Intra- and Inter-Day Reliability of the NIRS Portamon Device after Three Induced Muscle Ischemias

(1) Background: Near-infrared spectroscopy (NIRS) is an innovative and non-invasive technology used to investigate muscular oxygenation. The aim of this study is to assess the within- and between-session reliability of the NIRS Portamon (Artinis, Elst, Netherlands) device following three sets of induced muscle ischemia. (2) Methods: Depending on the experimental group (G1, G2 or G3), a cuff was inflated three times on the left upper arm to 50 mmHg (G1), systolic blood pressure (SBP) + 50 mmHg (G2) or 250 mmHg (G3). Maximum, minimum and reoxygenation rate values were assessed after each occlusion phase, using a Portamon device placed on the left brachioradialis. Reliability was assessed with intraclass correlation coefficient (ICC) value and ICC 95% confidence interval (CI-95%), coefficient of variation (CV) and standard error of measurement (SEM) (3) Results: Our results showed a good to excellent reliability for maximums and minimums within-session. However, the reoxygenation rate within sessions as well as measurements between sessions cannot predominantly show good reliability. (4) Conclusions: Multiple measurements of maximums and minimums within a single session appeared to be reliable which shows that only one measurement is necessary to assess these parameters. However, it is necessary to be cautious with a comparison of maximum, minimum and reoxygenation rate values between sessions.


Introduction
Near-infrared spectroscopy (NIRS) is an innovative and non-invasive technology used to investigate muscular oxygenation through several indicators. By using nearinfrared wavelengths (~700-900 nm), the light is able to penetrate biological tissues where the main absorbing chromophores in skeletal muscle are hemoglobin (Hb), myoglobin (Mb), and to a lesser extent, cytochrome oxidase (cytox) [1,2]. Thus, NIRS technology provides information on four variables, depending on the device used: oxy[Hb + Mb], deoxy[Hb + Mb], total[Hb + Mb] (i.e., the sum of oxy-and deoxy[Hb + Mb]) and tissue saturation [1]. The concept of recording muscle oxygenation light dates back to 1937 [3], and human skeletal oxygenation measurements to the end of the 1980s [4].
Due to their moderate cost, convenient small size and wireless connectivity, NIRS devices have been widely used at rest and during exercise [5][6][7][8][9]. Moreover, NIRS oximeters offer the advantage of acceptable signal-to-noise ratios, even during dynamic exercise [8].
Indeed, NIRS provides robust information on skeletal muscle oxidative capacity. NIRSderived parameters such as the muscle reoxygenation rate after exercise [5,10,11], the muscular consumption of O 2 (m . VO 2 ) [12], as well as the size of the post-occlusive reactive hyperemia (PORH) [13] may be used to assess performance, vascular reactivity and training status. PORH is characterized by an increase in blood flow following an arterial occlusion, representing the significance of limb reperfusion after ischemia [14,15]. This physiological Participants visited the laboratory twice within a 24 h to 72 h period. During the first visit, anthropometric measurements were recorded (stature, body mass, and adipose tissue thickness), while limb lengths were measured to ascertain NIRS device placement.

Study Design
Using an experimental design, participants were randomly assigned to one of three groups with different occlusion pressures: 50 mmHg (G1), SBP + 50 mmHg (G2) and 250 mmHg (G3) [28,29]. The protocol is described in Figure 1 and consists of a baseline period followed by 3 × 7 min occlusion phases interspersed with three reperfusion phases where the cuff was deflated. Participants were lying on a medical couch for the duration of the session in the dark and asked to avoid any movement which could disturb the signal. After a 15 min baseline period, the cuff placed on the left arm was inflated to induce three occlusion phases (O1, O2, O3) each of seven minutes. Each occlusion phase was followed by a reperfusion period (R1 = 10 min, R2 = 10 min, R3 = 20 min).
Participants visited the laboratory twice within a 24 h to 72 h period. During the first visit, anthropometric measurements were recorded (stature, body mass, and adipose tissue thickness), while limb lengths were measured to ascertain NIRS device placement.

