川嶋 大介

Heparin concentration c in a blood extracorporeal circulation has been real-timely predicted based on the relaxation strength Δεm at relaxation frequency fm extracted by relaxation time distribution (RTD). The simulated extracorporeal circulation was conducted to optimize the number of Δεm for the prediction of c using the porcine whole blood (WB) and low-leukocyte and −platelet blood (LLPB) under the condition of the gradual increment of c from 0 to 8 U/mL with constant flow rate and blood temperature. The experimental results show that among the three relaxation strengths Δε1, Δε2 and Δε3 (in ascending order of frequency), Δε2 at f2 = 5.2 ∼ 6.2 MHz and Δε3 at f3 = 42 ∼ 50 MHz were correlated to c. The Δε3 was decreasing with increasing c in both cases, which was influenced by the plasma macromolecular concentrations, while the Δε2 was increased with increasing c in WB case but was hardly changed in LLPB case because the Δε2 is influenced by the blood cell concentrations and the shape changes of blood cell membranes. Heparin concentration c is estimated by the linear regression formula cPRE=a1(Δε2-Δε2c=0)+a2(Δε3-Δε3c=0) (a1 = -0.991, a2 = -0.123) within the mean absolute percentage error (MAPE) of 0.291.

Albumin andγ-globulin concentrations in subcutaneous adipose tissues (SAT) have been quantified by multivariate regression based on admittance relaxation time distribution (mrARTD) under the fluctuated background of sodium electrolyte concentration. ThemrARTD formulatesP=Ac+Ξ(P: peak matrix of distribution function magnitudeγˆand relaxation timesτˆ,c: concentration matrix of albumincAlb,γ-globulinGloc, and sodium electrolyteNac,A: coefficient matrix of a multivariate regression model, andΞ: error matrix). ThemrARTD is implemented by two processes which are: (1) the training process ofAthrough the maximum likelihood estimation ofPand (2) the quantification process ofcAlb,Gloc, andNacthrough the model prediction. In the training process, a positive correlation is present betweencAlb,Gloc, andNactoγˆ1atτˆ1= 0.1 as well asγˆ2atτˆ2= 1.40 μs as under a fixed concentration of proteins solution into a porcine SAT (cAlb= 0.800-2.400 g/dL,Gloc= 0.400-1.200 g dl-1andNac= 0.700-0.750 g dl-1). ThemrARTD method quantifiescAlb,Gloc, andNacin SAT with an absolute error of 33.79%, 44.60%, and 2.18%, respectively.

The multi-frequency complex electrical impedance tomography (mfc-EIT) has been proposed to image the spatiotemporal extra/intra-cellular ion concentrations. In mfc-EIT, four frequency-dependent Jacobian matrices described by the partial derivatives of complex impedance (resistance and reactance) with respect to complex conductivity (conductivity σ and relative permittivity ε) as a function of radial frequency are newly introduced as a complex Jacobian matrix. The mfc-EIT inversely reconstructs the σ and ε images as a function of frequency from the complex impedance. Based on the electric field simulation, the σ and ε images reconstructed by mfc-EIT and conventional EIT with a general Jacobian matrix are evaluated by spectral stabilities <SS> which is an indicator of reasonable spectral trend. The σ and ε images by mfc-EIT have lower <SS> (σ: 2.7 × 10-9 in and ε: 3.0 × 10-4 in) than those by conventional EIT (σ: 5.5 × 10-8 and ε: 5.6 × 10-3), which indicates mfc-EIT reconstructs the reasonable frequency response. The mfc-EIT is also applied to the imaging of the extra/intrace-cellular ion transfer around cell-spheroids in experiment. According to the reconstructed σ and ε images, the extracellular ion concentration reflected by σ at f = 1 kHz are increased over time, while the intracellular ion concentration reflected by σ and ε at f = 1 MHz are decreased over time, which is attributed to the ion transfer via the ion channel. Thus, mfc-EIT captures the extra/intra-cellular ion concentration changes by ion transfer via ion channels.

The admittance relaxation time distributions (aRTD) has been proposed for the quantification of protein (albumin and -globulin) concentration c. The aRTD extracts peaks of distribution function [S] at relaxation times τ [s] from measured admittance τ [S]. In order to clarify the relationship between , τ, and c, aRTD was performed on solutions and porcine subcutaneous adipose tissue (SAT) with an injected current frequency f from 10Hz to 1MHz under the albumin concentration Albc of 0.80 g/dL to 2.30 g/dL, the -globulin concentration Gloc of 0.40 g/dL to 1.15 g/dL, and protein mixtures PMc. As the results, two peaks (P1 and P2) of P1 at τP1 from 0.014 s to 0.083 s and P2 at τP2 from 0.98 μs to 1.27 μs were extracted, which changes alongside an increase in Albc, Gloc, and PMc. In order to evaluate the predictability and accuracy of aRTD, the aRTDP1 and aRTDτP1 extracted by aRTD are compared with iRTDP1 and iRTD τP1 extracted by typical impedance RTD (iRTD). Regression line analysis shows c is linearly proportional to aRTDP1 and inversely proportional to aRTDτP1 with a coefficient of determination R2 above 0.8, which is much larger than R2 between c and P1 extracted by iRTD. The aRTD on SAT provides additional correlation through P2 at τP2, which increases proportionally to Gloc and inversely proportional to Albc and PMc. Moreover, the average fitting error ϵ of aRTD is smaller than ϵ of iRTD by 12%, which highlights aRTD accuracy in the fitting.