1.

Introduction

Near-infrared spectroscopy (NIRS) at a wavelength between 700 and $900\mathrm{nm}$ is useful for continuously and noninvasively assessing the oxygenation state in vivo. This method is based on the relatively high tissue transparency of near-infrared light in addition to hemoglobin absorption in the region. There are many reports on the clinical applications of NIRS, such as monitoring cerebral oxygen metabolism during surgery^{1, 2} or for the respiratory management of neonates.^{3, 4} In particular, it may prove useful to monitoring cerebral circulation during cardiac surgery under extracorporeal circulation, which is associated with high risks of cerebral ischemia, embolism, hyperperfusion after release from the artificial heart-lung machine, and cerebral complications such as postoperative deterioration of neural recognition.^{5, 6} Commercially available NIRS monitors have been used for comparing oxygen saturation $\left({\mathrm{SO}}_2\right)$ with internal jugular vein oxygen saturation $\left({\mathrm{SjvO}}_2\right)$ ^{7, 8, 9, 10, 11, 12, 13, 14} or electroencephalographs^{15, 16, 17} and for assessing changes in blood volume.¹⁸

However, the NIRS apparatus used employs continuous wave (CW) near-infrared light. Therefore, absolute values of hemoglobin concentration and highly reproducible ${\mathrm{SO}}_2$ cannot be obtained. To overcome these difficulties, more sophisticated technologies have been introduced.

Phase modulation spectroscopy (PMS) ^{19, 20, 21, 22, 23} and time-resolved spectroscopy (TRS) ^{24, 25, 26, 27} have been developed for this purpose. However, these new technologies have not been evaluated in detail in a clinical study.²⁸

In TRS, the distribution of optical path lengths is directly measured. Its temporal profile is analyzed using a photon diffusion equation²⁵ and the Microscopic Beer-Lambert law,²⁹ thereby enabling determination of the hemoglobin concentration.

Our three-wavelength time-resolved spectroscopy system (TRS-10)³⁰ was tested in cardiopulmonary bypass surgery, during which the absolute concentrations of oxygenated, deoxygenated, and total hemoglobin were estimated. ${\mathrm{SO}}_2$ was simultaneously calculated as a percentage of oxygenated hemoglobin of total hemoglobin. The optically estimated values were compared with those of ${\mathrm{SjvO}}_2$ and hematocrit (Hct) in arterial blood. The validity and usefulness of our optical measurements are demonstrated in this clinical study.

2.

Materials and Methods

2.2.

Time-Resolved Spectroscopy System (TRS-10)

Our time-resolved spectroscopy system (TRS-10) uses the time-correlated single photon counting (TCPC) method to measure the temporal profile of the detected photons, as shown in Fig. 1 . The apparatus is described elsewhere.³⁰

Fig. 1

(a) Appearance of TRS-10 (Hamamatsu Photonics K.K.). (b) Block diagram of the TRS-10 system.

In brief, the system consists of a three-wavelength ( $759\mathrm{nm}$ , $797\mathrm{nm}$ , and $833\mathrm{nm}$ ) light pulser (PLP, Hamamatsu Photonics K.K., Hamamatsu, Japan) as the light source, which generates light pulses with a peak power of $60\mathrm{mW}$ , pulse width of $100\mathrm{ps}$ , pulse rate of $5\mathrm{MHz}$ , and an average power of $30\mu \mathrm{W}$ . For the detection, a photomultiplier tube (PMT, H6279-MOD, Hamamatsu Photonics K.K.), a constant fraction discriminator (CFD), a time-to-amplitude converter (TAC), an A/D converter, and a histogram memory were all assembled. The three PLPs emit light pulses on a time series, and three-wavelength light pulses are guided into one illuminating optical fiber by a fiber coupler (CH20G-D3-CF, Mitsubishi Gas Chemical Company, Inc., Japan). A single optical fiber (GC200/250L, Fujikura Ltd., Japan) with a numerical aperture (N.A.) of 0.21 and a core diameter of $200\mu \mathrm{m}$ was used for illumination. An optical bundle fiber (LB21E, Moritex Corporation, Japan) with an N.A. of 0.21 and a bundle diameter of $3\mathrm{mm}$ was used to collect diffuse light from the tissues.

3.

