Cerebral perfusion and blood–brain barrier assessment in brain trauma using contrast-enhanced near-infrared spectroscopy with indocyanine green: A review

Abstract

Contrast-enhanced near-infrared spectroscopy (NIRS) with indocyanine green (ICG) can be a valid non-invasive, continuous, bedside neuromonitoring tool. However, its usage in moderate and severe traumatic brain injury (TBI) patients can be unprecise due to their clinical status. This review is targeted at researchers and clinicians involved in the development and application of contrast-enhanced NIRS for the care of TBI patients and can be used to design future studies. This review describes the methods developed to monitor the brain perfusion and the blood–brain barrier integrity using the changes of diffuse reflectance during the ICG passage and the results on studies in animals and humans. The limitations in accuracy of these methods when applied on TBI patients and the proposed solutions to overcome them are discussed. Finally, the analysis of relative parameters is proposed as a valid alternative over absolute values to address some current clinical needs in brain trauma care. In conclusion, care should be taken in the translation of the optical signal into absolute physiological parameters of TBI patients, as their clinical status must be taken into consideration. Discussion on where and how future studies should be directed to effectively incorporate contrast-enhanced NIRS into brain trauma care is given.

Introduction

Near-infrared spectroscopy (NIRS) represents a valuable complementary non-invasive monitoring tool for use in the context of traumatic brain injury (TBI).^1,2 However, to date, measurements on these patients in this context with commercially available NIRS devices have not shown parity with the invasive techniques regarding their ability to detect episodes of ischemia.^3–5 Changes in optical signal due to the passage of a contrast dye, such as indocyanine green (ICG), are potentially more readily detected than changes due to endogenous chromophores alone (e.g. hemoglobin).⁶

The use of contrast-enhanced NIRS optodes applied to the scalp has definite advantages over other contrast-enhanced imaging techniques (e.g. computerized tomography (CT), magnetic resonance imaging (MRI), positive emission tomography) by potentially allowing continuous, repeatable bedside-monitoring as well economical, safety, logistical and radiation exposure benefits.^7–10 Despite these advantages, studies on TBI patients using contrast-enhanced NIRS have been limited and the utility of this technique in brain trauma care is still subject to investigation. This review aims to analyze the methods adopted so far to assess cerebral blood flow ($\mathit{CBF}$) and blood–brain barrier (BBB) integrity using contrast-enhanced NIRS with ICG on animals and humans; the limits of these assessments in TBI patients; the solutions proposed to overcome them; and to discuss how the use of contrast-enhanced NIRS could address some of the current clinical challenges in the care of TBI patients. Researchers interested in the clinical applications of contrast-enhanced NIRS may find this review useful with regard to the design of their future studies.

The dye kinetic in dynamic-contrast enhanced images can be divided into two phases: the first phase can be used for the evaluation of brain perfusion, as expressed by the Stewart-Hamilton equation, because the dye stays predominantly in the vascular space; subsequently, in the following 2–10 min, the passage of the dye into the extravascular space can be used for the assessment of vascular permeability.^11,12 It should be noted that these two phases are not rigidly separated, but they overlap at different time rates.¹³

Cerebral blood flow

Since the first description of the technique was made by Colacino et al.,¹⁴ multiple methods have been introduced to assess $\mathit{CBF}$ given the NIRS diffuse reflectance. The values considered were the optical density changes during the ICG passage or an estimation of the ICG concentration using the modified Beer–Lambert law.

Central volume principle

The convolution formula for the central volume principle allows measurement of $\mathit{CBF}$ using the dye’s concentration inside the brain volume of interest (VOI)¹⁵

$$Q\left(t\right)=\mathit{CBF}·{\int }_0^{t}Ca\mathrm{ }\left(\text{τ}\right)\cdot R(t\mathrm{ }-\mathrm{ }\text{τ})d\text{τ}$$ (1)

where $Q$ is the tissue residue function, equivalent to brain tracer concentration, $Ca$ is the dye’s concentration in the arterial blood and $R$ is the impulse residue function (IRF). The IRF is the fraction of all the tracer inputted into the tissue in an extremely short time period that is still present at time $t$.¹⁵

If a non-diffusible contrast dye is injected into the arteries so that all of it reaches the tissue instantaneously (i.e. $Ca\left(t\right)$ resembles a delta function) then IRF, multiplied by $\mathit{CBF}$, is equivalent to $Q$ (Figure 1). In this case, the width of $Q$ is associated to the mean transit time ($\mathit{MTT}$) and the area under the curve to the cerebral blood volume ($\mathit{CBV}$).^16,17

Figure 1.

