Synapse
⌘+K
Synapse
PulseExploreClubsResearchersJournals
Instagram
HomeClubsExplore
May 18, 2026Scientific ReportsOpen Access

Adaptive patch sampling and location-aware reasoning for whole body PET-CT multi-organ segmentation

View Full Paper
Ask AI
Bookmark
Share

Authors

JPJunha ParkACArthur ChoHPHae‐Jeong Park

Discussion

Loading...

Member takes

Overview

Randomized trial reveals improved segmentation performance in PET-CT imaging, indicating dynamic sampling's effectiveness.

Key Points

  • This research aims to enhance multi-organ segmentation in PET-CT imaging through adaptive sampling techniques.
  • Developed adaptive patch sampling (APS) for dynamic computation allocation based on voxel-wise uncertainty and prediction error.
  • Introduced a patch encoding (PE) block for inferring location information and modulating features using attention mechanisms.
  • Evaluated performance improvement through experiments on multi-organ PET-CT segmentation and external validations on the Synapse dataset.
  • Achieved faster convergence in training compared to existing methods.
  • Demonstrated consistent performance gains in multi-organ segmentation, confirmed by external validation on the Synapse dataset.
  • Mechanistic analyses validated the effectiveness of APS and the PE block through attention analysis.

Cite This Study

Park et al. (2026) studied this question.

synapsesocial.com/papers/6a0aac2b5ba8ef6d83b6fb92https://doi.org/10.1038/s41598-026-51023-x
View Full Paper
Ask AI
Bookmark
Share