Foundation Model-Enabled Efficient Data Sampling (FEEDS): A label-efficient training strategy for pan-cancer, multi-tracer PET/CT datasets

2026-08-11

AI Summary

This work addresses the performance and generalization bottlenecks in whole-body lesion segmentation caused by scarce annotations, proposing FEEDS—a single-step, efficient training paradigm grounded in vision foundation model embeddings. FEEDS selects informative and diverse samples from large-scale unlabeled PET/CT data in a single pass for expert annotation, eliminating the need for iterative pseudo-labeling or active learning loops, thereby substantially reducing both annotation and computational costs. Using only 30% of the annotated data, the method achieves segmentation performance on par with fully supervised models across multiple datasets—including AutoPET-III, DeepPSMA, and an internal multicenter, multi-tracer cohort—demonstrating significantly superior generalization compared to random sampling and semi-supervised baselines.

Abstract

Automated lesion segmentation in whole-body PET/CT imaging can assist clinicians with cancer detection, staging, and treatment planning across radiotracers and cancer types. However, training lesion segmentation models that capture variations in lesion size, distribution, and appearance requires large annotated datasets, whose creation is both time- and expertise-intensive. As a result, models trained on limited labeled PET/CT data often lack the accuracy and generalizability needed for clinical use. We present FEEDS (Foundation model-Enabled Efficient Data Sampling), a label- and compute-efficient learning strategy that uses vision foundation model embeddings to select the most informative and diverse unlabeled cases for expert annotation. Unlike unsupervised, semi-supervised, and active learning approaches, FEEDS is a one-step training paradigm requiring only a limited, representative training set, making it label- and compute-efficient. We train and validate FEEDS using the AutoPET-III dataset. We test its accuracy and generalizability on three held-out sets: AutoPET-III, DeepPSMA, and an internal Dartmouth-Hitchcock Medical Center dataset. We evaluate clinical utility at the voxel, lesion, and anatomic region level to assess performance in high-risk areas and treatment planning utility. FEEDS outperforms random-sampling-based labeling, pseudolabel-based semi-supervised learning, and training with limited labeled data alone. It generalizes across all three test sets, FDG and PSMA tracers, and multiple diseases, matching fully-labeled (100\%) training performance with 70\% less annotation burden. FEEDS addresses the challenge of label scarcity in an automatic lesion segmentation framework by providing a practical approach for constructing representative and diverse annotation queues from large, unannotated clinical repositories.

Problem

• lesion segmentation
• label scarcity
• PET/CT imaging
• pan-cancer
• multi-tracer

Innovation

• Foundation Model
• Efficient Data Sampling
• Label-Efficient Learning
• PET/CT Segmentation
• Pan-Cancer