Graph of the ICG concentration in the tissue ($\mathit{Q}$) when IRF is a delta function. The distance between points “a” and “b”, the height of $\mathit{Q}$, is associated with the cerebral blood flow ($\mathit{CBF}$); the distance between points “a” and “c”, the width of $\mathit{Q}$, is associated with the mean transit time ($\mathit{MTT}$); and the area colored in red under the curve is associated with the cerebral blood volume ($\mathit{CBV}$).

If instead the contrast agent is injected into the tissue from the peripheral venous system, the measurement of $Q$ is affected by the tracer delivery rate and a deconvolution between $Q$ and $Ca\left(t\right)$ is required in order to estimate $\mathit{CBF}$.^16,18 This deconvolution process results in estimating $\mathit{CBF}·\mathit{IRF}$, whose width is associated to $\mathit{MTT}$ and the area under the curve to $\mathit{CBV}$. Given the definition, the following relation holds¹⁸

$$\mathit{MTT}=\frac{\mathit{CBV}}{\mathit{CBF}}$$ (2)

In different studies, $Q$ and $Ca$ were directly measured using continuous-wave (CW)-NIRS system, subsequent to the ICG injection into the peripheral veins.^16,19–21

Diop et al.²² enhanced this method and calculated the absolute values of $\mathit{CBF}$ using a time-domain (TD)-NIRS device on piglets by measuring the absorption and scattering properties of the brain prior to ICG injection. These baseline tissue optical properties were also used by the same group to calibrate a diffuse correlation spectroscopy device, which alone can only measure relative changes of $\mathit{CBF}$, to monitor the absolute values of $\mathit{CBF}$ continuously.²³

Modified Fick principle

The Fick principle states that the flow passing through a system in a steady state can be measured from the tracer concentration’s inlet and outlet and its uptake from the system²⁴

$${FC}_{in}-{FC}_{\mathit{out}}=J$$ (3)

where $F$ is the flow, $C_{in}$ and $C_{\mathit{out}}$ are the inlet and outlet tracer concentrations, and $J$ is the uptake of a substance from the system.

This principle can be used to estimate the $\mathit{CBF}$ through the following formula¹⁵

$$\mathit{CBF}=\mathrm{ }\frac{{\Delta q}_x}{{\int }_0^{\Delta t}\left\{{\left[x\right]}_a\left(t\right)\right.-\mathrm{ }\left.{\left[x\right]}_v\left(t\right)\right\}dt}$$ (4)

where $\mathit{CBF}$ is the amount of flow in a finite time interval $\Delta t$, ${\left[x\right]}_a$ and ${\left[x\right]}_v$ are the concentrations of the substance $x$ in the arteries and in the veins, and ${\Delta q}_x$ is the amount of substance $x$ accumulated in the brain tissue per unit mass throughout the time interval.

Edwards et al.²⁵ measured a sudden increase of oxyhemoglobin concentration in the brain to analyze the $\mathit{CBF}$ using a modified version of the Fick principle. They measured changes in cerebral saturation for time periods less than the blood $\mathit{MTT}$ across the brain, which allowed them to consider the venous tracer concentration to be zero and to apply the following equation²⁵

$$\mathit{CBF}=\frac{Q\left(t\right)}{{\int }_0^{\Delta t}{C}_a\left(t\right)dt}$$ (5)

where $Q$ is the substance accumulation in the brain and $C_a$ is the concentration of the substance in the arteries.

