Assessing image quality and dose reduction of a new x-ray computed tomography iterative reconstruction algorithm...

Area of Science:

• Medical Imaging Physics
• Radiological Sciences
• Computational Imaging

Background:

• Radiation dose reduction is critical in X-ray computed tomography (CT).
• Task-based image quality assessment provides objective measures for comparing reconstruction algorithms.
• Iterative reconstruction (IR) is a promising alternative to conventional filtered back projection (FBP).

Purpose of the Study:

• To compare the task-based image quality of CT images reconstructed by FBP and a novel IR algorithm.
• To quantify the potential radiation dose reduction achievable with the IR algorithm while maintaining diagnostic image quality.

Main Methods:

• Utilized a channelized Hotelling observer (CHO) model with dense difference Gaussian (DDOG) channels to assess image quality.
• Employed body and head phantoms imaged at various dose levels for data acquisition.
• Evaluated observer performance using area under the receiver operating characteristic (ROC) curves and area under the localization ROC curves.

Main Results:

• The IR algorithm achieved FBP-equivalent image quality at significantly lower radiation doses.
• Dose reduction ranged from 50%-82% for signal known exactly tasks and 67%-77% for localization tasks.
• IR algorithm consistently outperformed FBP in model observer performance across tested tasks.

Conclusions:

• The developed task-based quantitative assessment validates the performance of the new CT IR algorithm.
• The IR algorithm enables substantial radiation dose reduction in CT imaging while preserving image quality.
• Further validation with clinical data and human observers is recommended.