Search papers, labs, and topics across Lattice.
This paper introduces the Trajectory-guided Uncertainty-aware Network (TrailNet), which enhances weakly supervised medical image segmentation by integrating gaze trajectory data to model both spatial semantics and temporal context. By addressing the challenges posed by noise in gaze data and exploratory fixations, TrailNet employs a spatio-temporal encoder and a multi-scale uncertainty decoder to improve segmentation accuracy. Experimental results show that TrailNet significantly outperforms existing methods, achieving Dice scores of 81.25% and 81.85% on two public datasets, highlighting its effectiveness in clinical applications.
Gaze trajectory can transform medical image segmentation, achieving over 81% accuracy without requiring pixel-level annotations.
Medical image segmentation heavily depends on labor-intensive and time-consuming pixel-level annotations. Eye tracking offers a cost-effective solution that can be naturally integrated into clinical workflows. Recorded by eye trackers, gaze conveys the spatial regions of clinicians'attention through fixations and the temporal context of clinicians'progressive visual perception from trajectories. Nevertheless, effective modeling of temporal trajectories remains challenging, and noise in gaze caused by exploratory fixations greatly limits segmentation performance. To overcome these limitations, we propose the Trajectory-guided Uncertainty-aware Network (TrailNet), which exploits gaze-supervised medical image segmentation from spatial semantics modeling to temporal context by jointly leveraging fixations and trajectories. Specifically, the proposed trajectory-guided spatio-temporal encoder models temporal context and establishes complementary interactions with image spatial semantics to strengthen target perception. Furthermore, the multi-scale uncertainty decoder leverages category mutual-exclusivity constraints to produce deterministic predictions and mitigate supervision uncertainty induced by noise. To enable gaze-free inference, we further introduce a cycle distillation strategy that transfers feature-level knowledge via teacher-student networks. Experimental results on two public datasets demonstrate that TrailNet outperforms state-of-the-art methods, achieving Dice scores of 81.25% and 81.85%, respectively.
