Active View Selection with Perturbed Gaussian Ensemble for Tomographic Reconstruction

2026-03-06

AI Summary

This work addresses the fidelity bottleneck in sparse-view X-ray computed tomography (CT) caused by insufficient projection data by proposing the first active view selection framework tailored for X-ray Gaussian splatting. The method constructs a Gaussian ensemble through a density-guided stochastic perturbation mechanism, effectively modeling geometric blur and attenuation characteristics inherent in X-ray imaging. It sequentially selects the next most informative view based on predicted structural variance. Integrating 3D Gaussian splatting, low-density Gaussian primitive identification, ensemble variance estimation, and sequential decision optimization, the proposed framework significantly suppresses geometric artifacts across arbitrary-trajectory CT benchmarks. Under a unified view selection protocol, it consistently outperforms existing approaches, markedly enhancing progressive reconstruction quality in sparse-view settings.

Technology Category

• Reasoning under Uncertainty: Stochastic Optimization
• Computer Vision: Segmentation
• Machine Learning: Active Learning

Application Category

• Security and Privacy: Data transparency and provenance
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems

Abstract

Sparse-view computed tomography (CT) is critical for reducing radiation exposure to patients. Recent advances in radiative 3D Gaussian Splatting (3DGS) have enabled fast and accurate sparse-view CT reconstruction. Despite these algorithmic advancements, practical reconstruction fidelity remains fundamentally bounded by the quality of the captured data, raising the crucial yet underexplored problem of X-ray active view selection. Existing active view selection methods are primarily designed for natural-light scenes and fail to capture the unique geometric ambiguities and physical attenuation properties inherent in X-ray imaging. In this paper, we present Perturbed Gaussian Ensemble, an active view selection framework that integrates uncertainty modeling with sequential decision-making, tailored for X-ray Gaussian Splatting. Specifically, we identify low-density Gaussian primitives that are likely to be uncertain and apply stochastic density scaling to construct an ensemble of plausible Gaussian density fields. For each candidate projection, we measure the structural variance of the ensemble predictions and select the one with the highest variance as the next best view. Extensive experimental results on arbitrary-trajectory CT benchmarks demonstrate that our density-guided perturbation strategy effectively eliminates geometric artifacts and consistently outperforms existing baselines in progressive tomographic reconstruction under unified view selection protocols.

Problem

• active view selection
• sparse-view CT
• X-ray imaging
• tomographic reconstruction
• geometric ambiguity

Innovation

• active view selection
• Gaussian Splatting
• tomographic reconstruction
• uncertainty modeling
• sparse-view CT