Method for Segmenting Metal Objects in Projection Images, Evaluation Device, Computer Program and Electronically Readable Storage Medium

Simple SummaryContent extracted from patent full text and abstract with AI.

This patent describes a computer-implemented method and system using artificial intelligence to segment and accurately identify metal objects in two-dimensional x-ray projection images taken from multiple angles, improving the reconstruction of three-dimensional x-ray images. By combining AI-driven segmentation with a consistency check across the reconstructed 3D volume (including regions beyond the primary area of interest), the approach enhances the detection of metal objects, even those partly outside the main scan area, and reduces artifacts caused by metal in the final images.

Use CasesContent extracted from patent full text and abstract with AI.

• Medical imaging, particularly in computed tomography (CT) or c-arm x-ray imaging during surgeries, to segment and reduce artifacts from metal implants or surgical instruments.
• Preoperative, intraoperative, or postoperative imaging when patients have metal implants (e.g., joint replacements, screws, plates) to improve image quality and diagnostic accuracy.
• Industrial or security x-ray imaging where metal objects might generate artifacts, and precise localization and segmentation is needed.
• Automated quality control in manufacturing using x-ray imaging to distinguish and analyze metal parts inside larger assemblies.

BenefitsContent extracted from patent full text and abstract with AI.

• Provides robust and consistent metal object segmentation, improving reliability over traditional AI-only or threshold-based approaches.
• Allows precise localization of metal objects both inside and outside the main region of interest, reducing the risk of missing hidden artifacts.
• Significantly decreases metal artifacts in tomographic images, leading to clearer and more diagnostically valuable x-ray results, even during complex procedures with metallic tools or implants.
• Automates and streamlines the correction and masking of metal-induced artifacts, potentially saving time and reducing manual intervention in clinical workflows.
• Adaptable as a software solution for integration into existing imaging hardware, or as part of a standalone analysis device.

Technical Classifications (CPCs)

Inventors & Applicants

Patent Abstract

Computer-implemented method for segmenting metal objects (8, 9) in multiple two-dimensional projection images acquired using different projection geometries, each projection image showing a region of interest (7), wherein a three-dimensional x-ray image is reconstructed from the two-dimensional projection images in the region of interest (7), the method comprising the steps of:- using a trained artificial intelligence segmentation algorithm to calculate first binary metal masks for each projection image,- reconstructing a three-dimensional intermediate data set of a reconstruction region (10), which is larger than the region of interest (7), by determining, for each voxel of the intermediate data set, as a metal value the number of first binary metal masks showing metal in a pixel associated with a ray crossing the voxel,- determining a three-dimensional binary metal mask, in which a voxel shows metal when the metal value is larger than a threshold value and no metal in all other cases,- determining second binary metal masks for each projection image by forward projecting the three-dimensional binary metal mask using the respective projection geometries.