Echo-power estimation from log-compressed video data in dynamic contrast-enhanced ultrasound imaging

Area of Science:

• Medical Imaging
• Ultrasound Technology
• Quantitative Analysis

Background:

• Ultrasound (US) scanners often apply non-linear processing to image data for visualization, storing it as compressed video.
• Dedicated manufacturer software exists for quantification but has limitations due to machine-specific data access and analysis functionalities.
• This hinders comparable quantitative analyses across different US systems, particularly for contrast-enhanced ultrasound.

Purpose of the Study:

• To investigate if linearization of compressed US video images can serve as an alternative to dedicated software for echo-power estimation.
• To assess the feasibility of performing comparable quantitative analyses across different US systems using a standardized approach.

Main Methods:

• Compared echo-power estimations from linearized compressed video images (VueBox, PixPower) against dedicated software (CHI-Q) using an Aplio 50 US system.
• Evaluated in vitro data for linear relationships between echo-power and contrast agent concentration.
• Analyzed in vivo data to determine differences in echo-power estimation between methods, excluding saturated/thresholded pixels.

Main Results:

• In vitro analysis showed strong linear relationships (R(2) ≥ 0.9996) between echo-power and concentration for all methods, with slopes near theoretical predictions.
• In vivo analysis revealed minimal mean differences in echo-power estimates between VueBox/PixPower and CHI-Q (-0.25 dB and -0.17 dB, respectively).
• VueBox requires calibration files, while PixPower needs empirical correction, highlighting the need for manufacturer data sharing.

Conclusions:

• Linearization of compressed US video images provides a reliable alternative for echo-power estimation, comparable to dedicated software.
• This approach facilitates multicentric studies and wider adoption of quantitative contrast-enhanced ultrasound techniques.
• Standardized linearization methods could overcome limitations of machine-specific software, improving data comparability across different ultrasound systems.