Search papers, labs, and topics across Lattice.
This paper introduces PSCT-Net, a geometry-aware framework for reconstructing pediatric skull CT images from sparse bi-planar X-rays, significantly reducing radiation exposure. By utilizing differentiable back-projection to create a spatially faithful volumetric prior and an Attention-Guided Projection module to establish voxel-wise correspondences, the method effectively mitigates depth ambiguity and enhances osseous boundary definition. The evaluation on a newly curated dataset, PedSkull-CT, demonstrates substantial improvements over existing geometry-agnostic techniques, making it a promising approach for pediatric imaging.
A geometry-aware reconstruction method that significantly improves pediatric skull CT imaging while minimizing radiation exposure.
Computed Tomography (CT) is essential for diagnosing pediatric craniofacial abnormalities, yet poses radiation risks to developing anatomies. Reconstructing 3D CT from sparse bi-planar X-rays offers a low-dose alternative but is severely ill-posed. Existing methods employ geometry-agnostic feature lifting, naively projecting 2D features into 3D without explicit spatial modeling, causing depth ambiguity and degraded osseous boundaries. We present PSCT-Net, a geometry-aware framework with differentiable back-projection. Differentiable back-projection establishes a spatially faithful volumetric prior, alleviating depth ambiguity. An Attention-Guided Projection (AGP-3D) module then learns non-linear voxel-wise correspondences between 2D regions and 3D locations. A Bidirectional Mamba (BiM-3D) module captures long-range volumetric dependencies with linear complexity. We further curate a private institutional pediatric skull CT cohort, PedSkull-CT, comprising normal and pathological cases for internal evaluation, addressing the gap in adult-centric, trunk-focused datasets.
