Pose estimation of known objects during transmission tomographic image reconstruction

Area of Science:

• Medical Physics
• Image Reconstruction
• Computational Imaging

Background:

• Transmission tomography is crucial for medical imaging but suffers from artifacts caused by high-density metal objects.
• Metal artifacts, such as streaking, degrade image quality and hinder accurate diagnosis in filtered back projection (FBP) images.
• Accurate reconstruction of images with metallic components like implants or brachytherapy devices remains a significant challenge.

Purpose of the Study:

• To develop and evaluate an algorithm for artifact reduction in transmission tomography with metallic objects of known composition and shape but unknown pose.
• To improve image quality by eliminating streaking artifacts originating from high-contrast metallic materials.
• To enhance the speed and accuracy of image reconstruction by incorporating prior knowledge of metallic object characteristics.

Main Methods:

• Utilized an alternating minimization (AM) algorithm based on a Poisson data model, minimizing I-divergence to maximize log-likelihood.
• Incorporated a steepest descent method to determine the unknown pose (position and orientation) of metallic objects.
• Constrained image pixels using known attenuation values of metallic objects or masked projection data in object shadows.

Main Results:

• The developed AM algorithm effectively eliminates streaking artifacts caused by metallic objects in transmission tomography.
• Exploiting prior knowledge of high-density materials significantly accelerated algorithm convergence compared to methods without this information.
• Demonstrated successful artifact reduction in two-dimensional simulations, with potential for three-dimensional extension.

Conclusions:

• The alternating minimization algorithm offers a robust solution for image reconstruction in transmission tomography with metallic artifacts.
• Prior knowledge of metallic object properties is key to improving the efficiency and accuracy of artifact correction.
• The presented method holds promise for enhancing diagnostic accuracy in medical imaging scenarios involving metallic implants or devices.