Human visual perception-based grayscale contrast enhancement with adaptive window settings in computed tomography...

Area of Science:

• Medical Imaging
• Radiology
• Human Visual Perception

Background:

• Grayscale contrast is crucial for visual examination in medical imaging.
• Current medical applications rely on grayscale images displayed on various devices.

Purpose of the Study:

• To propose a novel method for grayscale contrast improvement in computed tomography (CT) diagnostics.
• To achieve effective contrast enhancement that provides sufficient grayscale contrast and natural image appearance aligned with human visual perception (HVP).

Main Methods:

• Introduced a new quantitative metric, contrast-perceived to spatial frequency ratio (CPSFR), to assess perceptual image quality.
• Evaluated "Window Settings Automatic" (maximizing CPSFR) against "Window Settings Preset" and "Window Settings Manual" using phantom and patient images.
• Assessed target detectability and diagnostic satisfaction in phantom studies (720 images) and clinical studies (80 images with liver lesions).

Main Results:

• "Window Settings Automatic" showed statistically significant improvements in accuracy and sensitivity in phantom studies (p < 0.01).
• Clinical studies revealed "Window Settings Automatic" outperformed other methods, achieving the highest mean score and over 40% of top-scoring votes.
• The proposed method demonstrated superior performance in both quantitative and qualitative assessments.

Conclusions:

• The novel method effectively enhances grayscale contrast in diagnostically relevant regions.
• The approach yields natural image appearance, aligning with HVP to improve target detectability and diagnostic performance.
• This technique offers a valuable tool for improving CT diagnostic accuracy.