Study Design
Using an experimental design, participants were randomly assigned to one of three groups with different occlusion pressures: 50 mmHg (G1), SBP + 50 mmHg (G2) and 250 mmHg (G3) [28,29]. The protocol is described in Figure 1 and consists of a baseline period followed by 3 × 7 min occlusion phases interspersed with three reperfusion phases where the cuff was deflated. Participants were lying on a medical couch for the duration of the session in the dark and asked to avoid any movement which could disturb the signal. After a 15 min baseline period, the cuff placed on the left arm was inflated to induce three occlusion phases (O1, O2, O3) each of seven minutes. Each occlusion phase was followed by a reperfusion period (R1 = 10 min, R2 = 10 min, R3 = 20 min).
Participants visited the laboratory twice within a 24 h to 72 h period. During the first visit, anthropometric measurements were recorded (stature, body mass, and adipose tissue thickness), while limb lengths were measured to ascertain NIRS device placement.

Study Design
Using an experimental design, participants were randomly assigned to one of three groups with different occlusion pressures: 50 mmHg (G1), SBP + 50 mmHg (G2) and 250 mmHg (G3) [28,29]. The protocol is described in Figure 1 and consists of a baseline period followed by 3 × 7 min occlusion phases interspersed with three reperfusion phases where the cuff was deflated. Participants were lying on a medical couch for the duration of the session in the dark and asked to avoid any movement which could disturb the signal. After a 15 min baseline period, the cuff placed on the left arm was inflated to induce three occlusion phases (O1, O2, O3) each of seven minutes. Each occlusion phase was followed by a reperfusion period (R1 = 10 min, R2 = 10 min, R3 = 20 min).
Participants visited the laboratory twice within a 24 h to 72 h period. During the first visit, anthropometric measurements were recorded (stature, body mass, and adipose tissue thickness), while limb lengths were measured to ascertain NIRS device placement.

Study Design
Using an experimental design, participants were randomly assigned to one of three groups with different occlusion pressures: 50 mmHg (G1), SBP + 50 mmHg (G2) and 250 mmHg (G3) [28,29]. The protocol is described in Figure 1 and consists of a baseline period followed by 3 × 7 min occlusion phases interspersed with three reperfusion phases where the cuff was deflated.
Participants visited the laboratory twice within a 24 h to 72 h period. During the first visit, anthropometric measurements were recorded (stature, body mass, and adipose tissue thickness), while limb lengths were measured to ascertain NIRS device placement.

Study Design
Using an experimental design, participants were randomly assigned to one of three groups with different occlusion pressures: 50 mmHg (G1), SBP + 50 mmHg (G2) and 250 mmHg (G3) [28,29]. The protocol is described in Figure 1 and consists of a baseline period followed by 3 × 7 min occlusion phases interspersed with three reperfusion phases where the cuff was deflated. Participants were lying on a medical couch for the duration of the session in the dark and asked to avoid any movement which could disturb the signal. After a 15 min baseline period, the cuff placed on the left arm was inflated to induce three occlusion phases (O1, O2, O3) each of seven minutes. Each occlusion phase was followed by a reperfusion period (R1 = 10 min, R2 = 10 min, R3 = 20 min). Participants were lying on a medical couch for the duration of the session in the dark and asked to avoid any movement which could disturb the signal. After a 15 min baseline period, the cuff placed on the left arm was inflated to induce three occlusion phases (O1, O2, O3) each of seven minutes. Each occlusion phase was followed by a reperfusion period (R1 = 10 min, R2 = 10 min, R3 = 20 min).  (13.7 ± 6.6 mm) was also assessed before placing the NIRS instrument to avoid disturbance linked to adipose tissue thickness.