Similarly, other published studies measured the $\mathit{CBF}$ through the ICG contrast-enhanced NIRS signal using this modified version of the Fick principle.^6,20,26,27 Using an array of probes on infants, Kusaka et al.²⁸ compared $\mathit{CBF}$ in different brain regions in a topographic manner.

Regional cerebral blood volume and regional mean transit time

Keller et al.²⁹ introduced a method to measure the $\mathit{CBF}$ in the VOI, regional CBF ($\mathit{rCBF}$), using a formula derived from equation (2)

$$\mathit{rCBF}=\frac{\mathit{rCBV}}{\mathit{rMTT}}$$ (6)

where $\mathit{rCBV}$ and $\mathit{rMTT}$ are, respectively, the regional CBV and MTT in the VOI.

This method was used on an optical signal collected with probes on the scalp and inside the skull.^30–32

Regional cerebral blood volume

Wolf et al.³³ separated the signal from the arteries and veins based on the fluctuation of the optical density due to the cardiac and respiratory rates respectively. To measure the ICG concentration in the blood ($c_{\mathit{ICG}}^{\mathit{blood}}$), Keller et al.²⁹ compared the changes in optical density caused by the systolic and diastolic rates before and after the ICG injection

$$c_{\mathit{ICG}}^{\mathit{blood}}\left(t\right)=\left[\frac{A\mathrm{ }\left(t\right)}{A\mathrm{ }\left(t_0\right)}-1\right]\frac{a_{Hb}{c}_{Hb}}{a_{\mathit{ICG}}}$$ (7)

where $A$ is the amplitude of the pulse measured, $a_{Hb}$ is the hemoglobin extinction coefficient, $a_{\mathit{ICG}}$ the ICG extinction coefficient and $c_{Hb}$ the hemoglobin concentration, which was assumed constant.

Based on the modified Beer–Lambert law, the ICG concentration in the tissue $c_{\mathit{ICG}}^{\mathit{tissue}}$ was linked to the change in $OD$ before and after the ICG injection³⁰

$$c_{\mathit{ICG}}^{\mathit{tissue}}\left(t\right)=\mathrm{ }\frac{OD\left(t\right)-OD\left(t_0\right) }{a_{\mathit{ICG}}·\mathit{DPF}·d}$$ (8)

where $\mathit{DPF}$ is the differential pathlength factor and $d$ is the distance between the source and the detector.

The optical properties of ICG upon injection are unstable due to the inhomogeneous distribution of the dye in the blood, which affects the $OD$ analysis at the beginning of the recording.³⁰ The accuracy of equation (8) can be enhanced by considering this instability using a value, $\mathit{Cdo}$, which retrospectively analyses the $OD$ at the beginning of the injection based on the stabilized optical properties subsequent to sufficient intermixing of ICG and blood³⁰

$$c_{\mathit{ICG}}^{\mathit{tissue}}\left(t_0=\mathit{rMTT}\right)=\mathrm{ }\frac{\mathit{Cdo}}{a_{\mathit{ICG}}·\mathit{DPF}·d}$$ (9)

Finally, the $\mathit{rCBV}$ was measured from the following formula²⁹

$$\mathit{rCBV}=\frac{c_{\mathit{ICG}}^{\mathit{tissue}}\mathrm{ }}{c_{\mathit{ICG}}^{\mathit{blood}}}$$ (10)

Since the venous compartment does not show a pulsation related to the heart rate, it is not comprised within $c_{\mathit{ICG}}^{\mathit{blood}}$ in equation (7), and so the $\mathit{rCBV}$ obtained from equation (10) is not representative of the venous volume.

Regional mean transit time

Based on the Fick principle, Keller et al.²⁹ calculated the difference between the inflow and outflow rates of ICG by measuring the ICG concentration inside the VOI

$$c_{\mathit{ICG}}^{\mathit{tissue}}\left(t\right)={\int }_0^{t}i\left(t\right)\mathrm{ }-o\left(t\right)dt$$ (11)

where $i$ is the concentration-time course of ICG in the inlet and $o$ in the outlet. They then used the formula from dilution theory to obtain the transport function, $g\left(u\right)$

$$o\left(t\right)={\int }_0^{t}i\left(t-u\right)·g\left(u\right)du\mathrm{ }$$ (12)

where $u$ is the variable of integration in the convolution procedure.

