Introduction

The cerebellum is known to be engaged in motor control, cognition and memory. It is well connected to the cerebrum largely through the cerebellar peduncles, via tracts traversing the brainstem. Anatomically, the cerebellum consists of anterior, posterior and flocculonodular lobes. However, functional divisions may be more relevant for clinical purposes. While the vestibulocerebellum (flocculonodular lobe) controls balance, posture and ocular movements, the spinocerebellum (vermis) is engaged in truncal and gait posturing. Importantly for our study, the cerebrocerebellum (lateral hemispheres) are involved in fine motor planning, learning and controlling the limbs. Lesions in this region can result in dysdiadochokinesia and dysmetria elicited during clinical examination1,2.

Functional assessment of cerebellar activity has been studied using fMRI, PET, transcranial magnetic stimulation (TMS) and electrical stimulation. In particular, fMRI possess the disadvantages of motion artifacts, while PET necessitates the administration of radio-isotopes within a confined space. The latter 2 techniques may provide complementary information on neurovascular coupling during motor tasks which cannot be performed simultaneously in view of these limitations.

In contrast, near-infrared spectroscopy (NIRS) is a non-invasive functional imaging modality of ascertaining cerebral oxyhemoglobin (OxyHb) and deoxyhemoglobin (deOxyHb) concentration changes in real time with little interference by motion artefacts. Light in the near-infrared range of 600 to 900 nm is capable of penetrating scalp tissue and scattered by OxyHb and deOxyHb molecules present in the blood compartment. Resultant changes in their concentrations along the optical path of photons will be detected to reflect neurovascular coupling1,2,3. Real time motor activation with NIRS measurements as a form of cerebral oximetry may provide temporally relevant information to complement other neuroimaging methods.

Recent evidence on functional neuroanatomy of the cerebellum have suggested that it is more complex than previously thought4, and appears to be organized around a series of parasagittal stripes that partition underlying circuitry into zones anchored by Purkinje cells characterized by a laminated geometrical structure. Here, NIRS can provide novel data relating to neurovascular coupling of the cerebellar cortex while performing motor activation.

While NIRS has been utilized successfully in experiments involving the motor cortex and other brain regions5, some challenges remain with cerebellar recordings. This may be due to depth of the cerebellum relative to occipital cortex, low position and small volume. It also remains uncertain if cerebral oximetry is sufficiently robust to isolate cerebellar activity from that of the occipital, parietal and motor areas. Hence, most NIRS studies in conjunction motor activation tasks have focused on frontal lobe recordings to date.

Previously, attempts to measure cerebellar activity have been described6,7,8. These studies are differentiated from our study by using smaller sample sizes, different recording technique, equipment or detector locations and activating tasks. None had utilized validated tasks employed in bedside clinical testing of cerebellar function. Here, we ascertain the efficacy of NIRS in cerebellar function during standardized clinical motor tasks in healthy subjects by employing 2 novel protocols. Secondarily, we determine if our methods effectively evaluate cerebellar apart from corticospinal motor activity for each task.

Methods

All participating subjects were seated comfortably with eyes closed in a dark soundless room. The only audible input is the verbal cue to start and stop performing each task. Both hands were gently supported in the extended position. Prior subject briefing was conducted on the task and sequences before the actual study. The study has been approved by the hospital institutional review board, and informed consent has been obtained from all participants. All research methods were performed in accordance with relevant guidelines and regulations according to the Declaration of Helsinki.

For Task 1, subjects were instructed to perform alternating pronation and supination movements with one hand clapping onto the inactive opposite hand. This action is well-known to access diadochokinesia clinically, the ability to perform rapid alternating synchronous muscle movements in which the cerebellum is crucially engaged in9. Conversely, dysdiadochokinesia, the inability to perform such actions, is a sign of cerebellar dysfunction.

For Task 2, finger thumb opposition starting from the index to the little finger and repeated for the period of the experimental block was performed as fast as possible10. This action has been shown to be mediated by the corticospinal system, up to the level of the sensorimotor cortex using fMRI tractography11.

