Revisiting $Psi$DONet: microlocally inspired filters for incomplete-data tomographic reconstructions

2025-01-30

AI Summary

To address poor image quality and severe streak artifacts in sparse-angle and limited-angle CT reconstruction, this paper proposes ΨDONet-Lite, a lightweight unrolled deep network grounded in microlocal analysis. The key innovation lies in the first explicit modeling of the microlocal singular structure of streak artifacts and its incorporation into learnable filter design, enabling singularity-support-guided customized convolution. Compared to the original ΨDONet, ΨDONet-Lite achieves substantial parameter reduction while maintaining or slightly improving reconstruction fidelity on limited-angle tasks. Crucially, it is the first method validated for sparse-angle CT reconstruction, demonstrating both effectiveness and generalizability in this challenging regime. By bridging rigorous microlocal theory with practical deep learning architecture, ΨDONet-Lite establishes a new paradigm for low-dose CT reconstruction that balances theoretical soundness with engineering feasibility.

Technology Category

• Computer Vision: Learning & Optimization for CV
• Machine Learning: Learning on the Edge & Model Compression
• Reasoning under Uncertainty: Stochastic Optimization

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

In this paper, we revisit a supervised learning approach based on unrolling, known as $Psi$DONet, by providing a deeper microlocal interpretation for its theoretical analysis, and extending its study to the case of sparse-angle tomography. Furthermore, we refine the implementation of the original $Psi$DONet considering special filters whose structure is specifically inspired by the streak artifact singularities characterizing tomographic reconstructions from incomplete data. This allows to considerably lower the number of (learnable) parameters while preserving (or even slightly improving) the same quality for the reconstructions from limited-angle data and providing a proof-of-concept for the case of sparse-angle tomographic data.

Problem

• Sparse Angle Tomography
• Image Reconstruction
• Streak Artifact Reduction

Innovation

• ΨDONet Improvement
• Supervised Learning and Unfolding
• Sparse Angle Tomographic Imaging