Szczegóły publikacji
Opis bibliograficzny
Automatic brain tumor segmentation: advancing U-Net with ResNet50 encoder for precise medical image analysis / Shoffan SAIFULLAH, Rafał DREŻEWSKI, Anton Yudhana, Andiko Putro Suryotomo // IEEE Access [Dokument elektroniczny]. — Czasopismo elektroniczne ; ISSN 2169-3536. — 2025 — vol. 13, s. 43473–43489. — Wymagania systemowe: Adobe Reader. — Bibliogr. s. 43487–43489, Abstr. — Publikacja dostępna online od: 2025-03-03. — S. Saifullah - dod. afiliacja: Department of Informatics, Universitas Pembangunan Nasional Veteran Yogyakarta, Indonesia
Autorzy (4)
- AGHSaifullah Shoffan
- AGHDreżewski Rafał
- Yudhana Anton
- Suryotomo Andiko Putro
Słowa kluczowe
Dane bibliometryczne
| ID BaDAP | 159139 |
|---|---|
| Data dodania do BaDAP | 2025-03-29 |
| Tekst źródłowy | URL |
| DOI | 10.1109/ACCESS.2025.3547430 |
| Rok publikacji | 2025 |
| Typ publikacji | artykuł w czasopiśmie |
| Otwarty dostęp |  |
| Creative Commons | |
| Czasopismo/seria | IEEE Access |
Abstract
Automated brain tumor segmentation from MRI images is critical for accurate diagnosis and treatment planning. This study presents a novel ResUNet50-based approach, integrating ResNet50 as an encoder within the U-Net framework to achieve robust and precise segmentation. The proposed model was evaluated on two datasets: a Kaggle-based T1-CE MRI dataset and BraTS 2018, ensuring comprehensive assessment across different imaging conditions. ResUNet50 outperformed state-of-the-art models, achieving Dice coefficients exceeding 0.95 and Jaccard indices above 0.91 on the Kaggle dataset. Additionally, experiments on BraTS 2018 Whole Tumor segmentation across multiple MRI modalities (FLAIR, T1, T1-CE, and T2) demonstrated high accuracy on both High-Grade Gliomas (HGG) and Low-Grade Gliomas (LGG), confirming model generalization. Statistical significance tests (Paired t-tests and Wilcoxon Signed-Rank Tests, p < 0.05) validated the improvements over existing approaches. Furthermore, ResUNet50 reduced parameters by over 60% and accelerated inference time by 4.8× compared to U-Net, enhancing its potential for clinical deployment. Future work will focus on ensemble learning and bio-inspired optimization for improved robustness across multi-center MRI datasets. Explainable AI techniques such as Grad-CAM and saliency maps will be incorporated to enhance interpretability, improving clinical applicability.
