Scatter-glare corrections in quantitative dual-energy fluoroscopy

Area of Science:

• Medical Imaging
• Radiology
• Image Processing

Background:

• Cardiac and respiratory motion cause artifacts in traditional time subtraction angiography for ventricular imaging.
• Accurate quantification of parameters like ejection fraction requires correction for scatter and veiling glare (SVG) in digital subtraction angiography (DSA).

Purpose of the Study:

• To investigate a convolution filtering method for estimating scatter and veiling glare (SVG) in DSA images.
• To improve the accuracy of quantitative ventricular imaging by correcting for nonlinearities in DSA.

Main Methods:

• A two-step convolution filtering approach was used: grey level transformation followed by spatial frequency content adjustment.
• Gaussian convolution kernels with varying full width at half-maximum (FWHM) were applied to humanoid chest phantom images.
• Scatter and veiling glare (SVG) estimation was refined using a lookup table (LUT) adjusted for patient-specific x-ray settings and thickness.

Main Results:

• Convolution filtering with a 75-pixel FWHM kernel achieved an average root-mean-square (rms) percentage error of 9.7% in SVG estimation across 16 phantom cases.
• The method demonstrated effectiveness with various projections, thicknesses, and beam energies.
• The developed technique allows for accurate SVG estimation, crucial for quantitative analysis.

Conclusions:

• The proposed convolution filtering method effectively estimates scatter and veiling glare (SVG) in DSA images.
• This technique enhances the accuracy of quantitative parameters such as ejection fraction and left ventricular volume.
• The motion-immune dual-energy technique combined with SVG correction offers a robust solution for ventricular imaging.