Evaluation of Nonlinear Pixel-Wise Squaring for Enhancing Low-Contrast Resolution in CT Imaging

Area of Science:

• Medical Imaging
• Radiology
• Image Processing

Background:

• Early detection of hyperacute stroke relies on computed tomography (CT) scans.
• Subtle changes in early CT images are difficult to identify due to blurring.
• Improving low-contrast resolution is critical for accurate diagnosis.

Purpose of the Study:

• To investigate the effectiveness of nonlinear pixel-wise squaring in enhancing low-contrast resolution in CT images.
• To determine if this image processing technique can improve the visibility of subtle findings in early stroke detection.

Main Methods:

• Utilized the Catphan 700 phantom to generate standard and processed CT images (Square-5, Square, Square+5).
• Evaluated image quality using physical metrics like task transfer function (TTF) and contrast-to-noise ratio (CNR_LO).
• Conducted subjective visual assessments of image clarity and noise perception.

Main Results:

• Nonlinear processing effectiveness varied with scan mode (volume/helical) and display window (wide/narrow).
• The Square-5 method significantly improved subjective CNR for volume scans (p < 0.01).
• The offset-free Square method with a narrow window setting significantly improved CNR (p < 0.05) without increasing perceived noise.

Conclusions:

• Nonlinear squaring processing can improve low-contrast visibility in CT scans.
• Optimal performance is contingent on specific scan modes and display window settings.
• The Square-5 method for volume scans and the offset-free Square method with a narrow window offer significant improvements in low-contrast visibility.