The transport function is the function of particles’ transit times from the beginning to the end of the VOI.³⁴ $\mathit{rMTT}$ was measured from the transport function using a deconvolution algorithm.²⁹

Blood flow index

Based on a model introduced by Perbeck et al., Kuebler et al. evaluated the blood flow index (BFI) in a study on piglets.^35,36 The BFI is the ratio between the peak of ICG concentration in the examined tissue and the time interval between 10% and 90% of this peak (Figure 2). The value obtained is an arbitrary value linked to the CBF with an unknown coefficient.³⁷

Figure 2.

Graph of measurements of the BFI derived from the changes of ICG optical density or ICG brain tissue concentration during dye passage. After the venous injection (point “a”), there is an increase of ICG signal from the brain until a peak (point “d”). The ratio between this peak and the time for concentration/optical density to vary from 10% to 90% of its maximum value (points “b” and “c”, respectively), highlighted in red, forms the BFI.

This method has been tested in humans and translated into clinical practice on infants and adults.^20,37–41 It should be noted that Wagner et al.,³⁷ unlike others, considered the optical density changes without transforming them into ICG concentration.

Different results demonstrated method limitations. In a study on healthy volunteers, the BFI values derived did not correlate with those measured using the method described in section ‘Regional cerebral blood volume and regional mean transit time’.³⁰ In adults who had suffered an ischemic stroke, the BFI differed significantly between the affected and unaffected hemispheres, but not compared to the controls.^38,39 In a study of TBI patients undergoing decompressive craniectomy, a surgical microscope saw no abnormal BFI via ICG fluorescence on the visible brain parenchyma.⁴² In similar patients, Rothoerl et al.⁴³ found no correlation between the values obtained using the BFI and a radioisotopes technique. However, in the aforementioned study, there were two elements that differed from the methodology introduced by Kuebler et al. Firstly, the optical data were not converted into ICG concentration but into an oxygen saturation index instead. Secondly, the time interval considered was from the point of contrast dye injection to the peak of tissue concentration.

Distribution of time of flight of photons

TD-NIRS allows one to measure the distribution of time of flight of photons (DTOF).

Liebert et al.⁴⁴ separated the absorption changes in the intracranial tissue (ICT) and extracranial tissue (ECT) during the ICG passage using three statistical moments of DTOF: total number of detected photons, mean time of flight of photons and variance. This analysis is an estimation of $\mathit{CBF}$ based on the different ICG inflow rates between the ICT, represented mainly by the variance, and that of the ECT, represented by the total number of detected photons.^44–46 A slower ICG passage through the brain was measured in TBI and stroke patients, in agreement with their abnormal cerebral perfusion or brain lesion.^47–49 Elliott et al.⁴⁵ extended this method to quantify the $\mathit{CBF}$ in pigs by using the variance in the algorithms described in sections ‘Central volume principle’ and ‘Modified Fick principle’.

It should be mentioned that a comparison between ICG inflow into the ICT and into the ECT can also be performed on the distribution of times of arrival of fluorescence photons (DTA).^50–52 The analysis on DTA serves to more accurately assess cerebral perfusion than that on DTOF.^51,53

BBB damage

As is true for other contrast dyes (e.g. Evans blue), the ICG passage across the BBB could be ascribed to the multiple mechanisms that compound BBB damage.⁵⁴

The ICG passage into the extravascular space has been measured through the dye absorption or fluorescence properties. On animals, a comparison of the optical signal against CT perfusion scans and histological samples showed that the ICG retention in the interstitial brain tissue is linked to BBB damage.^46,55–58 The high ICG tissue concentration is limited to the brain areas affected by the BBB damage rather than the whole brain.^46,57 Ergin et al.⁵⁹ showed that certain drugs leak into the interstitial tissue as well as dye, which suggests potential clinical application of contrast-enhanced NIRS by means of optimizing end-organ dosage.

