EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

LightFormer-3D: a lightweight hierarchical CNN-transformer hybrid for 3D medical image segmentation

Qingzheng Hu, Ji Li, Wenqiu Zhu

Frontiers in Computer Science · 09.07.2026 · 8

DOI: 10.3389/fcomp.2026.1837091 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Frontiers in Computer Science, online erschienen am 09.07.2026. Verfasst von Qingzheng Hu und 2 weiteren Autor:innen. Die Arbeit nennt 42 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Introduction

Brain tumor segmentation from 3D MRI is critical for treatment planning, yet deploying accurate Transformer-based models in clinical settings remains challenging due to prohibitive computational costs. State-of-the-art methods often exceed 90M parameters and require seconds per inference, precluding real-time use on resource-constrained hardware.

Methods

We propose LightFormer-3D, a lightweight hierarchical CNN-Transformer hybrid that applies a co-design principle to jointly optimize patch embedding, self-attention, and multi-scale feature fusion, ensuring that efficiency gains compound across all stages. Specifically, a Depthwise Separable Convolution Patch Embedding (DSC-PE) reduces embedding parameters by 70%; a Multi-head Self-Attention with Spatial Sequence Reduction (MSA-SR) compresses self-attention complexity from O(N 2 ) to O(N 2 /r 3 ); and a Lightweight Scalable Feature Fusion (LSFF) module adaptively weights multi-scale features using only five parameters.

Results

Evaluated on the BraTS 2017 and BraTS 2021 benchmarks, LightFormer-3D achieves a mean Dice score of 87.0% on BraTS 2021 with 1.78M parameters and 9.3 GFLOPs, surpassing UNETR (80.2% Dice, 92.5M parameters) by 6.8 pp and the best competing lightweight method, U-Net_ASPP_EVO (86.0%), by 1.0 pp, while delivering 10× faster inference than UNETR (0.51s vs. 5.41s per case on an NVIDIA RTX 4060). Five-fold cross-validation (86.7% ± 0.2%) confirms stable generalization. Ablation studies validate each module's contribution, and zero-shot external validation on the independent MU-Glioma-Post dataset (203 post-treatment cases) confirms robust whole-tumor localization (WT Dice 85.0%) under severe domain shift.

Discussion

LightFormer-3D establishes that co-designed lightweight modules can simultaneously improve accuracy and efficiency, achieving a new Pareto-optimal point for brain tumor segmentation.

Zitieren

APA

Hu, Q., Li, J., & Zhu, W. (2026). LightFormer-3D: a lightweight hierarchical CNN-transformer hybrid for 3D medical image segmentation. Frontiers in Computer Science, 8. https://doi.org/10.3389/fcomp.2026.1837091

Vancouver

Hu Q, Li J, Zhu W. LightFormer-3D: a lightweight hierarchical CNN-transformer hybrid for 3D medical image segmentation. Frontiers in Computer Science. 2026;8. doi:10.3389/fcomp.2026.1837091

BibTeX
@article{Hu2026,
  author = {Hu, Qingzheng and Li, Ji and Zhu, Wenqiu},
  title = {LightFormer-3D: a lightweight hierarchical CNN-transformer hybrid for 3D medical image segmentation},
  journal = {Frontiers in Computer Science},
  year = {2026},
  volume = {8},
  doi = {10.3389/fcomp.2026.1837091}
}
Kostenlos starten →