RADIANT-PET improves PET/CT lesion segmentation accuracy by layering LLM-based adjudication and RL optimization on top of a high-sensitivity voxel segmentation stage, with largest gains when radiology reports are available.
Title resolution pending
3 Pith papers cite this work. Polarity classification is still indexing.
3
Pith papers citing it
fields
cs.CV 3years
2026 3verdicts
UNVERDICTED 3representative citing papers
The autoPET3 challenge finds that leading AI models reach a mean Dice score of 0.66 for multitracer PET/CT lesion segmentation, with compositional generalization to unseen tracer-center pairs remaining an open problem driven by volume overestimation and case heterogeneity.
An uncertainty-aware framework with Bayesian ensembling and epistemic uncertainty-augmented training improves lesion segmentation robustness on public multi-tracer PET/CT datasets over standard nnU-Net.
citing papers explorer
-
RADIANT-PET: Reasoning-Augmented PET/CT Lesion Segmentation with Large Language Models and Reinforcement Learning
RADIANT-PET improves PET/CT lesion segmentation accuracy by layering LLM-based adjudication and RL optimization on top of a high-sensitivity voxel segmentation stage, with largest gains when radiology reports are available.
-
The autoPET3 Challenge: Automated Lesion Segmentation in Whole-Body PET/CT $\unicode{x2013}$ Multitracer Multicenter Generalization
The autoPET3 challenge finds that leading AI models reach a mean Dice score of 0.66 for multitracer PET/CT lesion segmentation, with compositional generalization to unseen tracer-center pairs remaining an open problem driven by volume overestimation and case heterogeneity.
-
Improving PET/CT-Based Whole-Body Lesion Segmentation Using Prediction Uncertainty-Augmented Models
An uncertainty-aware framework with Bayesian ensembling and epistemic uncertainty-augmented training improves lesion segmentation robustness on public multi-tracer PET/CT datasets over standard nnU-Net.