A Review of Functional Near-Infrared Spectroscopy Studies of Motor and Cognitive Function in Preterm Infants

Abstract

Preterm infants are vulnerable to brain injuries, and have a greater chance of experiencing neurodevelopmental disorders throughout development. Early screening for motor and cognitive functions is critical to assessing the developmental trajectory in preterm infants, especially those who may have motor or cognitive deficits. The brain imaging technology functional near-infrared spectroscopy (fNIRS) is a portable and low-cost method of assessing cerebral hemodynamics, making it suitable for large-scale use even in remote and underdeveloped areas. In this article, we review peer-reviewed, scientific fNIRS studies of motor performance, speech perception, and facial recognition in preterm infants. fNIRS provides a link between hemodynamic activity and the development of brain functions in preterm infants. Research using fNIRS has shown different patterns of hemoglobin change during some behavioral tasks in early infancy. fNIRS helps to promote our understanding of the developmental mechanisms of brain function in preterm infants when performing motor or cognitive tasks in a less-restricted environment.

Introduction

In 2015, a survey of children under five showed that 2.7 million died during the neonatal period [1]. Preterm birth is the leading cause of child mortality in almost all high- and middle-income countries [2]. The complications of preterm birth are the largest direct cause of neonatal mortality, accounting for 35% of the world’s 3.10 million deaths each year, and at least 50% of all neonatal deaths [3]. Although advances in perinatal medical care have contributed to an increase in the survival rate of newborns, disorders of higher brain functions among surviving infants are still a crucial issue in perinatal medicine [4–7]. The gestational age (GA) of preterm neonates is generally <37 complete weeks [8]. Incomplete development of some organs may cause preterm infants to be unable to adapt to the outside world and prone to hypoxia and infection [9]. These may increase the risk of cerebral palsy, learning disabilities, and visual impairment in preterm infants, affecting long-term physical health and cognitive function [10]. The effects of preterm birth may last a lifetime, placing a heavy burden on the families, society, and medical institutions [11].

Brain imaging techniques can provide valuable information about brain development, and opportunities to link brain development to human behavior [12]. The traditional brain imaging techniques are mainly magnetic resonance imaging (MRI), electroencephalography (EEG), and computerized tomography (CT). However, CT scans generate radiation, and frequent scanning may affect brain development, so they are not recommended for preterm infants. MRI scans are not suitable for motor or cognitive function studies in preterm infants either. First, MRI has little tolerance for head movements [13]. Since it is almost impossible for a preterm infant to keep still for any length of time, the MRI scan must be done while the infant is asleep. The alternative is to sedate or anesthetize the infant before performing the scan, which may make parents reluctant to allow an MRI scan on their child. Second, a cognitive task is the most effective method for studying cognitive development, but sleeping infants cannot accomplish such tasks. Third, studying the brain function of preterm infants while they are performing natural behaviors requires long-term bedside monitoring with an instrument that is portable and possesses high ecological validity. These goals are difficult to achieve using MRI. Finally, the cost of MRI equipment is relatively high, and the availability of MRI units in low-income countries remains poor. A survey has shown that, in West Africa, 84 MRI units serve a population of 372,551,411 [14]. The lack of brain imaging equipment has led to higher mortality in preterm infants in low-income countries. A report from the World Health Organization showed that the mortality rate of preterm infants in low-income countries is ~9 times higher than that in high-income countries [15]. Although EEG is less restricted and more readily accepted by parents than MRI or CT, it is not suitable for developmental studies in preterm infants due to poor spatial localization. Accurate localization of the active areas in the brain is vital in studies of motor and cognitive function.

Functional near-infrared spectroscopy (fNIRS) is a brain imaging technology that is much more suitable for preterm infants than MRI, CT, or EEG. fNIRS devices are usually attached to the participant’s scalp, where optodes (optical sensor devices) emit near-infrared light. The light penetrates the soft tissues and bone and is absorbed by oxy- and deoxy-hemoglobin. Light detectors absorb the scattered light emitted by the optode and quantify the concentration of hemoglobin. fNIRS captures changes in cerebral oxygen saturation by measuring the intensity of scattered light in the active areas of the brain. Compared with traditional functional brain imaging technologies, fNIRS is less susceptible to noise and requires less restriction of head or body activity than MRI and CT [16]. Also, it has more accurate spatial localization of the active cerebral areas than EEG. In addition, fNIRS is relatively low-cost and so can be promoted on a large scale. Some fNIRS studies in resource-poor and remote areas have confirmed its suitability during infancy and its potential for field-based neuroimaging research on cognitive function [17, 18]. If fNIRS is widely used, more families with preterm infants will have access to brain imaging, even in remote and underdeveloped areas. Studies have shown that early intervention is helpful for neurodevelopmental outcomes in preterm infants, especially those who may have motor or cognitive deficits [19–21]. The common use of fNIRS also helps the early identification and diagnosis of these deficits, so that the affected infants can benefit from early intervention.

Infancy is the most vigorous stage of growth and development in a person’s life. Motor, hearing, and visual skills are all developing at this stage. In the late 1970s to mid-1980s, fNIRS was first used to study infants [22–24] and it is now widely used in research and clinical environments to monitor cerebral oxygen saturation in preterm infants with various conditions, such as apnea [25], intraventricular hemorrhage [26], and medication effects [27]. Many studies have confirmed the feasibility of using fNIRS to assess cerebral autoregulation [28, 29], and researchers have found that developmentally linked or acquired dysfunction of cerebral autoregulation is closely associated with the occurrence of brain injury. In recent years, some research has used fNIRS to compare cerebral oxygenation in preterm and term infants and assess the brain functions in preterm infants to explore their developmental trajectories.