After an initial 2-min rest for baseline stabilization, each performed the tasks at maximal velocity with simultaneous fNIRS recording for 3 min, followed by a 3-min rest. Subjects were then cued to perform the motor task on the opposite hand. The cycle was repeated 3 times in total for each side.

fNIRS recordings obtained from a NIRSPORT system (NIRx Medical Technology, New York, USA). Subjects wore a cap with LED source and detector probes embedded, such that a 2–3 cm distance was maintained between 2 optodes, using the international 10–20 EEG system as a reference. The LED source consisted of 2 wavelengths (760 nm and 850 nm). In accordance to the modified Beer-Lambert law5 equation, data comprising OxyHb and deOxyHb were obtained.

Protocol 1 (Fig. 3) consisted of 5 optodes (4 source, 1 detector, 4 recording channels) spaced 3 cm apart and comprising single-tipped LED light sources emitting light simultaneously in conjunction with single-tipped fiber-optic detectors. This configuration will correspond to 4 recording channels. Protocol 2 (Fig. 3) consisted of 10 optodes (2 source, 8 detector, 8 recording channels) spaced 2 cm apart and comprising single-tipped LED light sources in conjunction with dual-tipped fiber-optic detectors. This configuration will correspond to 8 recording channels. Optodes are embedded in a firm-fitting cap to ensure optimal scalp contact.

We measured mean values of baseline to peak changes for OxyHb and deOxyHb in real time during each task. In addition, mean values of time to peak concentration were also analyzed for OxyHb and deOxyHb. The baseline to peak changes and time to peak analyses were recorded bilaterally regardless of the hand performing each task. For deOxyHb measurements, the peak was ascertained as the value of the time whereby maximum reduction was observed, as opposed to maximum increase for OxyHb. Stable recordings were defined as a waveform following a rest period showing a distinct onset, maximum or minimum peaks, all within the recording period in every channel (Fig. 4). To that end, baseline to peak values reflected averaged values of differences at the end of the baseline stabilization phase to the maximum value during the motor task. Time to peak value reflected averaged values of the time difference from the end of the stabilization phase to the time when the peak value is achieved during the motor task.

Each subject completed Tasks 1 and 2 beginning with one hand, followed by the opposite. In Protocol 1, the configuration corresponds to 4 channels, and data values were averaged across the channels. In Protocol 2, the configuration corresponds to 8 channels, and data values were averaged across the channels. NIRS recordings were repeated using Protocols 1 and 2 separately. Data was analyzed with SPSS Version 26 (Chicago, IL, USA), using the Mann–Whitney U test to compare mean values of subjects. P value of < 0.05 denoted statistical significance. Time units were in seconds (s) and NIRS measurements in microM (mM).

Table 1 shows data values for OxyHb & deOxyHb amplitude (m/M) for Protocol 1.

Table 1 Data values for OxyHb & deOxyHb Amplitude (m/M) (Protocol 1).
Full size table

Table 2 shows data values of time to peak (s) for Protocol 1.

Table 2 Data values of time to peak (s) (Protocol 1).
Full size table

Table 3 shows data values for OxyHb & deOxyHb amplitude (m/M) for Protocol 2.

Table 3 Data values for OxyHb & deOxyHb Amplitude (m/M) (Protocol 2).
Full size table

Table 4 shows data values of time to peak (s) for Protocol 2.

Table 4 Data values of time to peak (s) (Protocol 2).
Full size table

Figures 1 depicts results graphically Hb amplitude values for Protocol 1. Shaded bars represent Task 1 and unshaded bars represent Task 2.

Fig. 1
figure 1

Task 1 performed on the right and left hands did not result in significantly increased OxyHb amplitude recordings compared to Task 2. Task 1 performed on the right and left hands also did not result in significantly reduced deOxyHb amplitude recordings compared to Task 2. Shaded bars represent Task 1 (cerebellar) and unshaded bars represent Task 2 (Corticospinal).

Full size image

Figure 2 depicts results graphically Hb amplitude values for Protocol 2. Shaded bars represent Task 1 and unshaded bars represent Task 2.

