Accurate iodine quantification and residual error reduction with principal component analysis multimaterial...

Area of Science:

• Medical Imaging
• Quantitative CT
• Image Processing

Background:

• Dual-energy CT (DECT) enables multimaterial decomposition (MMD) for iodine quantification, vital for diagnostics.
• Residual errors from non-iodine materials in DECT iodine maps reduce accuracy, particularly in thoracic regions and low-dose scans.

Purpose of the Study:

• To assess the accuracy of iodine quantification and residual error using a principal component analysis multimaterial decomposition (PCA-MMD) algorithm.
• To evaluate PCA-MMD performance across diverse phantom sizes and radiation dose levels in DECT.

Main Methods:

• A photon-counting CT system scanned thorax phantoms with iodine and calcium inserts across three sizes (20.9-33.2 cm diameter) and dose levels (3-55 mGy).
• The PCA-MMD algorithm utilized PCA transformation and barycentric coordinates, avoiding matrix inversion instability.
• Iodine quantification was assessed via linear regression, RMSE, and CV; residual error was quantified; clinical validation involved comparing virtual noncontrast (VNC) to true noncontrast (TNC) images.

Main Results:

• PCA-MMD demonstrated near-unity regression slopes (0.98-0.99, R² ≥ 0.996) and reduced RMSE by up to 65% compared to standard MMD (0.10-0.39 vs. 0.20-0.72 mg/mL).
• Residual error was substantially lower with PCA-MMD (0.7-1.6%) versus standard MMD (16.1%-54.9%).
• At 3 mGy, PCA-MMD achieved lower RMSE (0.60 mg/mL vs. 0.95 mg/mL); clinical VNC images showed improved accuracy (15.5 HU vs. 20.9 HU, p=0.012) with excellent reproducibility (85% CV < 2%).

Conclusions:

• PCA-MMD significantly enhances iodine quantification accuracy and minimizes residual error in both phantom and clinical DECT applications.
• The algorithm's consistent performance across varying dose levels highlights its potential for clinical translation in quantitative imaging.