Similarly to these animal models, Liebert et al.⁶⁰ described a prolonged optical signal from the ICG in patients with identified BBB damage, due to a slower washing out of the dye from the brain tissue. These results are in agreement with Kamp et al., who measured the ICG retention in the parenchyma of severe TBI patients with acute subdural hematoma using a surgical microscopic during a decompressive craniectomy.⁴² They reported a higher retention of the dye in the patients who went on to achieve less favorable outcomes.

Limitations

There are multiple limitations in the translation of the ICG optical signals into physiological parameters. These limits are further amplified in clinical scenarios where patients’ statuses can influence the values obtained with unpredictable variables.

Central volume principle's assumptions

The analysis of flow using the central volume principle can lead to an overestimation in the case of BBB damage.¹⁷ This is because, in the case of a diffusible tracer, the concentration inside the blood volume must be considered as in equilibrium with the volume in which it diffuses.¹⁵ In TBI patients, the presence of BBB damage can change the diffusive properties of the dye and so affect the validity of the central volume principle. These changes would depend on the extent of the areas affected by the BBB damage, and they would not be consistent between patients.

Fick principle's assumptions

The Fick principle’s assumptions are not completely fulfilled and accounted for in the modified Fick principle. The basic principle assumes a steady state, wherein a constant rate of indicator inflow and outflow exists.²⁴ However, the level of arterial ICG inflow changes over time ($dt$ of equation (6)), meaning the tissue accumulation rate varies. Furthermore, measurement of this rate in TBI patients is impaired because the rate of arterial ICG inflow is not consistent between patients or even in the same patients at different times. Secondly, time $t$ in equation (5) must be earlier than the dye’s passage into the cerebral veins; this was previously considered as the mean transit time of blood flow through the brain. However, the mean transit time that should instead have been considered is the one pertaining to the volume of tissue illuminated. This is because all the substances which have inflowed from the artery during $dt$ must be present in the illuminated volume in order for the denominator of equation (5) to be valid. The analysis of the mean transit time of the illuminated areas can be particularly complicated in cases, as in TBI patients, where there could be different levels of perfusion across the brain. It should also be noted that Edwards et al.²⁵ considered $Q$ of equation (5) as the substance accumulation rate. However, the units of $Q$ are concentration per unit volume rather than concentration per unit volume per second.²⁵ The measurement of the tracer concentration at a single time $t$ would not precipitate measurement of its accumulation rate, as to do so, the former must be compared with at least a second measurement at another time.

Similarly to the ICG arterial inflow, $J$ can be inconsistent in TBI patients because of the BBB damage. Consequently, the Fick principle’s assumed steady state cannot be entirely satisfied due to the ICG leakage into the interstitial space. As discussed in section ‘Central volume principle assumptions’, the grade of this interference would vary with the severity of the disease.

Indocyanine green as an indicator of plasma volume

The blood volume mainly comprises red blood cell volume and plasma volume. For a correct analysis of the blood volume, both these components have to be measured.^61,62 Due to the Fåhraeus effect, the ratio between the two is different depending on the size of the vessels.⁶³ The brain volume illuminated would encompass vases of different sizes and consequently with different ratios between the cellular and plasma volumes. Since the ICG remains confined within the plasma, its optical signal is linked to the plasma volume only, and is proportional to the fraction of vessels of different sizes within the illuminated tissue.

Based on an experiment on healthy volunteers using single-photon emission computed tomography, Keller et al. assigned a cerebral-to-large-vessel hematocrit ratio to the plasma volume in the tissue illuminated.^29,64 However, a meta-analysis showed that different disease states can result in large variations in the amount of hematocrit in the microcirculation, which makes the cerebral-to-large-vessel hematocrit ratio different in these patients.⁶² Similarly, in trauma patients, changes in either the macro- or microcirculation can result in different cerebral-to-large-vessel hematocrit ratios than those obtained from analyses on non-trauma patients.^65–67

Extracranial tissue interference

Analysis of the diffuse reflectance on the scalp can lead to an underestimation of the true $\mathit{CBF}$ because of the high percentage of optical pathlength covered by the ECT and the lower blood flow in the scalp compared to the brain (Figure 3(a)).⁶⁸ Due to the high level of photon absorption by ICG, the dye passage through the ECT may further exacerbate this interference. The degree of interference is linked to the dye’s concentration in the ICT and ECT, which changes according to the ICG kinetic in each.⁴⁶

Figure 3.

