Pixel-based small-window parametric ultrasound imaging for liver tumor characterization.

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Radiology

Background:

• Characterizing liver tumors is clinically challenging.
• Ultrasound parametric imaging can enhance contrast but relies on specific scatterer distributions.
• B-mode imaging has limitations in liver tumor visualization.

Purpose of the Study:

• To propose a pixel-based small-window parametric ultrasound imaging method.
• To improve liver tumor detectability using weighted horizontally normalized Shannon entropy (WhNSE) and fuzzy entropy (FE).
• To evaluate the effectiveness of FE and WhNSE in enhancing ultrasound image contrast.

Main Methods:

• A pixel-based parametric imaging approach using a sliding window was employed.
• Novel fuzzy entropy (FE) and weighted horizontally normalized Shannon entropy (WhNSE) were calculated for each pixel.
• Contrast-to-noise ratio (CNR) and region of interest (ROI) detection abilities were assessed via simulations and clinical data.

Main Results:

• Fuzzy entropy (FE) imaging demonstrated the highest improvement in detecting hyperechoic ROIs, achieving up to 457.31% CNR gain (p < 0.01).
• Weighted horizontally normalized Shannon entropy (WhNSE) imaging showed superior hyperechoic ROI detection performance with a CNR of 1.607 ± 0.816 (p = 0.05).
• Both methods significantly enhanced CNR compared to B-mode ultrasound.

Conclusions:

• Pixel-based parametric imaging using FE and WhNSE effectively enhances ultrasound image contrast and tumor detectability.
• The pixel-based fuzzy entropy imaging method, considering neighboring pixel relationships, achieved superior detection performance.
• These novel entropy-based methods offer promising advancements for liver tumor characterization in clinical ultrasound.