Abstract
Liver steatosis, a type of fatty liver disease, has been gaining significant attention in the medical field due to its global prevalence. Currently, large-scale evaluations of liver steatosis are hindered by the variability of the available non-invasive imaging tools. While conventional ultrasound often serves as the primary imaging modality for liver assessments, its accuracy is limited by its qualitative nature and operator dependency. To address these challenges, this study introduces a novel approach that reduces the degrees of freedom in simultaneously estimating ultrasonic backscatter (BSC) and attenuation (AC) coefficients. By exploiting the Rayleigh scattering behavior where the BSC trend remains approximately constant at lower frequencies, our method focuses on estimating just two parameters, offering increased precision. Ultrasound measurements were performed on a cohort of 29 Sprague-Dawley rats subjected to either a control or methionine and choline deficient diet. Postimaging, histological evaluations were conducted. Using support vector machine classification with leave-one-out cross-validation, our approach utilizing two degrees of freedom outperformed the three degrees of freedom method, achieving an accuracy of 96.6% compared to 72.4% in detecting liver steatosis. The findings indicate that this new method can significantly reduce estimation variability and enhance liver steatosis classification, paving the way for more consistent and reliable non-invasive liver assessments.