Representation of the ICG kinetic in the ICT and the ECT without a brain lesion. (a) Standard physiological state. (b) First ICG bolus passage. The ICG reaches the vessels in the ICT before the ECT due to the different perfusion levels between the two layers. (c) First ICG bolus passage. The ICG is being washed out from the ICT, while its concentration in the ECT is increasing due to the slower blood transit time there than in the ICT. (d) ICG kinetic a few minutes post-injection. The ICG blood concentration is reduced due to dilution and clearance; the ICG remains in the venous system of the microcirculation of the two layers with a higher concentration in the ECT than in the ICT.

Indocyanine green’s kinetics in the intracranial and extracranial tissue

The ICG kinetic within the two layers is related to their hemodynamic and vascular permeability. As described in section ‘Introduction’, the relevance of these components depends on the time from the injection.

First indocyanine green’s passage

Firstly, there is a single bolus ICG inflow into the ICT and then into the ECT, due to a different perfusion velocity in the two layers (Figure 3(b)).^30,40

Secondly, once in the tissue, the ICG passage is quicker in the ICT than in the ECT due to the different mean transit times (Figure 3(c)).^31,32

Due to the kinetics of the first passage, the optical density curve generated by the ICG is initially represented predominantly by a signal coming from the ICT alone, and subsequently, by a combination of the ICT and ECT, with an increasing component from the latter as the ICG leaves the ICT while still passing through the ECT.⁴⁶

The ICG kinetic between the two layers and the optical curve obtained are generally in agreement with the results observed from the gadolinium’s passage in an MRI reported in the literature.⁴¹ A comparison between the passage of the bolus of gadolinium in the ICT and ECT showed a peak delayed by a few seconds, as well as a slower recovery to baseline in the former than in the latter.⁴¹

As explained in section ‘Distribution of time of flight of photons’, abnormal cerebral perfusion can delay the passage of the bolus into the brain and eliminate the difference between the time of inflow into the ICT and ECT (Figure 4(b) and (c)). This different ICG kinetic would change the shape of the optical curve retrieved.⁷

Figure 4.

Representation of the ICG kinetic in the ICT and the ECT with a brain lesion. (a) Standard physiological state. (b) First ICG bolus passage. The ICG does not reach the ICT before the ECT due to a reduction of brain perfusion. (c) First ICG bolus passage. The ICG reaches the ICT and the ECT at approximately the same time. (d) ICG kinetic a few minutes post-injection. The ICG blood concentration is reduced due to the dye’s dilution and clearance; the ICG remains in the venous system of the microcirculation of the two layers with a higher concentration in the ECT than in the ICT; the ICG leaks into the interstitial tissue in the areas with BBB damage.

Indocyanine green’s washing out and retention

After the peak of bolus passage, the optical signal from the ICG does not return immediately to the baseline. This suggests that the ICG is partially retained in the vascular space and is being slowly washed out (Figure 3(d)). This washing-out process can be delayed if the ICG leaks into the interstitial tissue potentially due to loss of BBB integrity (Figure 4(d)).

Washing out from the intravascular space

The ECT slower perfusion compared to that of the ICT results in a slower ICG washing out from the ECT than from the ICT (Figure 3(d)). The ICG is more likely to remain in the venous system (e.g. venules) where the blood pressure and transit velocity are lower (Figure 3(d)).⁶⁹ Due to the different washing-out rates, the percentage of optical signal coming from the ECT, compared to that from the ICT, increases over time.⁴⁶

Retention in the interstitial space

In case of BBB damage, the dye firstly inflows into the interstitial tissue and subsequently leaves once the concentration inside the blood is lower than that in the interstitial tissue.¹³ Consequently, the ICG optical signals from injured brain areas are higher than the surrounding areas without BBB damage until the ICG has washed out from the interstitial tissue (Figure 4(d)). It should be mentioned that in the case of inflamed ECT, the ICG also outflows into its interstitial space.^70,71