Ankle-Brachial Index (ABI)
The ABI was calculated as the ratio of the highest SBP value of posterior and dorsal tibial arteries and the highest SBP value of the brachial artery. Values are reported in Table 1. The assessment order was the following: right brachial artery; right tibial posterior and anterior arteries; left tibial posterior and anterior arteries; left brachial artery; right brachial artery [30]. SBP was assessed with a blood pressure monitor (Easy 3, Hol-tex+, Aix-en-Provence, France) and a manual stethoscope (Classic III, 3M Littman Stethoscopes, Maplewood, MN, USA) for the first measurement of the right brachial artery and with a mini-Doppler (Sonotrax Lite, Edan Instruments Inc., Shenzhen, China) for subsequent measures.

Near-Infrared Spectroscopy (NIRS)
A wireless NIRS device (PortaMon, Artinis, Elst, The Netherlands), connected with Bluetooth with a sampling rate of 10 Hz, was used on the left arm. This was dualwavelength (760 and 850 nm), with three pairs of LEDs spaced 30, 35, and 40 mm from the receiving continuous-wave NIRS system using the modified Lambert-Beer law. It calculates the absolute concentration of tissue oxy-, deoxy-and total hemoglobin (O 2 Hb, HHb, tHb, respectively). Tissue saturation index (TSI), expressed in % and reflecting the dynamic balance between O 2 supply and consumption, was calculated as stated in Equation (1): On the arm, the NIRS device was placed on the brachioradialis muscle, at two-thirds on the line from the styloid process to the central point between the lateral and medial epicondyles ( Figure 2) in order to make device placement consistent for all participants [31]. The device was adhered to the limb with double-sided auto-adhesive tape (Coheban, 3M, Cergy-Pontoise, France) and wrapped in black cloth and an elastic bandage to prevent any disturbance due to light interference or unintentional movement.

Data Analysis
NIRS data were acquired at 10 Hz, and the signal was smoothed using a 10th-order low-pass zero-phase Butterworth filter (cut-off frequency 0.8 Hz) using Pandas software library functions for Python (Python 3.8.8, Python Software Foundation, The device was adhered to the limb with double-sided auto-adhesive tape (Coheban, 3M, Cergy-Pontoise, France) and wrapped in black cloth and an elastic bandage to prevent any disturbance due to light interference or unintentional movement.

Data Analysis
NIRS data were acquired at 10 Hz, and the signal was smoothed using a 10th-order lowpass zero-phase Butterworth filter (cut-off frequency 0.8 Hz) using Pandas software library functions for Python (Python 3.8.8, Python Software Foundation, https://www.python.org, accessed on 19 October 2021) [32]. To avoid any disturbance linked to the beginning of the protocol, TSI baseline , HHB baseline and O 2 HB baseline were calculated as the average of the last minute before the first occlusion.
The difference (∆TSI, ∆HHB and ∆O 2 HB) between the maximum reached corresponding to the hyperemia spike during the reoxygenation (for TSI max and O 2 HB max ) and the minimum reached during the occlusion (TSI min and O 2 HB min ) and inversely for HHB max and HHB min were calculated ( Figure 3).   and HBdiff between the start and the end of the reoxygenation curve, as well as the intercept and the coefficient of determination (r²) (Figure 3). The start, the end, and the highest velocity value (Vpeak) of the reoxygenation curve were calculated using automatic peak detection Python routine, with a 5% threshold for both start and end, applied to speed values. Speed values were obtained by derivative collected data and smoothed using a 10th-order low-pass zero-phase Butterworth filter (cut-off frequency 0.8 Hz) (Python 3.8.8).

