DPEA-Net: a clinically-oriented lightweight 3D CNN for glioma segmentation in multiparametric MRI - Takeaways - MDSpire

DPEA-Net: a clinically-oriented lightweight 3D CNN for glioma segmentation in multiparametric MRI

  • By

  • Caijian Hua

  • Xuerong Jing

  • Liuying Li

  • Xia Zhou

  • May 25, 2026

  • 0 min

Share

  • 1

    DPEA-Net is a lightweight 3D convolutional neural network designed for accurate glioma subregion segmentation from multiparametric MRI.

  • 2

    The model features a Dynamic Hierarchically Decoupled Convolution unit that reduces parameters by 99% compared to 3D U-Net.

  • 3

    DPEA-Net achieves mean Dice scores of 90.43% for whole tumor, 85.56% for tumor core, and 81.89% for enhancing tumor on validation sets.

  • 4

    The Cross-Dimensional Region-Specific Enhancement Attention module improves the modeling of 3D spatial relationships for better boundary refinement.

  • 5

    DPEA-Net enables sub-2-second inference on standard clinical hardware, supporting efficient integration into neuro-oncological workflows.

Original Source(s)

DPEA-Net: a clinically-oriented lightweight 3D CNN for glioma segmentation in multiparametric MRI

  • by Caijian Hua, Xuerong Jing, Liuying Li, Xia Zhou

  • May 25, 2026

Related Content

VANGL1 links angiogenesis–stemness programs and tumor microenvironment remodeling: a pan-cancer, multi-omics study with translational validation

From molecular networks to translational intervention: current progress in the mechanisms of gemcitabine resistance in bladder cancer

Novel approaches for separating graft-versus-leukemia effects from graft-versus-host disease