Quantification of the extracranial tissue interference

Monte Carlo simulations have been used on healthy volunteers’ MRIs at different physiological perfusion levels to simulate the diffuse reflectance obtained from the changes in optical density within ICT and ECT due to the ICG passage.^72,73 The percentage of optical signal from the brain during the first ICG bolus passage was estimated to be approximately 70%, mainly due to the higher perfusion and blood volume in the ICT than in the ECT. The optical signal coming from the brain during the ICG passage was significantly higher than that estimated before injection, which led to the conclusion that the passage of dye enhances the brain signal.⁷

However, the ratio assigned in the Monte Carlo simulations mentioned above is not representative of the entire ICG passage, because the fraction of ICG in the ECT progressively increases over the ICT (Figure 3(b) and (c)). Furthermore, this increase has a more profound effect on the percentage of diffuse reflectance from the brain than the simple proportional effect of different ICG ratios between the layers. This is because the optical properties in the ECT change the depth of the photon pathway and thus the volume of illuminated brain tissue.⁷⁴

Finally, brain-injury patients may have different perfusion levels and blood volumes than the physiological parameters considered by the Monte Carlo simulations. Consequently, the ICG ratio between the two layers would be different from the one assigned (Figure 4(b) and (c)).

Techniques to remove the extracranial tissue interference

The ECT contribution to the optical signal differs according to the SD distance of the probe and the type of NIRS method.

Single-channel continuous wave

A single-channel CW-NIRS has been used to measure the $\mathit{CBF}$ on healthy volunteers.^30,75 However, the optical signal coming from the brain detected by a single long channel can be highly contaminated by the ICG passage through the ECT.⁷⁶ Separation of the two layers using a single-channel CW-NIRS monitor is not possible and so neither is correction for the aforementioned ECT interference.

Single-channel frequency-domain

Steinbrink et al.⁴⁰ used one source and one detector of an FD-NIRS device (e.g. ISS OxiplexTS) on four patients during cardiopulmonary bypass surgery to measure the ICG concentration change in the ICT and ECT. The analysis of the mean time of flight of photons allowed for the separation of the ICG concentration in each layer.^40,77 While the FD-NIRS device monitored optical changes at different levels of $\mathit{CBF}$ during the various stages of the intervention, these were not seen using a non-depth resolved CW-NIRS device due to the ECT interference with the brain signal.

Multiple source-detector distances

Different studies have shown that multiple optodes at increasing SD distances can be a valuable tool to separate the optical changes in the ICT and ECT during the ICG passage.^21,41,78 This is because the longer channels are more affected by the ICG passage through the brain than the shorter channels, which are mainly confined in the ECT.⁷⁹ The difference in intensity between channels allows for regression of the ECT signal from the brain signal.⁸⁰ Alternatively, plotting the decrease in intensity against the SD distances makes it possible to measure the optical properties of each layer.⁸¹ This can be done using a CW- or a FD-NIRS.⁸²

The use of the FD-NIRS facilitates measurement of the phase-shift according to the SD distance as well as the intensity levels.^83–87 Therefore, the ICG passage through the layers measured with a FD-NIRS can also be divided using the model described in section ‘Single-channel frequency-domain’ adapted to a multidistance approach system.⁴¹

A comparison between multidistance depth-resolved NIRS and an intracranial probe showed no correlation between the results obtained, which led to the conclusion that this method has certain limitations in the brain signal measurement, possibly due to effects from the ECT.³²

Time-domain

The combined thickness of the ECT and the skull is not consistent between individuals and this can be further pronounced after surgical interventions (e.g. decompressive craniectomy).^7,48 Consequently, the depth of the photon pathway differs between individuals.

