search

menu

  • Research Research
    • Where science meets inspired minds

    • Back
    • Research
    • Our Science
    • Research Groups
    • Facilities & Platforms
    • Clinical research
    • Find a researcher
    • Publications
    • Knowledge Transfer
  • Careers & study Careers & study
    • Become a leader in cancer research

    • Back
    • Careers & study
    • Vacancies
    • Faculty
    • Scientific staff
    • Scientific support staff
    • Postdoctoral fellows
    • PhD Students
    • Operational staff
    • Clinical fellows
    • Life in Amsterdam
    • Student internships
  • News & Events News & Events
    • Check out our stories and events

    • Back
    • News & Events
    • News
    • Media & Press
    • Calendar
  • About us About us
    • Maximum impact for cancer patients

    • Back
    • About us
    • Our vision
    • Organization
    • Collaborations
    • Responsible Research
    • Support us
    • Visit us
    • Contact us
  • Support us
Support us
  • Home
  • Publications
  • Research
  • Publications
  • Article

Leveraging modality-guided pre-training for dual-prompt-driven multi-cancer PET-CT segmentation.

Xinglong Liang ,
Jiaju Huang ,
Tianyu Zhang ,
Luyi Han ,
Xin Wang ,
Yuan Gao ,
Chunyao Lu ,
Yue Sun ,
Jonas Teuwen ,
Tao Tan ,
Ritse Mann

Abstract

PET-CT lesion segmentation remains challenging due to heterogeneous lesion appearance, small and dispersed lesions, physiological FDG uptake, and limited annotations. Existing self-supervised methods are mostly designed for unimodal imaging and therefore fail to fully exploit the complementary anatomical and metabolic information in PET-CT. Meanwhile, conventional multi-cancer segmentation strategies often treat different cancer types as a unified task, which weakens cancer-specific features, and existing prompt-based methods still have limited task adaptation and sensitivity to small lesions. To address these limitations, a unified two-stage framework for multi-cancer PET-CT segmentation is presented. First, a modality-guided probabilistic masked autoencoder is introduced to enhance cross-modal PET-CT representation learning through modality-specific masking. Second, a dual-prompt downstream segmentation network is designed to model both cancer-specific characteristics and cross-cancer shared knowledge, with prompt-aware heads further improving task adaptation and small-lesion delineation. Experiments on a multi-cancer PET-CT dataset show consistent improvements over the best-performing non-prompt and prompt-based baselines, with average Dice gains of 2.51% and 2.18%, respectively. The framework is further applied to an unannotated breast cancer cohort for survival analysis, demonstrating promising generalizability and improved risk stratification. The code is available at: https://github.com/XinglongLiang08/DpDNet.

More about this publication

Medical image analysis

Volume 113
Pages 104182
Publication date 20-06-2026

Full text links

Publisher website (DOI) 10.1016/j.media.2026.104182
Europe PubMed Central 42341389
Pubmed 42341389

Where science meets inspired minds

Contact

Plesmanlaan 121
1066CX Amsterdam

020 512 9111 communicatie@nki.nl

Quick links

  • Vacancies
  • News
  • Contact us
  • Media & Press

Follow us on

Disclaimer
Privacy statement
Cookies
Change cookie settings

This site uses cookies

This website uses cookies to ensure you get the best experience on our website.