Abstract:PURPOSE Metal artifact remains a challenge in cone-beam CT images. Many image domain-based segmentation methods have been proposed for metal artifact reduction (MAR), which require two-pass reconstruction. Such methods first segment metal from a first-pass reconstruction and then forward-project the metal mask to identify them in projections. These methods work well in general but are limited when the metal is outside the scan field-of-view (FOV) or when the metal is moving during the scan. In the former, even reconstructing with a larger FOV does not guarantee a good estimate of metal location in the projections; and in the latter, the metal location in each projection is difficult to identify due to motion. Single-pass methods that detect metal in single-energy projections have also been developed, but often have imperfect metal detection that leads to residual artifacts. In this work, we develop a MAR method using a dual-layer (DL) flat panel detector, which improves performance for single-pass reconstruction. METHODS In this work, we directly detect metal objects in projections using dual-energy (DE) imaging that generates material-specific images (e.g., soft tissue and bone), where the metal stands out in bone images when nonuniform soft tissue background is removed. Metal is detected via simple thresholding, and entropy filtration is further applied to remove false-positive detections. A DL detector provides DE images with superior temporal and spatial registration and was used to perform the task. Scatter correction was first performed on DE raw projections to improve the accuracy of material decomposition. One phantom mimicking a liver biopsy setup and a cadaver head were used to evaluate the metal reduction performance of the proposed method and compared with that of a standard two-pass reconstruction, a previously published sinogram-based method using a Markov random field (MRF) model, and a single-pass projection-domain method using single-energy imaging. The phantom has a liver steering setup placed in a hollow chest phantom, with embedded metal and a biopsy needle crossing the phantom boundary. The cadaver head has dental fillings and a metal tag attached to its surface. The identified metal regions in each projection were corrected by interpolation using surrounding pixels, and the images were reconstructed using filtered backprojection. RESULTS Our current approach removes metal from the projections, which is robust to FOV truncation during imaging acquisition. In case of FOV truncation, the method outperformed the two-pass reconstruction method. The proposed method using DE renders better accuracy in metal segmentation than the MRF method and single-energy method, which were prone to false-positive errors that cause additional streaks. For the liver steering phantom, the average spatial nonuniformity was reduced from 0.127 in uncorrected images to 0.086 using a standard two-pass reconstruction and to 0.077 using the proposed method. For the cadaver head, the average standard deviation within selected soft tissue regions ( σ s ) was reduced from 209.1 HU in uncorrected images to 69.1 HU using a standard two-pass reconstruction and to 46.8 HU using our proposed method. The proposed method reduced the processing time by 31% as compared with the two-pass method. CONCLUSIONS We proposed a MAR method that directly detects metal in the projection domain using DE imaging, which is robust to truncation and superior to that of single-energy imaging. The method requires only a single-pass reconstruction that substantially reduces processing time compared with the standard two-pass metal reduction method.

DUALWISE-MAR: Dual-domain Multi-view Weakly supervised Segmentation Network for CBCT Metal Artifact Reduction.

PDS-MAR: a fine-grained projection-domain segmentation-based metal artifact reduction method for intraoperative CBCT images with guidewires

b-MAR: bidirectional artifact representations learning framework for metal artifact reduction in dental CBCT

Metal artifact reduction in 2D CT images with self-supervised cross-domain learning

SemiMAR: Semi-Supervised Learning for CT Metal Artifact Reduction

An Improved Dual-Domain Network for Metal Artifact Reduction in CT Images Using Aggregated Contextual Transformations.

Deep learning based projection domain metal segmentation for metal artifact reduction in cone beam computed tomography

Encoding Metal Mask Projection for Metal Artifact Reduction in Computed Tomography

View-Consistent Metal Segmentation in the Projection Domain for Metal Artifact Reduction in CBCT -- An Investigation of Potential Improvement

Building a Bridge: Close the Domain Gap in CT Metal Artifact Reduction

A comprehensive dual energy method for CBCT metal artifact reduction

Adaptive and Iterative Learning With Multi-Perspective Regularizations for Metal Artifact Reduction

A New Dental CBCT Metal Artifact Reduction Method Based on a Dual-Domain Processing Framework.

Stay In The Middle: A Semi-Supervised Model for CT Metal Artifact Reduction

DAN-Net: Dual-domain adaptive-scaling non-local network for CT metal artifact reduction

DuDoNet: Dual Domain Network for CT Metal Artifact Reduction

Convolutional Neural Network Based Metal Artifact Reduction in X-ray Computed Tomography

Coupling Model- and Data-Driven Networks for CT Metal Artifact Reduction

Metal Artifact Reduction in Dual Energy CT by Sinogram Segmentation Based on Active Contour Model and TV Inpainting

Single-pass Metal Artifact Reduction Using a Dual-Layer Flat Panel Detector.