One of the advantages of using the method described in section ‘Distribution of time of flight of photons’ with TD-NIRS instruments is that the depth of the photon pathway does not have to be assumed as it must be with CW- and FD-NIRS.^21,44 This is because the latest photons have traveled via the deepest layers, which shape the characteristics of the DTOF.^45,46,50

High-resolution diffuse optical tomography

Habermehl et al.⁸⁸ plotted the optical signal obtained from a CW-NIRS high-resolution diffuse optical tomography on an MRI mash to record the ICG passage on four healthy volunteers. The resultant brain 3D-reconstruction facilitated the separation of the ICG passage in the two layers.

It should be noted that by incorporating optical data into a structural image, the variables related to the ECT thickness can be measured, and the depth of the photon pathway estimated.

Indocyanine green’s clearance

Abnormalities in the liver perfusion or liver function can affect the rate of the ICG clearance from the blood.⁸⁹ Since liver injuries are relatively common in trauma patients, the rate and the amount of ICG leakage into the interstitial tissue in cases of BBB damage can be confounded by different rates of ICG clearance across TBI patients, or in the same patient at different time-points, independent of the status of the BBB.

Aims for future studies

The assessment of the $\mathit{CBF}$ or the BBB integrity in TBI patients would not give a comprehensive picture of the status of the brain after injury because this is compounded by multiple completely or partially unrelated elements, such as energy crisis, neuroinflammation, level of oxygen extraction, axonal damage, etc.^90–94 Furthermore, the evolution of these elements would not depend on the $\mathit{CBF}$ or the BBB integrity.^95–98 Therefore, an analysis of the $\mathit{CBF}$ and BBB integrity would still miss significant physiological elements in TBI patients. It is unlikely that, in clinical practice, it could be used alone or be suitable for all clinical needs and for the entire wide spectrum of brain lesions. This highlights the importance that clinical studies that use contrast-enhanced NIRS on TBI patients aim to be comprehensive of multimodal techniques (e.g. CT, microdialysis, etc.) and to address specific clinical needs where the understanding of the $\mathit{CBF}$ and BBB integrity is relevant.

Appropriate clinical decisions are not necessarily based on analysis of absolute values. For example, a comparison between the outcomes derived from clinical decisions in brain trauma care based on absolute parameters of intracranial pressure (ICP) monitored through a bolt, and on an ICP assessment using serial CT scans and neurological testing, showed no significant differences.⁹⁹ Similarly to the ICP values, the transformation of the optical values obtained from the ICG passage into absolute parameters of $\mathit{CBF}$ and BBB integrity may not be needed and a more broad analysis could be sufficient. For instance, in a case report of a patient with vasospasm following subarachnoid hemorrhage, a possible scenario after brain trauma, Keller et al.¹⁰⁰ reported changes in the shape of the ICG-inflow curves depending on the treatment applied. Similar analyses could be adopted in other clinical scenarios in order to assess response to treatments or to monitor the trend of the patient’s status without transforming these changes into absolute parameters.

Conclusion

In conclusion, contrast-enhanced NIRS with ICG could be used to analyze the brain perfusion and the BBB in TBI patients. However, the translation of the optical signal acquired into quantitative values must account for the clinical variables present in TBI patients. Different solutions have been used to address these problems, either partially or totally. Further research is needed to implement these methods in clinical practice on trauma patients. Clinical studies with contrast-enhanced NIRS should be designed to meet specific current clinical needs and should consider the possibility of addressing them by measuring relative optical and physiological parameters rather than absolute values, as well as combining the results obtained with other neuromonitoring techniques.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This article presents independent research funded by the project “Brain Injury and Trauma Monitoring Using Advanced Photonics”, financed by the European Union Horizon 2020 Research and Innovation Program, under grant agreement 675332, through the National Institute for Health Research Surgical Reconstruction and Microbiology Research Centre (NIHR SRMRC), partnership between University Hospitals Birmingham NHS Foundation Trust, the University of Birmingham, and the Royal Centre for Defence Medicine. The views expressed are those of the authors and not necessarily those of the NHS, the NIHR or the Department of Health.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

ORCID iDs

Mario Forcione https://orcid.org/0000-0002-4045-8011

Piotr Sawosz https://orcid.org/0000-0002-0697-3036