Szczegóły publikacji
Opis bibliograficzny
Lightweight Grey Wolf Optimization-driven hyperparameter tuning of U-Net for robust brain tumor segmentation / Shoffan SAIFULLAH, Rafał DREŻEWSKI, Anton Yudhana // W: Computational Science – ICCS 2026 workshops : 26th International Conference : Hamburg, Germany, June 29–July 1, 2026 : proceedings , Pt. 1 / eds. Maciej Paszyński, Amanda S. Barnard, Yongjie Jessica Zhang. — Cham : Springer, cop. 2026. — ( Lecture Notes in Computer Science ; ISSN 0302-9743 ; LNCS 16786 ). — ISBN: 978-3-032-29911-6; e-ISBN: 978-3-032-29912-3. — S. 485–493. — Bibliogr., Abstr. — Publikacja dostępna online od: 2026-06-26. — S. Saifullah - dod. afiliacja: Universitas Pembangunan Nasional Veteran Yogyakarta, Indonesia
Autorzy (3)
- AGHSaifullah Shoffan
- AGHDreżewski Rafał
- Yudhana Anton
Słowa kluczowe
Dane bibliometryczne
| ID BaDAP | 168906 |
|---|---|
| Data dodania do BaDAP | 2026-07-15 |
| DOI | 10.1007/978-3-032-29912-3_38 |
| Rok publikacji | 2026 |
| Typ publikacji | materiały konferencyjne (aut.) |
| Otwarty dostęp |  |
| Wydawca | Springer |
| Konferencja | International Conference on Computational Science 2026 |
| Czasopismo/seria | Lecture Notes in Computer Science |
Abstract
Accurate brain tumor segmentation from magnetic resonance imaging (MRI) is essential for clinical diagnosis and treatment planning, yet convolutional neural network performance is highly sensitive to hyperparameter selection. This paper proposes a lightweight Grey Wolf Optimization (GWO)–driven framework for automated hyperparameter tuning of a parameterized U-Net. The approach jointly optimizes architectural and training parameters under a constrained evaluation budget using reduced-resolution training, enabling efficient search without increasing model complexity. Experimental results on the Figshare Brain Tumor Segmentation dataset demonstrate strong performance, achieving Dice scores of up to 0.9820 under five-fold cross-validation. The optimized model generalizes well to BraTS 2021, reaching whole-tumor Dice scores up to 0.9778. These results demonstrate that lightweight and interpretable metaheuristic optimization can effectively improve segmentation performance while maintaining computational efficiency.
