AS-Mamba: Asymmetric Self-Guided Mamba Decoupled Iterative Network for Metal Artifact Reduction

2026-02-06

AI Summary

This work proposes AS-Mamba, a novel deep learning framework for metal artifact reduction in CT imaging that addresses the challenge posed by the directional geometric characteristics of metal artifacts, which severely degrade image quality and are poorly modeled by existing methods. AS-Mamba is the first to incorporate a state space model to explicitly capture the directional streaking patterns inherent in metal artifacts. It further integrates frequency-domain magnitude spectrum correction to mitigate intensity inhomogeneities caused by beam hardening effects and introduces a self-guided contrastive regularization strategy to enhance generalization across diverse clinical scenarios. Evaluated on both public and clinical dental CBCT datasets, AS-Mamba substantially outperforms current state-of-the-art methods, effectively suppressing directional artifacts while preserving fine structural details.

Technology Category

• Machine Learning: Learning with Manifolds
• Computer Vision: Multi-modal Vision
• Search and Optimization: Metareasoning and Metaheuristics

Application Category

• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

Metal artifact significantly degrades Computed Tomography (CT) image quality, impeding accurate clinical diagnosis. However, existing deep learning approaches, such as CNN and Transformer, often fail to explicitly capture the directional geometric features of artifacts, leading to compromised structural restoration. To address these limitations, we propose the Asymmetric Self-Guided Mamba (AS-Mamba) for metal artifact reduction. Specifically, the linear propagation of metal-induced streak artifacts aligns well with the sequential modeling capability of State Space Models (SSMs). Consequently, the Mamba architecture is leveraged to explicitly capture and suppress these directional artifacts. Simultaneously, a frequency domain correction mechanism is incorporated to rectify the global amplitude spectrum, thereby mitigating intensity inhomogeneity caused by beam hardening. Furthermore, to bridge the distribution gap across diverse clinical scenarios, we introduce a self-guided contrastive regularization strategy. Extensive experiments on public andclinical dental CBCT datasets demonstrate that AS-Mamba achieves superior performance in suppressing directional streaks and preserving structural details, validating the effectiveness of integrating physical geometric priors into deep network design.

Problem

• metal artifact reduction
• CT image quality
• directional artifacts
• structural restoration
• beam hardening

Innovation

• Mamba
• State Space Model
• metal artifact reduction
• frequency domain correction
• self-guided contrastive regularization