Multifrequency Electrical Impedance Tomography With Ratiometric Preprocessing for Imaging Human Body Compartments

Abstract

Reconstructing the structural images of human body compartments as point-of-care imaging could be possible using electrical impedance tomography (EIT), but the reconstructed image deteriorates due to high conductivity contrast <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\mu $ </tex-math></inline-formula> between anomaly and background. In this study, a multifrequency EIT (mf-EIT) with a ratiometric preprocessing (ratiometric EIT) has been developed in order to minimize <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\mu $ </tex-math></inline-formula> while maintaining high distinguishability without any <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$a$ </tex-math></inline-formula> <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">priori</i> information. The preprocessing of ratiometric EIT is achieved by extending the framework of ratiometric methods that uses the ratio of two measurement signals. Based on the proposed ratiometric preprocessing, ratiometric frequency-difference adjacent EIT (rfda-EIT) is newly derived. The rfda-EIT is qualitatively and quantitatively evaluated by numerical simulation under the variant conditions of <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\mu $ </tex-math></inline-formula> (=95%–10%) and experiment with eight subjects’ calves. As the results, the rfda-EIT scores the area ratio error of subcutaneous adipose tissue (SAT) 17.09% < ARE <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$^{\mathrm {SAT}} &lt; 28.13$ </tex-math></inline-formula> %, the position error of bone tibia 0.14% < PE <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$^{\mathrm {tibia}} &lt; 10.19$ </tex-math></inline-formula> %, the position error of bone fibula 9.51% < PE <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$^{\mathrm {fibula}} &lt; 1.74$ </tex-math></inline-formula> %, and the correlation coefficients <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$0.79&lt; $ </tex-math></inline-formula> CC < 0.91 in the numerical simulation. The total mean area ratio error <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\langle $ </tex-math></inline-formula> ARE <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$^{\mathrm {SAT}}\rangle $ </tex-math></inline-formula> by the rfda-EIT is lower than the error by the classical frequency-difference adjacent EIT (fda-EIT) from 12.70% to 5.18% in the experiment. Moreover, the true positive rate TPR <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$^{\mathrm {tibia}}$ </tex-math></inline-formula> for the bone tibia detection by the rfda-EIT is increased from 50.0% to 87.5% compared with the rate by the fda-EIT.