Comparison of algorithms for estimating ultrasound attenuation when predicting cervical remodeling in a rat model

The future of biomedical imaging involves obtaining quantitative diagnostic information about tissue structure and function rather than just relying on qualitative structural assessments. Ultrasound attenuation or loss of signal within tissue can be used to quantify micro-structural changes in the tissue consistent with cervical remodeling. In order to obtain the best estimates, the algorithms used to estimate ultrasound attenuation (spectral log difference method, spectral difference method, and hybrid method) need to be optimized and compared. In this study, the effects of the inhomogeneities within the ROI on the accuracy of the 3 algorithms were studied, and the optimal ROI size (number of independent echoes laterally and number of pulse lengths axially) was quantified for each method. The conclusions were then validated by analyzing an ultrasound image of a pregnant rat cervix from a custom-made high-frequency ultrasound imaging system.