長岡 亮, 長谷川 英之

Ryo NAGAOKA, Hideyuki HASEGAWA

富山大学学術研究部工学系

Graduate School of Science and Engineering for Research, University of Toyama

キーワード : pulse wave velocity, wavenumber analysis, phase, real part of complex signals

目的：脈波伝搬速度（Pulse Wave Velocity: PWV）を検出するための高分解能波数解析手法では，波数の推定に解析信号の位相情報が使用される．しかし，位相情報はフレーム方向における隣接信号の影響を受ける可能性があるため，我々は動脈壁の加速度波形の実部を使用した高分解能波数解析手法の改良を行った．手法：改良版波数解析手法では，我々は変動係数の2乗の逆数に対応する新たな評価関数を提案する．シミュレーションを実施してPWVの推定精度を評価し，in vivo実験で実現可能性も検討した．結果：シミュレーション実験では，提案法を使用した推定精度は，位相情報を使用した従来法による精度と同等であった．しかしながら，脈波に反射成分が含まれている場合，提案法を使用して推定したPWVは，従来法を使用して推定したPWVに比べて安定性に優れていた．さらに，in vivo実験では，提案法によって推定された心臓の駆出により発生する脈波の伝搬速度は，従来法によって推定された速度とほぼ同等であった（従来法：2.97 ± 1.2 m/s，提案法：4.82 ± 1.4 m/s）．一方，反射成分が存在する時相に従来法および提案法によって得られた推定PWV値は，それぞれ－1.13 m/sと－3.50 ± 0.9 m/sであった．2つの時刻で従来法により推定されたPWVは大きく異なり，PWVの推定結果は反射波の影響を受けやすいと考えられた．結論：シミュレーション実験およびin vivo実験の結果は，改良版高分解能波数解析手法が反射波の影響をより受けにくく，前進波および反射波の両方のPWVの推定精度が高いことを示唆した．

Purpose: In high-resolution wavenumber analysis for detection of pulse wave velocity (PWV), phase information of analytic signals is used to estimate the wavenumber. However, the phase information could be affected by the adjacent signals in the temporal direction. Therefore, we propose a modified high-resolution wavenumber analysis technique using real acceleration waveforms of the arterial wall. Method: In the modified wavenumber analysis, we propose a new evaluation function that corresponds to the inverse of the squared coefficient of variation. The accuracy of estimation of PWV was investigated by performing simulations, and the feasibility was also examined in an in vivo experiment. Results: In the simulation experiments, the estimation accuracy using the proposed method was comparable to that using the previous method using phase information. However, when the pulse wave included the reflection components, the PWV estimated using the proposed method was more stable than that estimated using the previous method. Also, in the in vivo experiments, at opening of the aortic valve, the velocity estimated by the proposed method was almost equal to that estimated by the previous method (previous: 2.97 ± 1.2 m/s, proposed: 4.82 ± 1.4 m/s). Meanwhile, when the reflection components were present, the estimated PWV values yielded by the previous and proposed methods were -1.13 and -3.50 ± 0.9 m/s, respectively. The PWVs at those two time points estimated by the previous method were quite different, and the PWV estimate was considered to be more affected by the reflected waves. Conclusion: The results of the simulations and in vivo experiments indicated that the modified high-resolution wavenumber analysis method was less affected by the reflected waves and more accurate in estimation of PWVs of both the forward and reflected waves.