Enhancing Tumor Edge Consistency in Multimodal MRI Synthesis for Improved Glioma Segmentation
IEEE Signal Processing Letters · 2025 · 32 · 2060-2064
DOI: 10.1109/lsp.2025.3562824 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Signal Processing Letters, 2025. Verfasst von Can Chang und 2 weiteren Autor:innen. Die Arbeit wurde bisher 2-mal zitiert (Crossref) und nennt 24 Literaturangaben. Gefördert von: National Natural Science Foundation of China, Key Program of the National Natural Science Foundation of China.
Abstract (Originalsprache)
Precise segmentation of glioma subregions using multimodal MRI is crucial for accurate diagnosis and effective treatment. However, the absence of certain MRI modalities in clinical settings often leads to incomplete information, necessitating cross-modality synthesis to fill the gaps. A significant challenge in this synthesis is the blurring of tumor subregion boundaries, which affects subsequent segmentation accuracy. Existing methods, while improving boundary clarity, fail to ensure consistent depiction across different modalities due to varying contrasts and sensitive areas. To address these issues, we propose CSEC-Net, a novel tumor-aware synthesis model that enhances tumor edge consistency through Specific Contrast extraction and Edge Consistency enhancement. Our model employs a Contrast-Specific Prototype Learning (CS-PL) method to extract contrast-specific prototype features and an Edge Consistency Contrast Learning (ECCL) method to improve tumor edge pixel sampling and feature learning. This innovative approach ensures consistent and clear tumor edge depiction across different modalities, significantly improving multimodal MRI synthesis and tumor segmentation accuracy.
Förderung
- National Natural Science Foundation of China
- Key Program of the National Natural Science Foundation of China
Zitieren
APA
Chang, C., Yao, L., & Zhao, X. (2025). Enhancing Tumor Edge Consistency in Multimodal MRI Synthesis for Improved Glioma Segmentation. IEEE Signal Processing Letters, 32, 2060-2064. https://doi.org/10.1109/lsp.2025.3562824
Vancouver
Chang C, Yao L, Zhao X. Enhancing Tumor Edge Consistency in Multimodal MRI Synthesis for Improved Glioma Segmentation. IEEE Signal Processing Letters. 2025;32:2060-2064. doi:10.1109/lsp.2025.3562824
BibTeX
@article{Chang2025,
author = {Chang, Can and Yao, Li and Zhao, Xiaojie},
title = {Enhancing Tumor Edge Consistency in Multimodal MRI Synthesis for Improved Glioma Segmentation},
journal = {IEEE Signal Processing Letters},
year = {2025},
volume = {32},
pages = {2060--2064},
doi = {10.1109/lsp.2025.3562824}
}
EUVIMED