Spectral CT metal artifact correction

点击翻译按钮获取中文摘要

PROJECT SUMMARY Radiation therapy treatment planning can be severely impacted by the presence of metal objects such as implants and orthopaedic hardware. Metal objects cause artifacts in computed tomography (CT) images that obscure anatomical structures and alter the CT numbers, both of which are critical to estimate accurately for the purpose of planning radiation therapy. These uncertainties can cause underdosing of tumors and overdosing of healthy tissue. Existing metal artifact reduction techniques do not fully mitigate all artifacts created by the metal objects and are known to introduce new artifacts. This project will develop a spectral CT imaging method to reduce metal artifacts while maintaining CT number accuracy and soft tissue contrast. We propose to reduce metal artifacts in CT imaging by using state-of-the-art acquisition techniques, combined with an optimization- based reconstruction framework. We developed a constrained `one-step' spectral CT image reconstruction (cOSSCIR) algorithm in previous work and preliminary studies demonstrate feasibility of the proposed algorithm to reduce metal artifacts to <8 HU error. The incorporation of physical effects into the data model is one method by which the algorithm reduces metal artifacts. The optimization framework developed by our group uniquely incorporates constraints that mitigate undersampling due to unreliable measurements that pass through metal and also enable acquisition approaches that will reduce the number of unreliable measurements. The methods are designed to correct metal artifacts broadly and automatically without requiring knowledge of the implant material. The project objective to reduce metal artifacts while maintaining soft tissue contrast and CT number accuracy will be achieved by further developing the cOSSCIR algorithm and investigating its application to both dual-kV and photon-counting spectral acquisition methods using simulations, phantom experiments, and clinical photon-counting CT datasets. The algorithm will also be evaluated relative to task of radiation therapy planning for prostate cancer in the presence of hip prostheses using simulations and phantom experiments. The developed spectral CT metal artifact correction method will be compared to gold-standard images and an established metal artifact reduction technique. Successful completion of the project aims will result in a method to reduce metal artifacts in CT images while maintaining soft tissue contrast and CT number accuracy that has been validated on simulated and experimental phantom data.