Fig. 2
figure 2

Task 1 performed on the right and left hands showed significantly increased OxyHb amplitude recordings compared to Task 2. Task 1 performed on the right and left hands showed significantly reduced deOxyHb amplitude recordings compared to Task 2. Shaded bars represent Task 1 (cerebellar) and unshaded bars represent Task 2 (corticospinal).

Full size image

Figure 3 illustrates the source (S) and detector (D) positions for Protocols 1 and 2 respectively. Arrow depicts position of inion corresponding to the horizontal joined by top of both ears.

Fig. 3
figure 3

Protocol 1 consists of 5 optodes (4 sourcss, 1 detector) spaced 3 cm apart and comprising single-tipped LED light sources in conjunction with single-tipped fiber-optic detectors. This corresponds to 4 recording channels. Protocol 2 consists of 10 optodes (2 sources, 8 detectors) spaced 2 cm apart and comprising single-tipped LED light sources in conjunction with dual-tipped fiber-optic detectors. This corresponds to 8 recording channels. Single (Protocol 1) and double-tipped (Protocol 2) detectors are shown below Optodes are embedded in a firm-fitting cap to ensure optimal scalp contact.

Full size image

Figure 4 shows actual NIRS recordings obtained for OxyHb (top row) and deOxyHb (bottom row) while performing Task 1 using Protocol 2. Distinct onset (vertical line) to maximum increase (OxyHb) or reduction (deOxyHb) depicted as peaks are illustrated.

Fig. 4
figure 4

Actual NIRS recordings obtained for OxyHb (top row) and deOxyHb (bottom row) while performing Task 1 using Protocol 2. Distinct onset (vertical line) to maximum increase (OxyHb) or reduction (deOxyHb), represented as the peak of earliest onset and largest amplitude (arrows) are illustrated.

Full size image

Results

Qualitative measurements

For Protocol 1, stable recordings were obtained in 7 of 14 (50%) subjects.

For Protocol 2, all 14 subjects (100%) achieved stable recordings.

Quantitatively measurements

Protocol 1

Amplitude

Task 1 performed on the right and left hands did not result in significantly increased OxyHb amplitude (Right: Z = 1.92713, p = 0.0674; Left: Z = 1.02481, p = 0.30772) recordings compared to Task 2.

Task 1 performed on the right and left hands also did not result in significantly reduced deOxyHb amplitude (Right: Z = 0.00819, p = 0.099202; Left: Z = 1.23665, p = 0.33232) recordings compared to Task 2 (Table1)(Fig. 1)(Fig. 4).

Time to peak

Task 1 performed on the right and left hands did not result in significantly different time to peak for OxyHb amplitude (Right: Z = 0.22942, p = 0.8181; Left: Z = − 1.2454, p = 0.2113) recordings compared to Task 2.

Task 1 performed on the right and left hands also did not result in significantly different time to peak for deOxyHb amplitude (Right: Z = 1.54036, p = 0.12356; Left Z = − 1.42565, p = 0.15272) recordings compared to Task 2 (Table 2).

Protocol 2

Amplitude

Task 1 performed on the right and left hands showed significantly increased OxyHb amplitude (Right: Z = − 4.0478, p < 0.0001; Left: Z = 2.77072, p = 0.0056) recordings compared to Task 2.

Task 1 performed on the right and left hands showed significantly reduced deOxyHb amplitude (Right: Z = 1.99716, p = 0.040552; Left Z = 8.23665, p < 0.00001) recordings compared to Task 2 (Table 3, Fig. 2).

Time to peak

Task 1 performed on the right and left hands did not result in significantly changed time to peak for OxyHb amplitude (Right: Z = 0.38007, p = 0.70394; Lef: Z = − 1.20075, p = 0.23014) recordings compared to Task 2.

Task 1 performed on the right and left hands also did not result in significantly changed time to peak for deOxyHb amplitude (Right: Z = − 0.98588, p = 0.32218; Left: Z = − 0.59722, p = 0.5485) recordings compared to Task 2 (Table 4).

Protocol 1 vs Protocol 2

Overall, comparisons between Protocol 2 and 1 did not show any significant differences for time to peak amplitude (p > 0.05 for all) in corresponding tasks and sides individually. This is apparent for OxyHb and deOxyHb recordings.

