A regression approach to estimate the ultrasound contrast agent concentration

The quantification of contrast agent concentration is an ongoing research area to improve the ultrasound medical diagnosis capability. The accurate evaluation of local blood perfusion in a tissue and the detection of a tumor at early stage could be the main applications. The intensity of an ultrasound image is not directly correlated with the volumetric contrast agent concentration. The qualitative evaluation of the concentration carried out by the physicians could be noticeably imprecise. A method that can estimate the concentration, working only from the signals acquired by the ultrasound scanner, could be particularly useful in terms of precision, time consumption and cost. In this paper, a method to obtain the microbubble volumetric concentration, working from the radio frequency signals remotely acquired by means of an ultrasound scanner, is proposed. Firstly, a tissue rejection phase, based on a contrast pulse sequence technique, is applied to abate the tissue contribution. Then a non-linear regression approach, based on the support vector machine, is assumed to estimate the concentration in a region of interest. In vitro experiments, performed with non-specific and targeted microbubbles (able to selectively adhere to cancer cells), provided encouraging results.