PSCT-Net: Geometry-Aware Pediatric Skull CT Reconstruction via Differentiable Back-Projection and Attention-Guided Refinement

2026-06-18

AI Summary

This study addresses the severely ill-posed problem of reconstructing pediatric skull CT volumes from sparse biplanar X-rays, a task hindered by existing methods’ lack of explicit geometric modeling, which compromises 3D structural fidelity and bone boundary accuracy. To overcome these limitations, the authors propose a geometry-aware 3D reconstruction framework that leverages differentiable back-projection to establish spatially consistent voxel priors, thereby mitigating depth ambiguity. An attention-guided projection module (AGP-3D) is introduced to establish precise nonlinear 2D–3D correspondences, while a bidirectional Mamba module (BiM-3D) efficiently captures long-range voxel dependencies. The work also introduces PedSkull-CT, the first pediatric skull CT dataset. Experimental results demonstrate that the proposed method substantially improves geometric accuracy and bone boundary sharpness in reconstructed volumes.

Abstract

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.

Problem

• pediatric skull CT
• sparse-view reconstruction
• geometry-aware
• depth ambiguity
• low-dose imaging

Innovation

• differentiable back-projection
• geometry-aware reconstruction
• attention-guided refinement
• Bidirectional Mamba
• pediatric skull CT