Exploiting Longitudinal Context in Clinician-Verified Interactive Lesion Tracking

2026-05-21

AI Summary

This work addresses the challenge of balancing automation and clinical controllability in tumor tracking, particularly in ambiguous lesion segmentation. The authors propose a clinically validated interactive tracking paradigm that unifies registration-derived spatial cues with baseline appearance within a single framework and introduces a temporal difference weighting mechanism to enable longitudinal information-guided precise segmentation. Key contributions include the first clinically intervenable tracking mechanism, the establishment of PanTrack—the first longitudinal pancreatic cancer tracking benchmark supporting out-of-distribution generalization evaluation—and the use of large-scale synthetic data pretraining to mitigate real-data scarcity. The method achieved top performance in the MICCAI autoPET IV Challenge, surpassing training-from-scratch approaches by 4.5 Dice points and significantly outperforming existing methods under both fully automatic and expert-verified tracking settings.

Abstract

Tracking tumor lesions across serial CT scans is essential for oncological response assessment. Existing automated methods face a fundamental trade-off: end-to-end trackers achieve high automation but offer no opportunity to correct silent tracking failures, while decoupled registration-segmentation pipelines permit user verification yet discard the lesion's prior appearance, limiting accuracy in ambiguous cases. In this work, we propose a Verified Tracking paradigm: a clinician verifies a registration-proposed prompt, which the model leverages alongside the baseline lesion appearance to resolve segmentation ambiguities. We present a unified framework combining early spatial prompt fusion with latent temporal difference weighting for longitudinally-informed segmentation. To address data scarcity, we leverage large-scale synthetic pretraining, proving essential for exploiting longitudinal context, improving performance by up to 4.5 Dice points over training from scratch. Our approach secured first place in the MICCAI autoPET IV challenge. We further curate and release PanTrack, a new longitudinal pancreatic cancer benchmark, to assess out-of-distribution generalization. Experiments show that our model outperforms prior work in both fully automatic and the proposed verified tracking setting offering a clinically safe middle ground between automation and control. Code, model and dataset will be released at https://github.com/MIC-DKFZ/LongiSeg

Problem

• lesion tracking
• longitudinal context
• tumor segmentation
• clinician verification
• response assessment

Innovation

• Verified Tracking
• Longitudinal Segmentation
• Synthetic Pretraining
• Temporal Difference Weighting
• Interactive Lesion Tracking