1.

Introduction

Functional near-infrared spectroscopy (fNIRS) is a noninvasive neuroimaging technique hinging on the properties of light in the red and near-infrared wavelength range to investigate hemodynamic activity occurring in the brain,¹ enabling researchers to detect changes in oxy- (HbO) and deoxy-hemoglobin (HbR) concentration. When multiple sources and detectors of near-infrared light are employed, creating multidistance channels, the technique is usually referred to as diffuse optical tomography (DOT).^2,3 Depth-resolved images of HbO/HbR concentration changes in the brain can be effectively reconstructed with DOT.⁴

Optical imaging is particularly suitable for the (term and preterm) neonatal population, because of its noninvasive nature, portability—it can be used in the neonatal intensive care unit (NICU) at the cradle-side—and technical compatibility with other clinical recording systems.^5–12 For instance, it can be employed in multimodal acquisitions with electroencephalography (EEG), which is used for seizure detection in clinical practice. So far, DOT studies in neonates have monitored hemodynamic changes in the whole brain up to a maximum of 2 h.^6,13–15 Prolonging this “monitoring window” would, however, increase the chance to document important intracranial events in the neonatal population, such as seizures or intracranial hemorrhages.¹⁶

The reasons that could explain the limited monitoring time lie primarily in (1) the discomfort of the optode holder for the baby and the skin damage that it can possibly produce and (2) a possible interference of the cap with clinical procedures and nursing. This is particularly true for DOT acquisitions, which require several fiber bundles and the placement of optodes over the whole head. This may force the newborn to a strained position and may cause skin lesions or pressure sores, especially in very preterm infants.

To this end, we propose a low-cost, easy-to-dress custom-made cap with whole-head coverage for DOT and 10 electrode positions for EEG, designed for continuous DOT monitoring in the very preterm neonates. We tested the cap on a very preterm newborn in NICU with a 7-day continuous DOT monitoring in order to validate its effectiveness.

2.

Materials and Methods

2.1.

Head Cap Manufacturing Process

The hand-made neonatal cap (hereafter NeoCAP) was designed by merging two components: an outer layer consisting of a soft tissue cap (EasyCap, Germany) marked with the $10-10$ reference points [Fig. 1(a)] and an inner layer made of foam flexible polyurethane [Fig. 1(b)], a standard material employed for prevention of skin bedsores in the presence of prolonged forced immobilization and previously employed in preterms.^17–19 Supports for optodes were created with truncated, hand-made, plastic cones, whose diameter was adapted to fit the optodes. Electrodes were inserted into white disks located in the outer layer. Holes were created in the inner layer aligned with optodes and electrodes, and the two layers were finally sewed together. With this solution, neither the plastic supports for optodes nor that for electrodes were in direct contact with the inner surface [Figs. 1(a) and 1(b)], so as to avoid the friction of the apparatus with the skin of the baby’s head.

Fig. 1

(a) The original outer layer, with the $10-10$ reference points marked on it (left) and the outer layer with all optodes’ support and electrodes’ support. (b) The inner layer created to match the dimension of the baby’s head. (c) Outer and inner layers were sewed together to create the cap (left). The holes made in the inner layer aligned with the optodes’ position and electrodes’ position of the outer layer can be clearly seen. On the right, the cap positioned on the neonate’s head, with all fibers connected to it. (d) Configuration of optodes and electrodes. The $10-5$ locations, where optodes/electrodes were positioned, are displayed in red rectangles for sources, light blue ellipses for detectors, and black circles for electrodes. Green lines connecting sources and detectors show the 105 DOT channels.

The cap was designed of a single size (26 cm circumference) corresponding to the 50th percentile of head circumference of neonates at 28 weeks gestation.

Electrodes were placed in F3, Fz, F4, C3, C4, P3, P4, and FT10 according to the international $10-5$ system.²⁰ The ground electrode was placed on the inion and the reference electrode on FT9. Optodes were positioned in a subset of locations of the international $10-5$ system [see Fig. 1(c)].

3.

Results

All the PIPP scale values were below 5 over the whole study period, indicating no pain associated with NeoCAP.²¹ No skin lesions were recorded during the observation period (Fig. 2). Accidental dislocation of optodes occurred four times over 7 days with prompt rearrangement by the attending nurse.

Fig. 2

Pictures of the neonate’s head taken after removal of the cap toward the end of the 7-day monitoring period. Neither skin lesions nor evident pressure marks due to the cap can be observed.

The neonate was successfully monitored with DOT for 122.6 h over the 7-day monitoring period (73% of the total duration).

Figure 3(a) shows a box-plot of the percentage of reliable channels for each DOT segment. Figure 3(b) shows three examples of spectrograms computed on data acquired at different days. The presence of the heartbeat component is clear as well as the variability of the frequency of the heartbeat over time, a well-established and distinctive feature in the preterm population. The mean cardiac frequency was $2.4\pm .2\mathrm{Hz}$. The absence of constant drifts was attested by a mean slope angle of $.03\mathrm{deg}\pm .1\mathrm{deg}$, which can be considered a negligible deviation from the expected 0 deg.

Fig. 3

(a) Box-plot showing the % of reliable channels during all data segments. The median number of reliable channels across data segment was 92.38%, with only three segments of data having $<78.1\%$ of reliable channels. (b) Three representative spectrograms, showing the results of the time-frequency analysis. In the upper panel, the spectrogram depicts one segment of data ($\sim 377\min$, i.e., $\sim 6\mathrm{h}$ and a half) acquired during the first days, in the bottom panel on the left, the spectrogram depicts one segment of data ($\sim 98\min$) acquired halfway through the recording phase, and in the bottom panel on the right, the spectrogram depicts one segment of data ($\sim 73\min$) acquired toward the end of the 7-day period. The black ellipses highlight in each spectrogram the heartbeat component, which is consistently observable for the entire duration of the three time-series.

