CMRA-DETR: A Lightweight and High-Accuracy Detection Framework for MRI-Based Brain Tumor Identification - Takeaways - MDSpire

CMRA-DETR: A Lightweight and High-Accuracy Detection Framework for MRI-Based Brain Tumor Identification

  • By

  • Weng, Cai

  • Huang, Bowei

  • Chen, Jinghui

  • Hu, Wei

  • Huang, Zhiqing

  • Weng, Punan

  • Zhao, Hongjia

  • Zheng, Minqin

  • April 21, 2026

  • 0 min

Share

  • 1

    CMRA-DETR is a novel framework designed for accurate brain tumor detection in MRI scans, addressing challenges in low-contrast imaging.

  • 2

    The framework incorporates a CSP-MambaOut backbone to enhance local texture recognition of tumor margins through gated feature selection.

  • 3

    An AIFI-MALA module is included to improve attention mechanisms, addressing distributional smoothing issues in conventional linear attention.

  • 4

    CMRA-DETR achieves high performance metrics, including P = 95.5% and mAP@50 = 97.9%, while reducing parameters and GFLOPs significantly.

  • 5

    The model demonstrates strong cross-dataset generalization, achieving mAP@50 = 96.6% on an external dataset without fine-tuning.

Original Source(s)

CMRA-DETR: A Lightweight and High-Accuracy Detection Framework for MRI-Based Brain Tumor Identification

  • by Weng, Cai , Huang, Bowei , Chen, Jinghui , Hu, Wei , Huang, Zhiqing , Weng, Punan , Zhao, Hongjia , Zheng, Minqin

  • April 21, 2026

Related Content

The role of NAD+ metabolic reprogramming in colorectal cancer chemoresistance: mechanistic insights, clinical translation challenges and opportunities

Advanced Brain Tumor Care, Close to Home

  • by

  • May 28, 2026

  • 2 min
Synthetic short RNA in cancer