MAPSeg: self-supervised colorectal polyp segmentation via memory-augmented framework and synthetic polyp simulation - Takeaways - MDSpire

MAPSeg: self-supervised colorectal polyp segmentation via memory-augmented framework and synthetic polyp simulation

  • By

  • Claudia Delprete

  • Domenico Buongiorno

  • Roberto Maria Scardigno

  • Elena Sibilano

  • Antonio Brunetti

  • Vitoantonio Bevilacqua

  • May 14, 2026

  • 0 min

Share

  • 1

    MAPSeg is a self-supervised framework for colorectal polyp segmentation that eliminates the need for annotated datasets.

  • 2

    The framework utilizes SIMPO, a synthetic augmentation strategy that generates realistic polyp shapes from healthy mucosa images.

  • 3

    MAPSeg outperforms existing unsupervised methods by 23% in Intersection over Union and 12% in DICE score on the Hyper-Kvasir dataset.

  • 4

    The approach demonstrates strong generalization capabilities across various out-of-distribution benchmarks.

  • 5

    MAPSeg significantly reduces reliance on manual annotations while maintaining high accuracy in polyp segmentation.

Original Source(s)

MAPSeg: self-supervised colorectal polyp segmentation via memory-augmented framework and synthetic polyp simulation

  • by Claudia Delprete, Domenico Buongiorno, Roberto Maria Scardigno, Elena Sibilano, Antonio Brunetti, Vitoantonio Bevilacqua

  • May 14, 2026

Related Content

Dye-free versus dye-containing submucosal injection solutions in ESD for colorectal LST: a multi-center retrospective study on efficacy and safety

Comparative efficacy of therapeutic modalities for metastatic uveal melanoma: a systemic review and network meta-analysis

Clinical significance and functional characterization of RRN3 in gastric cancer: insights from pan-cancer analysis and experimental validation