Improved reference frequency method for attenuation imaging using multi-frequency coupling

Abstract

A well-known acoustical parameter used for tissue characterization is the attenuation coefficient slope (ACS), which has shown potential in clinical applications, such as quantization of liver fat content. Conventional methods estimate ACS from backscattered echo data in the spectral domain. However, they are affected by system dependencies and require a calibrated reference phantom to compensate for diffraction effects. To overcome these limitations, the Reference Frequency Method (RFM) was introduced, enabling ACS estimation without a reference phantom. Building on this framework, a method named TNV-RFM that leverages the multi-frequency coupling using the Total Nuclear Variation is proposed. Data from simulated and tissue-mimicking phantoms, and in vivo liver acquisitions from healthy and metabolic dysfunction–associated steatotic liver disease (MASLD) volunteers—diagnosed through clinical evaluation, cardiometabolic profile, and histopathological analysis of laparoscopic biopsies—were used to compare both methods. Results demonstrated a consistently lower coefficient of variation with TNV-RFM (12.2%, 23.9%, and 30.7% for simulations, phantoms, and liver samples, respectively) vs RFM (20.7%, 38.9%, and 54.1%). These findings suggest that TNV-RFM provides more stable and reliable ACS estimates, further improving the conventional RFM framework in attenuation imaging.