Laterality of peak amplitude

For Task 1, we compared peak amplitude of right and left-sided optodes while performing using the right hand, for both OxyHb and deOxyHb. This was repeated using the left hand. None of the side-to-side comparisons were statistically significant (p > 0.05 for all).

Discussion

Our study has demonstrated the enhanced feasibility and superiority of NIRS for evaluating neurovascular coupling in the cerebellum during a cerebellum-activating motor task, both qualitatively and quantitatively, using Protocol 2 compared to Protocol 1. Apart from increased peak OxyHb, reduced peak deOxyHb was observed during Task 1.

Stable and robust recordings for Protocol 2 compared to Protocol 1 can be attributed to several reasons. As scalp hair is a major factor affecting NIRS recordings, the dual-tipped sensors used in Protocol 2 comprising small diameter tips within spring-loaded holders, separated by a gap enabled better hair separation. In addition, the two times increased overall tip surface areas of the active sensors in Protocol 2 compared to the single tip design used in Protocol 1 improved OxyHb and deOxyHb amplitude changes recorded, as well as in stability of recordings. Protocol 1 has a central detector (D1) corresponding to the cerebellar vermis region and 4 separate sources (S3 to S6). This 4-channel configuration may capture less neurovascular activity compared to Protocol 2 comprising 8 channels of 8 detectors (D1 to D8) and 2 sources (S1 and S2) located on bilateral lateral cerebellar hemispheres. As cerebellar neurovascular activity was also found to be bilateral even when either hand was performing a task, detection of neurovascular activity may be more effective in Protocol 2 comprising a larger bilateral area coverage compared to Protocol 1.

We employed 2 clinically validated tasks intended to closely engage the cerebellar and corticospinal systems respectively within an environment which eliminated confounding inputs, including visual and sound stimuli. Task 1 is a well-recognized bedside maneuver to assess a fundamental cerebellar function, and correspondingly, Task 2, for corticospinal motor activity (Table 5).

Table 5 Experimental protocols for each subject.
Full size table

Previously, similar but not identical publications had attempted to examine neurovascular coupling with NIRS. Rocco et al.7 presented a single subject study to explore neurovascular coupling at left and right cerebellar positions in conjunction with a finger tapping task. This is seen in contrast to our study which employed validated Tasks 1 and 2. This may help explain their findings, which obtained 2 peaks in OxyHb during each task, but no corresponding change described for deOxyHb. The finger tapping task is a combination of forearm flexion with rhythmicity, which may only partially engage the lateral cerebellar hemispheres for motor function9.

Nishida et al.8 utilized NIRS in conjunction with a relatively complex bow tying task and measured OxyHb signals bilaterally at a level situated 3 cm below the inion. However, the complexity of the task did not eliminate visual attention and other cognitive activity, including attention, tactile and associative processes. No change was described for deOxyHb.

Walia et al.6, in a healthy human subject study, showed the feasibility of NIRS of the cerebellum in conjunction with transcranial AC stimulation at 4 Hz, which was postulated to facilitate cerebellar inhibitory activity. The optode positions appear to be similar to Protocol 1 in our study, but no motor activation tasks were utilized.

In NIRS, the normal hemodynamic response to increased metabolic demand from neuronal activity, such as activation of cerebellar circuitry in our study protocol, is a focal increase in blood inflow exceeding oxygen consumption8. This results in an increase of OxyHb, decrease of deOxyHb, or both, as a fundamental process of neurovascular coupling. Specifically, increased oxygen delivery may result from vasodilation or enhancement of local diffusion, and increased oxygen extraction may be resultant of increased tissue consumption12,13. Hence, increase in Oxy Hb with a concomitant decrease in deOxyHb levels is more representative of focal neurovascular coupling activity than a change only in either parameter.

Protocol 2 had utilized a higher number of closer spacing optodes on both sides, with each customized position comprising LED light sources in conjunction with dual-tipped fiber-optic detectors. This results in 8 recording channels, compared to 4 in Protocol 1. These improvisations may contribute to better results both qualitatively and quantitatively in cerebellar recordings14. Our optode configuration (Fig. 3) in Protocol 2 likely corresponds to recording activity from the lateral cerebellar hemispheres.