Figures 4 and 5 show an example of DOT signals of all channels in one of the segments of data and an example of EEG data, respectively.

Fig. 4

Example of DOT intensity data of all channels at both wavelengths (210 channels) in one of the data segment. No filtering was applied to the plotted data. In the bottom panel, a particular of the above time-series is displayed. The heartbeat component is appreciable in most of the channels.

Fig. 5

Example of EEG data. The data were offline re-referenced to the average of FT9 and FT10. A notch filter ($\text{cut-off frequency}=50\mathrm{Hz}$) was applied to remove the line noise. A band-pass filter ($\text{cut-off frequency}=0.5$ and 60 Hz) was further applied to remove higher frequency noise and low frequency drifts.

Approximately 17 min were identified in the EEG trace as featuring the tracé discontinu. During this time period, nine motion artifacts were identified in the DOT data, with most of these occurring only in a few channels.

4.

Discussion

DOT has an undisputed potential for investigating whole-brain hemodynamics in the neonatal population.^6,13,25–31 Prior studies in this field have raised the issue related to the cap tolerance by the neonates as a possible factor limiting the duration of DOT data acquisition. Singh et al.⁶ and Chalia et al.¹³ adopted a soft and flexible cap (EasyCap, Germany, i.e., our outer layer), with a whole-head coverage for a multimodal DOT-EEG acquisition. Ferradal and colleagues¹⁴ employed a custom-made optode holder, composed of five triangular patches joined together with a neoprene headband, fixed to the head via Velcro straps. A layer of flexible plastic was the structural support of the cap and a layer of soft silicone was used to embed the optical fibers. The optode array covered the occipital, temporal, and inferior parietal cortices. Austin et al.³² employed a custom-made helmet with whole-head coverage, made of an outer shell of thermoplastic and an inner part of NIR-absorbing foam. Crucially, in all these DOT studies, the length of registration did not exceeded 2 h, with a minimal length of 60 min^6,14 up to a maximum of 115 min¹³ and the gestational age ranged from 34 weeks gestation¹³ to term.

Here, we showed that the adoption of our hand-made cap (NeoCAP) enabled a safe continuous DOT monitoring for a 7-day period in a very preterm infant. Of import, DOT data quality was remarkable, as indicated by the three metrics used to assess the reliability of the measurements; the vast majority of the channels exhibited a high SNR with no noticeable drift, and the heartbeat was easily identifiable in most of the high SNR channels. A (reassuringly) small number of artifacts were identified while the neonate was in quiet sleep, according to EEG traces. Given, however, that identifying quiet sleep in very preterm neonates from an EEG trace is particularly challenging,²⁴ our analysis likely overestimated such artifacts, because the probability that the baby was likely to be awake was not negligible. Future studies should consider direct monitoring of sleep stages while acquiring DOT/EEG data.

To date, this represents the longest DOT monitoring ever performed on newborns, and we deem NeoCAP as the critical factor to achieve such a breakthrough, both in terms of monitoring time and signal quality. One of its points of strength is constituted by the presence of a soft inner layer that prevents direct contact between optodes’ support/electrodes’ support and the neonatal skin, thus avoiding pressure lesions and skin damages. The presence of a spongy inner layer allows postural changes of the newborn without determining increased pressure on a single side due to optodes’ support. The current proof-of-concept NeoCAP demonstrates that the appropriate cap may substantially increase the time of continuous DOT monitoring of whole-brain hemodynamics in very preterm neonates. Furthermore, NeoCap can be easily and quickly fitted on and removed from the infant’s head as the DOT device could be quickly and easily switched on and off. Nurses could carry on their routine neonatal care with minimal effort required to take care of the cap and the DOT device. The data quality results suggest that nurses were able to successfully fit the cap back on the infant’s head after each removal. Furthermore, the use of EasyCap as an external layer had the great advantage to help nurses reliably (re-)position the cap on the newborn’s head and reduce the likelihood of misplacement errors. Common landmarks that were used to check the position of the cap on the head were the location of the EasyCap borders, of the holes for the ears and the strip under the chin. The study’s physician and the nurse in charge of the newborn checked the positioning of the cap on the newborn’s head at least four times per day, and never noticed a misplacement of the cap. Taking pictures of the cap after each repositioning may be a viable solution to quantitatively assess the potential misplacement error.

NeoCAP permits the simultaneous recording of EEG signals when required (e.g., for functional stimulation or for clinical practice). In comparison to DOT, the EEG set up takes longer and electrodes are less comfortable for the infant’s skin, due to the presence of the electrolyte gel. The use of cup-like electrodes, filled with the electrolyte gel, could be more appropriate for future studies, because they might reduce the risk of spreading the gel over the inner surface of the cap. Another option would be to embed the electrodes in the inner layer, inside the sponge, so that less electrolyte gel would be needed to lower the impedance.

The low-cost manufacturing process of this head-cap can be replicated to create head-caps for different head sizes, spanning across preterm and term age ranges. Furthermore, optodes’ positions and electrodes’ positions can be modified during the manufacturing process according to the needs of the study and the number of sources and detectors available. Future studies should attempt to further reduce motion artifacts, which are quite common during the acquisition (although this comes to us as no surprise, given that the cap was not removed during parental visits or other clinical interventions to the baby).

Our work provides evidence that a prolonged continuous DOT monitoring is feasible in preterm infants, highlighting the crucial need for a cap adapted to this specific population to avoid some of the pitfalls in this research field. Adopting NeoCAP might open a new range of exciting opportunities for comparing whole-brain hemodynamic changes with other vital parameters and it might enable to consistently observe the impact of multiple daily stressful events on the brain of preterm infants.