Statistical Analysis
The three reperfusions per session are named R1, R2 and R3 for reperfusion 1, 2 and 3, respectively. Sessions 1 and 2 are named S1 and S2, respectively. Hemoglobin difference (HB diff ) was calculated as the difference between oxygenated hemoglobin (O 2 Hb) and deoxygenated hemoglobin (HHb).
Reoxygenation rate was calculated for each parameter (TSI reoxy_rate , HHB reoxy_rate , O 2 HB reoxy_rate and HB diff_reoxy_rate ) as the upslope (r 2 = 0.968 ± 0.016) of TSI, [HHb], [O 2 HB] and HB diff between the start and the end of the reoxygenation curve, as well as the intercept and the coefficient of determination (r 2 ) ( Figure 3). The start, the end, and the highest velocity value (Vpeak) of the reoxygenation curve were calculated using automatic peak detection Python routine, with a 5% threshold for both start and end, applied to speed values. Speed values were obtained by derivative collected data and smoothed using a 10th-order low-pass zero-phase Butterworth filter (cut-off frequency 0.8 Hz) (Python 3.8.8).

Statistical Analysis
The three reperfusions per session are named R1, R2 and R3 for reperfusion 1, 2 and 3, respectively. Sessions 1 and 2 are named S1 and S2, respectively.
Standard error of measurement (SEM), an indicator of absolute reliability, was calculated as the standard deviation (SD) multiplied by the root mean square (RMS) of the difference of 1 and the ICC [34].
Coefficient of variation (CV) was used to assess the variability across multiple repeated measures. Within-participant CV was calculated as the mean of CV calculated over the three trials for each participant. Three CV between sessions were calculated for each participant for R1S1-R1S2, R2S1-R2S2 and R3S1-R3S2, respectively. A mean CV for each participant was then calculated, and the CV of each participant was averaged to obtain a mean between-session CV per group. Finally, a CV between participants was calculated for each trial and averaged.

Participants' Characteristics and Environmental Conditions
Non-significant differences were observed across all variables of participants' characteristics. Comparisons of temperature and humidity level between sessions showed a significant difference between means only for humidity level (t(30) = −2.580; p = 0.015; ES = −0.463). Participant characteristics are reported in Table 1. Environmental conditions were monitored during each session and are reported in Table 2.

Maximum and Minimum Responses
Results of within-participant variability assessed with CV are reported in Table 3. Results showed a lower mean CV between trials of each session across all variables for maximums (2.58 ± 1.82%) compared with minimums (5.14 ± 3.81%). The lowest variability for maximums was reached by TSI% (1.23 ± 0.62%), whereas the lowest variability was reached by HHb (2.30 ± 0.77%) for minimums.
Reliability between measurements assessed with ICC for minimums and maximums are reported in Tables 4 and 5, respectively. For both minimum and maximum values, ICC [ICC [CI-95%] showed a significant p-value for all conditions and all parameters (p < 0.001). Good to excellent reliability (0.75 > ICC [CI-95%] > 1) was found in all conditions for maximums and minimums of G2 and G3, for TSI% and HHb only.

Maximums and Minimums
Results of maximum and minimum variability between sessions, assessed with CV, are reported in Table 9. When averaged across groups, CV of HHb (3.70 ± 2.95%) and TSI% (2.43 ± 2.48%) show the lowest variability for minimums and maximums, respectively. Averaged across all groups and variables, minimums and maximums showed a CV of 7.08 ± 7.17% and 4.30 ± 3.95%, respectively.  Tables 10 and 11. Results indicate a lack of reliability for both maximums and minimums between sessions, with the majority of CI-95% below the 0.5 threshold in all conditions. An example of the reliability of maximums both between-and within-session is provided by an analysis of the limits of agreement (Figure 4).

Reoxygenation Rate
Between-session variability of the reoxygenation rate parameters is reported in Table 12. For both slope and Vpeak, G1 showed CVs averaged across all variables of 41.60 ± 51.77% and 22.21 ± 32.79%, respectively, which are higher than the CVs reported for G2 (22.21 ± 32.79% and 13.07 ± 9.62%, respectively) and G3 (12.56 ± 16.40% and 8.67 ± 4.71%, respectively). The results of the reoxygenation rate analyses of reliability are reported in Tables 13 and 14. For TSI%, HHb and tHb, reliability could not elicit any reliable pairwise measurements of slope or Vpeak. For O 2 Hb Vpeak, only three conditions showed moderate reliability.