Results

The variations of mean optical path length, ${\mu }_s^{\prime }$ , tHb, and ${\mathrm{SO}}_2$ in 23 patients under resting conditions are plotted against age, as shown in Fig. 2 . The mean optical path length was $30.4\pm 2.8\mathrm{cm}$ , ${\mu }_s^{\prime }$ was $10.0\pm 0.8{\mathrm{cm}}^{-1}$ , tHb was $42.1\pm 9.2\mu \mathrm{M}$ , and ${\mathrm{SO}}_2$ was $67.3\pm 3.8\%$ . ${\mathrm{SO}}_2$ varied the least among the patients, while the tHb varied significantly.

Fig. 2

The variations of (a) mean optical path length, (b) ${\mu }_s^{\prime }$ , (c) tHb, and (d) ${\mathrm{SO}}_2$ in 23 patients under resting conditions are plotted against age.

Figure 3 shows a typical case (Case 4, F, 63Y) of photon count rates (light intensity), mean optical path lengths, ${\mu }_s^{\prime }$ , and ${\mu }_a$ at three wavelengths during coronary-artery bypass surgery. These values (except ${\mu }_s^{\prime }$ ) changed significantly due to hemodilution during extracorporeal circulation.

Fig. 3

Time course of TRS-10 parameters during surgery in a typical case (Case 4: F, 63Y): (a) count rate, (b) mean optical path length, (c) ${\mu }_s^{\prime }$ , and (d) ${\mu }_a$ . H: drawing arterial blood to determine hematocrit (Hct).

Figure 4 shows the relationship between the TRS parameters ( ${\mu }_a$ , ${\mu }_s^{\prime }$ , and the mean optical path length at three wavelengths) and the Hct value of arterial blood collected during surgery in a typical case (Case 4, F, 63Y). In this patient, the Hct value at the beginning of surgery was 23% but decreased to a minimum of 12% during extracorporeal circulation. ${\mu }_a$ changed linearly in relation to Hct changes, while ${\mu }_s^{\prime }$ remained almost unchanged. The mean optical path length increased with the decrease of Hct.

Fig. 4

Relationship TRS parameters at three wavelengths and hematocrit (Hct) value in the typical case (Case 4: F, 63Y): (a) ${\mu }_a$ , (b) ${\mu }_s^{\prime }$ , and (c) mean optical path length.

Figure 5a shows the time course of changes in the concentration of ${\mathrm{HbO}}_2$ , Hb, and tHb, and the ${\mathrm{SO}}_2$ and ${\mathrm{SjvO}}_2$ are shown in Fig. 5b. When extracorporeal circulation was started, ${\mathrm{HbO}}_2$ and tHb decreased abruptly, together with decreases in ${\mathrm{SO}}_2$ . After the release of extracorporeal circulation, those values returned to the initial levels.

Fig. 5

Hemoglobin concentrations and ${\mathrm{SO}}_2$ as measured by TRS-10 during surgery in the typical case (Case 4: F, 63Y): (a) ${\mathrm{HbO}}_2$ , Hb, and tHb; (b) TRS ${\mathrm{SO}}_2$ , ${\mathrm{SjvO}}_2$ . H: in arterial blood drawn to determine Hct.

The ${\mathrm{SO}}_2$ estimated by TRS was nearly the same as that of ${\mathrm{SjvO}}_2$ before extracorporeal circulation, but during circulation, they behaved differently. Especially when rewarming was carried out, ${\mathrm{SO}}_2$ did not vary as much as ${\mathrm{SjvO}}_2$ .

Figure 6 shows that fluctuations in ${\mathrm{SO}}_2$ and ${\mathrm{SjvO}}_2$ were separated during extracorporeal circulation in one case (Case 3, M, 52Y). The values of ${\mathrm{SjvO}}_2$ increased significantly when the pump was on, whereas ${\mathrm{SO}}_2$ decreased rapidly.

Fig. 6

Fluctuations of ${\mathrm{SO}}_2$ and ${\mathrm{SjvO}}_2$ were separated during extracorporeal circulation in one patient (Case 3: M, 52Y). When the pump was on, the values of ${\mathrm{SjvO}}_2$ increased significantly. Meanwhile, those of ${\mathrm{SO}}_2$ decreased rapidly. (a) ${\mathrm{HbO}}_2$ , Hb, and tHb; (b) TRS ${\mathrm{SO}}_2$ , ${\mathrm{SjvO}}_2$ . H: in arterial blood drawn to determine Hct.

