Improved accuracy of quantitative parameter estimates in dynamic contrast-enhanced CT study with low temporal...

Area of Science:

• Medical Imaging
• Radiology
• Image Processing

Background:

• Dynamic Contrast-Enhanced Computed Tomography (DCE-CT) is crucial for assessing tissue perfusion.
• Reducing radiation dose in DCE-CT is a significant clinical challenge.
• Principal Component Analysis (PCA) filtering can improve signal-to-noise ratio (SNR) in time-concentration curves.

Purpose of the Study:

• To investigate the efficacy of combining PCA filtering with arterial input function (AIF) estimation for DCE-CT.
• To assess the accuracy of kinetic parameter estimation at low temporal resolutions.
• To evaluate a pixel-by-pixel kinetic analysis method for DCE-CT data.

Main Methods:

• DCE-CT scans were performed with low temporal resolution to minimize patient radiation exposure.
• High temporal resolution AIF was generated using a previously published estimation method.
• PCA filtering was applied to 3x3 pixel regions to enhance SNR of tissue time-concentration curves.
• Kinetic analysis was performed using the modified Tofts' model and singular value decomposition.

Main Results:

• Accurate estimation of patients' AIFs was achieved.
• PCA filtering effectively reduced image noise by utilizing principal components of tissue curves.
• The combined method maintained accuracy in quantitative histogram parameters (volume transfer constant, rate constant, blood volume, blood flow) at sampling intervals up to 15 seconds.
• Results were superior to down-sampling alone.

Conclusions:

• The proposed PCA filtering combined with AIF estimation enables low-frequency scanning in DCE-CT to reduce patient radiation dose.
• This method is effective for pixel-by-pixel kinetic analysis of DCE-CT data, particularly for cervical cancer patients.