Discussion
The present study aims to assess the reliability of NIRS measurements of maximum, minimum and reoxygenation rate values following three occlusions. The main finding of this study is that NIRS assessment of maximums and minimums was highly reliable across trials of a single session but was not reliable on two separate occasions. Moreover, if the 95% CI of the ICC is reported, the slope of reoxygenation and Vpeak cannot be considered reliable parameters both within and between sessions.

Maximums and Minimums Reliability
Between sessions, our results are in agreement with the findings of Lacroix et al. (2012) for O 2 Hb and tHb maximums [35]. Their study focused on forearm oxygenation response to occlusion. However, they had a unique 5 min occlusion at 100 mmHg over the SBP (mean: 219 ± 7 mmHg), close to our G3 pressure (250 mmHg). Lacroix et al. (2012), found an ICC of 0.63 and 0.31, whereas we reported an ICC of 0.53 and 0.40 for O 2 Hb and tHb maximums, respectively (Table 11). Unfortunately, 95% CI of ICC was not reported and the O 2 Hb maximum was considered as good, while our result reported with CIs is considered as not reliable (ICC = 0.53 [0.07, 0.81]). Additionally, the authors also reported similar between-session CVs for O 2 Hb (6.68%) compared to 5.10% in the current study.
Our findings indicate lower results for minimums and for maximums of TSI in G3 (25.57 ± 9.96% and 73.48 ± 4.89%, respectively), compared to de Oliveira et al. (2021), who found a TSI of 44.00 ± 10.39% and 79.98 ± 5.11% for the minimum and maximum, respectively. Indeed, the authors assessed the StO 2 on a similar population of young healthy adults, utilizing the same equipment (PortaMon, Artinis, Elst, Netherlands) and occlusion pressure (250 mmHg) but on a different forearm muscle (flexor carpi radialis) and with a 5 min occlusion period. These differences, particularly for TSI min responses, could be attributed to the 2 min occlusion difference between studies since the duration of occlusion induces differences in microcirculatory responses [36] and significant differences in maximums and minimums of TSI% [27]. It may also result from the spatial heterogeneity of tissue responses, which makes comparison between different muscles difficult [1,37]. Moreover, differences between findings from this study for minimums recorded at brachioradialis and those at the lower limb after 5 min of occlusion at 250 mmHg (46.2 ± 7.5%) [24] could be further explained by limb-specific variation in arterial function [38].
To our knowledge, our study is the first to investigate the intra-day and day-to-day pairwise reliability of NIRS maximums and minimums after three arterial occlusions. Since the minimum and maximum values directly influence the calculation of other parameters, such as the slope of reoxygenation or the amplitude [39,40] or the maximal physiological range [23], and have an effect on the results of index calculation, such as the area under the curve (AUC), it is essential to know the reliability of those parameters. Thus, these should be ascertained both between successive measurements within a single session and also between separate occasions, such as pre-and post-intervention. Our findings highlighted the highly acceptable reliability after arterial occlusions of maximums and minimums for intra-day measurements but suggest caution when comparing values between sessions.
When compared to results found in trained participants (2.83 ± 0.48%·s −1 ) [40], our results seem to be much lower and inconsistent with previous findings, despite similar population demographics and characteristics. However, in our study, walking was considered a physical activity, contrary to the work of McLay et al. (2016). Thus, it is noteworthy that our population elicits results closer to a reperfusion slope for an untrained group (1.26 ± 0.28%·s −1 ) than for a trained group [40].
The reliability of the reperfusion slope of NIRS-derived parameters has not been widely investigated in the literature.  reported an excellent intraday ICC (0.92) with a CV of 9 ± 4% [24]. We reported ICC values of 0.91, 0.82 and 0.92 for pairwise measurements between the three occlusions in session 2. However, we reported a higher CV of 27.59 ± 13.01%. Interestingly, between-session slope measurements have been reported to be highly reliable with ICCs of 0.88 [27] and 0.94 [24] between sessions. Those findings are high when compared to our results of 0.43, 0.43 and 0.62. Moreover, we are unable to ratify a level of good reliability between sessions for our measurements, since the 95% CI reported all elicit a lower value under the threshold of 0.50. In this way, we proffer cautiousness with a comparison of reperfusion slope measurements, particularly between sessions.
The higher variability found for intra-day reoxygenation rate measurement may be explained by the extended durations of the successive occlusion and reperfusion phases, leading to a 76 min long protocol. This duration and a resting state may induce changes in skin temperature or in blood pressure levels [43], which could affect the results.
Our findings showed a dissonance with results in the literature regarding inter-day reliability. This dissonance may be due to room temperature changes between two sessions for the same participant. Even if no significant difference was found between S1 and S2, the temperature may vary between both sessions, inducing changes in both blood pressure level and muscle oxygenation kinetics [44,45]. Another explanation may result from the blood volume changes during successive occlusions and across each session [46]. Indeed, during ischemia, the muscle blood flow increases, which affects the quantity of oxy-and deoxyhemoglobin measured with the NIRS device. Thus, these blood volume changes may bring confusion in slope measurements, and some researchers suggest correcting the NIRS signal with blood volume changes [12]. However, deoxyhemoglobin is supposed to be less sensitive to blood volume changes [47], but our findings showed that HHb was not more reliable between sessions than other parameters.
This discrepancy between our findings and that of the literature is emphasized by the way in which ICCs are reported in reliability studies focusing on NIRS-derived parameters. Following recent guidelines on selecting and reporting ICCs [25], we indicated reliability according to the lower bound value of 95% CI. However, the level of reliability ("poor", "moderate", "good" and "excellent") seems to differ among guidelines [48]. In the present study, 50% of within-session and no between-session measurements of reperfusion slope for TSI% are considered reliable according to the guidelines [25]. However, if ICC was reported without the 95% CI, 89% of the within-session and 22% of the between-session measurements would have been reported as reliable. This huge difference, particularly for within-session measurements, highlights the significance of reporting the 95% CI in reliability studies.

