Quantitative image quality evaluation of pixel-binning in a flat-panel detector for x-ray fluoroscopy

Area of Science:

• Medical Imaging
• Radiological Physics
• Detector Technology

Background:

• Flat-panel (FP) detectors in X-ray fluoroscopy require low-noise electronics and high data transfer rates.
• Pixel-binning is a technique to lower data transfer rates and improve signal-to-noise ratio (SNR).

Purpose of the Study:

• To quantitatively assess the image quality of different pixel-binning methods in X-ray fluoroscopy.
• To compare conventional data-line (D) and gate-line (G) binning with a novel alternate binning method.

Main Methods:

• Four acquisition methods were evaluated: no-binning, D binning, G binning, and alternate binning.
• Image quality was assessed using synthetic images and a guidewire detection task in a four-alternative forced-choice paradigm.
• Two detector orientations (data lines parallel and perpendicular to the guidewire) were tested at low (0.6 microR) and high (4.0 microR) exposures.

Main Results:

• At low exposure, D binning showed significantly better performance regardless of detector orientation.
• At high exposure, D binning was superior with data lines parallel to the guidewire, while G binning was better with data lines perpendicular.
• Alternate binning performed best on average across orientations and provided temporally fused images with smooth guidewires.

Conclusions:

• Exposure-dependent detector binning is suggested for fluoroscopy, switching between D binning and alternate binning based on exposure levels.
• Alternate binning offers a robust solution for enhancing image quality and guidewire detection in X-ray fluoroscopy.
• Computational observer models can effectively predict the performance of different acquisition methods.