceedings of the International Conference of Medical and Life Science (ICoMELISA 2021)
PB  - Atlantis Press
SP  - 37
EP  - 43
SN  - 2468-5739
UR  - https://doi.org/10.2991/978-94-6463-208-8_7
DO  - 10.2991/978-94-6463-208-8_7
ID  - Mahasin2023
ER  -