It is also important to note that no systematic evaluation employing NIRS similar to Protocols 1 and 2 in the cerebellar or posterior head regions have been published to our knowledge.

Regarding the overall aim of the study, it is well recognized that a cardinal challenge of NIRS recordings is a stable, effective and comfortable optical interface with the scalp in the presence of hair and hair follicles, which are strong attenuators, resulting in poor signal quality14. In the presence of thick hair, it can also be time-consuming to ensure that the minimum amount of hair is under each optode15. While NIRS recordings are more easily obtained in the frontal lobe regions, evaluation in the posterior head region of the cerebellum has not been systematically evaluated. Protocol 1 is a current reference standard in which Protocol 2 is tested against. Protocol 1 utilized older and more readily available equipment, in conjunction with an optode montage with fewer source and detector positions. Protocol 2 represents a significant innovation comprising increased number of dual-tipped LED detectors arranged in a more closely spaced configuration, resulting in superior detection of neurovascular coupling activity as shown in this study.

The lack difference for time to peak amplitude between Protocol 1 vs 2 validates that a similar trajectory of neurovascular coupling temporally, but more effective and discriminatory detection was obtained using Protocol 2.

In addition, lack of laterality in recordings during Task 1 may reflect bilaterality of cerebellar participation, but it can be further evaluated in future studies employing other combinations of source detector/optode configurations.

The cerebellum plays a crucial role in coordinating motor function, in which ensuring smoothness and accuracy is achieved by integration of sensory inputs from vestibular receptors and proprioceptors. In turn, it modulates commands to motor neurons to compensate for shifts in body position or changes in load upon muscles.

From the anatomical standpoint, mossy fibers convey sensory information centrally, and error signals are conveyed by the climbing fiber inputs, which become activated when an unexpected event occurs, such as when a greater load than expected is placed on a muscle. Divergence of input from the mossy fibers to the granule cells to the parallel fibers creates complex representations of the cerebellar sensory representation and the desired motor output. When predicted output is not achieved, climbing fibers signal this error and trigger a calcium spike in the Purkinje cell, resulting in influx of calcium which changes the connection strengths between parallel fibers and Purkinje cells, such that when the next desired action impulse is effected, the motor output will be modified to more closely approximate the desired output. Hence, the cerebellum acts as a feed forward controller of effective ballistic movements16.

In line with cerebellar control of movement coordination and error correction, we have utilized a validated task, whereby Task 1 has been utilized clinically for the bedside evaluation of cerebellar incoordination17. The cerebellum coordinates the function of agonist and antagonist movements required for specific alternating movements. This altered coordination forms the basis for dysdiadochokinesia in cerebellar injury or dysfunction18. The cerebellum also coordinates the acceleration and velocity of muscular activity. The clinical signs of cerebellar dysfunction include ataxic gait, dysarthria, intention tremor or dysmetria, ocular movement abnormalities and dysdiadokokinesia. Practically, upper limb motor function is least feasible with gait, speech or eye movements for our study. Dysdokokinesia testing consisting of alternating hand movements demands higher accuracy relative to evaluating dysmetria as a simple past-pointing test alone. Thus, we have employed a validated clinical test incorporating accuracy, alternation of movements, error correction and kinetic control of motor action. Further validation of our findings can be made in future studies correlating similar NIRS recordings with neuroimaging of subjects with cerebellar lesions.

In our study, Protocol 2 did not completely isolate cerebellar neurovascular coupling activity, but it had detected increased neurovascular coupling activity above and beyond any inherent corticospinal activity in the cerebellum when Task 1 was performed compared to when Task 2 was performed. In addition, stable recordings were obtained in 100% of patients using Protocol 2 compared to only 50% with Protocol 1. Hence, our study has demonstrated the enhanced feasibility and superiority of NIRS for evaluating neurovascular coupling in the cerebellum during a cerebellum-activating motor task, both qualitatively and quantitatively, using Protocol 2 compared to Protocol 1 (Table 6).

Table 6 Comparison of Protocol 1 and Protocol 2.
Full size table