The Reliability When Assessing Tissue Oxygenation
NIRS technology is also used to investigate cerebral oxygenation (fNIRS). In healthy participants, fNIRS recordings of changes in cerebral oxygenation indicate good reliability (ICC = 0.83) [49]. Initial drop amplitude recorded during an orthostatic test showed good reliability for O 2 Hb (ICC = 0.83) and for TSI% (ICC = 0.99) [50]. Once again, confidence intervals are not reported in these studies. Moreover, a good to excellent test-retest reliability is found for O 2 Hb and HHb in response to postural changes [51].

Limitations
The main limitation of this study may be the inability to control the room temperature, as microcirculation is sensitive to skin temperature [54]. However, no significant differences between groups or sessions have been found.

Conclusions
In conclusion, the present study demonstrated that maximum and minimum values reached after 3 periods of 7 min occlusions are highly reliable within a single session, thus it may not be necessary to perform multiple occlusions to obtain reliable values of maximums and minimums. However, NIRS-derived parameters such as the slope of reperfusion or Vpeak during reperfusion cannot show reliable reliability when ICC is reported with 95% CI. It is thus important to be cautious when comparing values between sessions, particularly when comparing pre-and post-intervention values.

Institutional Review Board Statement:
The study was conducted in accordance with the Declaration of Helsinki of 1975, revised in 2013. The protocol was evaluated and approved by the local Ethics Committee of Paris-Saclay University (Comité d'éthique pour la Recherche de l'Université Paris-Saclay) (reference: CER-Paris-Saclay-2020-006, approved the 9 December 2020).
Informed Consent Statement: All subjects gave their informed consent for inclusion before they participated in the study.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author. The data are not publicly available due to ethical restrictions.