Figure 7 shows the correlation between the Hct in arterial blood and tHb in nine patients. The correlation among patients was high, with $r^2=0.63$ , showing that tHb measured by TRS has a good linearity with Hct. In addition, intraindividual correlations were also very high ( $r^2=0.72\pm 0.22$ ; data is not shown).

Fig. 7

Relationship between tHb by TRS-10 and hematocrit (Hct) in nine patients.

Figure 8 shows the linear regression plot and the Bland-Altman plot of ${\mathrm{SO}}_2$ and ${\mathrm{SjvO}}_2$ for all data of six patients: during the whole surgery (a), during the surgery except perfusion (b), during partial perfusion (c), and during total perfusion (d). The correlation coefficients were $r^2=0.33$ in (a), $r^2=0.46$ in (b), $r^2=0.29$ in (c), and $r^2=0.42$ in (d). The bias and the precision ( $\pm 2$ SD) were $0.66\pm 19.6\%$ in (a), $2.90\pm 18.7\%$ in (b), $0.92\pm 16.5\%$ in (c), and $-5.21\pm 17.9\%$ in (d). The correlation coefficient in (c) and bias in (d) were lower than those in other stages of the operation. This variation was split into inter- and intraindividual analysis, and the results of correlation, bias, and precision among six patients and the mean of these values within six patients are shown in Table 1 . Precisions become smaller for intraindividual analysis, although biases did not show such changes.

Fig. 8

Linear regression correlation between ${\mathrm{SO}}_2$ and ${\mathrm{SjvO}}_2$ (upper) and difference plots for ${\mathrm{SO}}_2$ and ${\mathrm{SjvO}}_2$ using the Bland-Altman method (lower) in all data points of six patients: (a) during the whole surgery, (b) during the surgery except perfusion, (c) during partial perfusion, and (d) during total perfusion.

Figure 9a shows the plot of the ${\mu }_s^{\prime }$ during the surgery in one patient (Case 5, M, 69Y). The ${\mu }_s^{\prime }$ decreased gradually during perfusion only in this case. The other eight patients did not show such changes of ${\mu }_s^{\prime }$ during surgery [Fig. 9b, Case 7, M, 66Y].

Fig. 9

(a) ${\mu }_s^{\prime }$ decreased gradually during perfusion in one case (Case 5: M, 69Y). This trend was observed only in this case. (b) ${\mu }_s^{\prime }$ remained unchanged in a typical case (Case 7: M, 66Y). A similar pattern was observed in the other eight patients except Case 5.

Table 1

The correlation and agreement between SO2 and SjvO2 .

4.

Discussion

In the present study, the observed values of tHb showed wide variation in resting conditions, reflecting different preoperative blood drawing for each patient. In contrast, ${\mathrm{SO}}_2$ in resting conditions was very similar among patients. This parameter, which is independent of patient age and Hct, is very useful for monitoring the brain tissue oxygenation of each patient.

During surgery under extracorporeal circulation, the Hct value decreased further due to hemodilution to prevent the formation of thrombi. The various parameters (count rate, mean optical path length, ${\mu }_a$ ) determined by TRS and tHb responded well to changes in such hemodilution as shown in Figs. 3 and 4. In particular, ${\mu }_a$ varied linearly with Hct. Although several conventional approaches use CW-methods that assume a constant value of the mean optical path length in the calculation, it should be noted that there are large fluctuations among the patients. Thus, the quantitative measurements employed here are necessary to continuously monitor the condition of patients during the operation. Although ${\mu }_a$ and the mean optical path length were affected by Hct changes, ${\mu }_s^{\prime }$ showed almost constant values without being affected by Hct changes. This result supports the proposition that absorption and scattering can be handled independently.

In this study, we observed the decrease of ${\mu }_s^{\prime }$ during perfusion in one patient, as shown in Fig. 9a. The trend of ${\mathrm{HbO}}_2$ , Hb, tHb, and ${\mathrm{SO}}_2$ were not much different from other cases. System drift was significantly smaller than the change in ${\mu }_s^{\prime }$ of the patient. The chief contributor to the scattering of tissues is their organelle contents such as mitochondria, endo-plasmic reticulum, nucleus, and so forth.³⁴ There are reports on the relationship between the ${\mu }_s^{\prime }$ and the developmental stages of a newborn baby’s brain.³⁵ Moreover, change in ${\mu }_s^{\prime }$ of a piglet brain caused by the depolarization has been demonstrated when the piglets were exposed to hypoxia.^{27, 36} The possibility of inducing scattering changes by almost any kind of solute (particularly glucose, mannitol, and sucrose) has been suggested.³⁴ Although more research is necessary for ascertaining the clinical implications of scattering change, it has been reported that such information is important. ^{27, 34, 35, 36} It may be used to prevent risks leading to brain damage, such as cerebral edema.

