Search papers, labs, and topics across Lattice.
This paper introduces a hybrid registration framework for intraoperative liver registration that combines biomechanical modeling with data-driven learning to address the challenges of soft-tissue deformation and sparse measurements. By learning a residual deformation function through a graph neural diffusion approach, the method corrects linear biomechanical predictions using context samples derived from sparse intraoperative correspondences. Experimental results on a deformable liver phantom dataset show significant improvements in registration accuracy and generalization, particularly for out-of-distribution scenarios, compared to existing methods.
Learning a residual deformation function from sparse intraoperative measurements leads to superior liver registration accuracy, especially in challenging geometries.
Accurate intraoperative liver registration is challenging due to substantial soft-tissue deformation yet sparse intraoperative measurements. Biomechanical models regularize this ill-posedness with prior knowledge but exhibit persistent prediction bias due to simplifying assumptions, while data-driven learning solutions struggle with data efficiency, generalization, and physical plausibility. We propose a hybrid registration framework that adapts a biomechanical prior using sparse intraoperative correspondences. Rather than learning a full deformation field, we learn a residual deformation function that corrects linear biomechanical predictions, modeled as a graph neural diffusion function with geometry-aware attention over the 3D liver mesh. To enable long-range information transfer of sparse observations, we take a novel perspective of sparse intraoperative measurements as \textit{context} samples where input-output pairs of the residual deformation function are fully observed, casting the problem into learning-to-learn this residual function from intraoperative context samples with feedforward meta-learners. Experiments on a deformable liver phantom dataset demonstrate improved registration accuracy and generalization compared to rigid, biomechanical, and data-driven baselines, particularly for out-of-distribution geometries and deformations.
