Quantification of contrast agent materials using a new image- domain multi material decomposition algorithm based on...

Area of Science:

• Medical Imaging
• Computational Imaging
• Materials Science

Background:

• Dual-Energy CT (DECT) enables material separation using dual energy spectra.
• Current DECT material decomposition methods face challenges with accuracy and workflow.
• Decomposing more than two materials is crucial for clinical and industrial applications.

Purpose of the Study:

• To propose a novel material decomposition algorithm for DECT.
• To improve the efficiency and accuracy of multi-material decomposition (MMD).
• To develop a semi-automatic MMD algorithm using clustering techniques.

Main Methods:

• Implemented a local clustering method for barycentric coordinates assignment, reducing search domains for increased precision.
• Utilized a fast bi-directional Hausdorff distance measurement for optimizing coordinate selection.
• Employed the Doubly Local Wiener Filter Directional Window (DLWFDW) algorithm to mitigate noise interference.

Main Results:

• The algorithm achieved separation errors below 2% for blood and 9% for fat in clinical images.
• Phantom data demonstrated high accuracy in separating materials with varying concentrations.
• Achieved up to 93% accuracy for calcium plaque and 97.1% for iodine contrast agents.

Conclusions:

• A new, easily implementable material decomposition algorithm for DECT was developed.
• The algorithm enhances the MMD workflow through semi-automatic material coordinate assignment via clustering.
• The proposed method offers comparable qualitative and quantitative performance to existing techniques.