We have demonstrated the usefulness of our TRS system for monitoring the state of brain oxygenation, as well as tissue optical parameters related to physiological conditions of cerebral tissue. The TRS allows the quantification of ${\mathrm{HbO}}_2$ , Hb, and tHb of the brain. Both correlations between Hct and tHb within patients and among patients were high, even though the anatomical structure such as the thicknesses of the scalp and skull might be different in individuals. The effect of the outer layer may be smaller in time-resolved spectroscopy than that of the conventional CW method.^{37, 38}

The trends of ${\mathrm{SO}}_2$ and ${\mathrm{SjvO}}_2$ matched during the surgery. Especially in the rewarming period, both values temporarily declined (Figs. 5 and 6), reflecting the oxygen demand-supply imbalance due to the increased metabolism of the brain and temperature effect on the hemoglobin dissociation curve. However, the ${\mathrm{SjvO}}_2$ value changed between approximately 40% and 90%, while the ${\mathrm{SO}}_2$ value changed within 50% and 80%. Thus, the ${\mathrm{SO}}_2$ variation tended to be smaller than the ${\mathrm{SjvO}}_2$ variation. In addition, the correlation and bias between the ${\mathrm{SO}}_2$ and ${\mathrm{SjvO}}_2$ also differed depending on the conditions of extracorporeal circulation, although a good correlation was observed except for perfusion for the period. These values do not necessarily match, as seen in Fig. 6, because ${\mathrm{SjvO}}_2$ shows a balance between blood flow and metabolism in the whole brain, making it difficult to measure very small changes. Moreover, there is a report of increase in the shunt of arterial and venous blood flow,¹⁰ resulting in increase of ${\mathrm{SjVO}}_2$ .

In contrast, the TRS may be able to perform localized tissue measurements. Thus, the ${\mathrm{SO}}_2$ is not necessarily higher than the ${\mathrm{SjvO}}_2$ , due to the ${\mathrm{SO}}_2$ showing average saturation of arterial, capillary, and venous blood, because the ${\mathrm{SjvO}}_2$ shows information of the whole brain hemodynamics, including the deeper region that extracts less oxygen than the neocortex region.¹² Some studies on the correlation between NIRS ${\mathrm{SO}}_2$ and ${\mathrm{SjvO}}_2$ have shown that it almost matches,^{7, 28} while others have shown that it does not. ^{8, 9, 12, 39} The contradiction between these results is attributable to surgery without extracorporeal circulation or differences in the method of inducing anesthesia or the temperature control conditions.

${\mathrm{SjvO}}_2$ is used as an indicator of cerebral oxygenation, where higher ${\mathrm{SjvO}}_2$ shows better oxygenation conditions. However, there is a report of postoperative disturbance of higher brain functions in patients who showed high ${\mathrm{SjvO}}_2$ during extracorporeal circulation.⁴⁰ Therefore, the judgment of normoxia-hypoxia by ${\mathrm{SjvO}}_2$ may be insufficient, and simultaneous measurements of ${\mathrm{SO}}_2$ by TRS can give more accurate information about the oxygenation conditions of patients.

5.

Conclusion

We have developed a time-resolved spectroscopy system (TRS-10) and have used it in cardiopulmonary bypass surgery as a brain oxygenation monitor. ${\mathrm{SO}}_2$ measured by TRS paralleled fluctuations in ${\mathrm{SjvO}}_2$ , although these correlations depended on the perfusion conditions. Further detailed studies are necessary to clarify the relation between these values.

A good correlation was noted between tHb by TRS and Hct value (12% to 37.8%) in arterial blood $(r^2=0.63)$ . Intra-cerebral hemodynamics of patients were captured well by TRS in real time and noninvasively.

Consequently, our results indicate that TRS is a valid spectroscopic method for quantifying cerebral hemodynamics and is useful for clinical monitoring.

Clarification of the relationship between ${\mu }_s^{\prime }$ and the pathological condition of the patient and improvement of the analytical approach ^{40, 41, 42, 43} for measuring deeper regions are future challenges.