EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

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

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

Frontiers in Digital Health · 14.05.2026 · 8

DOI: 10.3389/fdgth.2026.1797571 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Frontiers in Digital Health, online erschienen am 14.05.2026. Verfasst von Claudia Delprete und 5 weiteren Autor:innen. Beteiligte Einrichtung: Department of Electrical and Information Engineering, Polytechnic University of Bari, Bari, Italy. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 40 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Introduction

Colorectal cancer originates in most cases from polyps that progressively become malignant over time and colonoscopy offers a highly studied early diagnostic strategy to ensure a timely treatment plan. Automatic image segmentation of polyps, using intelligent supervised approaches, achieved good performance, but the need of their large annotated datasets limits the clinical applicability.

Methods

This study presents MAPSeg (Memory-Augmented Polyp Segmentation), a fully self-supervised and annotation-free framework for colorectal polyp segmentation, trained exclusively on images of healthy mucosa within an anomaly detection paradigm. The key novelty of MAPSeg mainly lies in SIMPO (Simulation of Polyps), a synthetic augmentation strategy that generates realistic polyp shapes and textures in a colon-specific context, combined with a memory-augmented encoder that models structural priors of normal tissue.

Results

Extensive experiments demonstrate that MAPSeg outperforms the strongest unsupervised methods by approximately 23% in Intersection over Union and 12% in DICE score on the Hyper-Kvasir dataset and consistently maintains this performance margin across multiple out-of-distribution benchmarks, indicating strong generalization capability.

Discussion

The results highlight MAPSeg supported by SIMPO as a viable solution for unsupervised colorectal polyp segmentation, significantly reducing the dependency on manual annotations while maintaining high segmentation accuracy.

Schlagwörter

Stichwörter der Autor:innen

anomaly detectionmedical image analysismemory-augmented networkspolyp segmentationsynthetic data generationunsupervised learning

Autor:innen und Institutionen

  • Department of Electrical and Information Engineering, Polytechnic University of Bari, Bari, Italy.

Zitieren

APA

Delprete, C., Buongiorno, D., Scardigno, R. M., Sibilano, E., Brunetti, A., & Bevilacqua, V. (2026). MAPSeg: self-supervised colorectal polyp segmentation via memory-augmented framework and synthetic polyp simulation. Frontiers in Digital Health, 8. https://doi.org/10.3389/fdgth.2026.1797571

Vancouver

Delprete C, Buongiorno D, Scardigno RM, Sibilano E, Brunetti A, Bevilacqua V. MAPSeg: self-supervised colorectal polyp segmentation via memory-augmented framework and synthetic polyp simulation. Frontiers in Digital Health. 2026;8. doi:10.3389/fdgth.2026.1797571

BibTeX
@article{Delprete2026,
  author = {Delprete, Claudia and Buongiorno, Domenico and Scardigno, Roberto Maria and Sibilano, Elena and Brunetti, Antonio and Bevilacqua, Vitoantonio},
  title = {MAPSeg: self-supervised colorectal polyp segmentation via memory-augmented framework and synthetic polyp simulation},
  journal = {Frontiers in Digital Health},
  year = {2026},
  volume = {8},
  doi = {10.3389/fdgth.2026.1797571}
}
Kostenlos starten →