There are already some reviews of fNIRS applications in exploring the brain function of preterm or term infants. Lloyd-Fox et al. [30] reviewed the achievements in neurodevelopment and the advantages of fNIRS in infant research. They focused on advances in techniques and the progress of methods in research design and the analysis of hemodynamic responses. Quaresima et al. [31] reviewed studies of cortical activation in the classical language region in newborns, children, and adults using fNIRS instruments of varying complexity. Wolf et al. [32] overviewed the instrumentation available, compared the advantages and limitations based on the underlying principles, and reviewed the application of fNIRS to measure oxygen saturation in the brain, liver, gastrointestinal tract, and peripheral tissue. However, there is no systematic review of fNIRS studies for the assessment of motor and cognitive function in preterm infants while performing tasks. In this review, we focused on studies assessing the brain function of preterm infants using fNIRS and were particularly interested in three functions: motor performance, speech perception, and facial recognition. We specifically concentrated on differences in cerebral hemodynamic responses between preterm and term infants when exposed to the same stimuli.

fNIRS Studies in Preterm Infants

We conducted a literature search first using PubMed. Subsequently, Google Scholar was used to supplement relevant articles. The keywords used in the searches were “near infrared spectroscopy”, “NIRS”, “preterm”, “premature”, “neonate”, “infant”, “speech”, “language”, “motor”, “movement”, and “face”. Next, the “Related Articles” function was used to broaden the search. The reference sections of all relevant studies were examined to identify additional papers. This literature search concluded in May 2019. Studies were limited to articles in English.

Included studies involved the use of fNIRS to assess brain function in preterm human neonates. Articles on animal experiments did not fulfill the inclusion criteria. Trials in both awake and sleep states were considered. Studies that examined the effect of drugs or breastfeeding on neonatal cerebral hemodynamics were excluded. In addition, research assessing cerebral oxygenation during hypoxic ischemic attacks or surgery in preterm infants was not considered. The search strategy is summarized in Fig. 1. Preterm infants were included as subjects in each study. Their GA was <37 weeks, and chronological ages were no more than one year. The brain function in preterm infants was assessed at hospitals in all studies.

Fig. 1.

Literature search strategy. The systematic search strategy highlights the number of articles acquired in the original search, the justification for rejection of certain articles, and the total number of articles included for systematic review, data extraction, and analysis.

According to the papers meeting the inclusion criteria, studies were classified into three areas: motor performance, speech perception, and facial recognition (Table 1). Our main concern was the cerebral hemodynamic responses in preterm infants exposed to visual, auditory, or motor stimuli. Also, we compared the cerebral hemodynamic responses between preterm and term infants, and explored the effect of preterm birth on different brain functions in infants.

Table 1.

Summary of fNIRS studies on motor and cognitive functions in preterm infants.

Study	fNIRS System	Method	Key finding
Motor performance
Isobe et al. [34]	NIR topography, Hitachi	Flexed and extended the right or left leg at the knee joint	HbO and tHb increase during contralateral knee movement
Kusaka et al. [35]	24 multichannel NIRS, Hitachi	Flexed and extended the knee or elbow joint	Significant difference in the area and degree of response between contralateral and ipsilateral movement
Kashou et al. [36]	NIRScout imaging system, NIRx	Palmar, plantar, and oromotor stimulation	Oromotor stimulus resulted in a 50% greater response than palmar or plantar stimuli
Pansy et al. [32]	INVOS Cerebral/somatic oximeter monitor	crSO2; assessed short-term neurological outcome by GMA	Motor impairment possibly associated with cerebral hypoxia
de Oliveira et al. [37]	NIRScout Tandem 1616, NIRx	Micro-direct-current motor held against right hand	Activation response bilateral and latency longer in preterms
de Oliveira et al. [38]	NIRScout Tandem 1616, NIRx	Micro-motor grasped in hand	12-month old preterm infants showed a contralateral response but the activation area was larger than 12-month-old term infants
Speech perception
Saito et al. [41]	NIRO200, Hamamatsu Photonics	Auditory stimuli (mother’s voice, nurse’s voice, or white noise)	Preterm infants reacted differently to different voice stimuli
Mahmoudzadeh et al. [40]	Imagent, ISS	Four syllables/ba/and/ga/produced by French male and female speakers; standard, deviant voice, and deviant phoneme trials	Preterm infants discriminated two syllables (/ba/vs/ga/) and human voice (male vs female).
Naoi et al. [42]	ETG-7000, Hitachi	Speech and non-speech auditory stimuli	Preterm infants: decreased activity in response to speech stimuli in the right temporal region, and increased interhemispheric connectivity
Arimitsu et al. [44]	Not mentioned	Phonemic contrast (/itta/vs/itte/), and prosodic contrast (/itta/and/itta?/)	Hemoglobin change pattern became the same as that of term infants at 38 weeks CGA
Arimitsu et al. [45]	ETG 4000, Hitachi	Phonemic contrast (/itta/vs/itte/), and prosodic contrast (/itta/and/itta?/)	Alterations in hemodynamic regulation and the functional system for phonemic and prosodic processing in preterm infants catch up by their projected due dates
Facial recognition
Carlsson et al. [57]	NIRO 300, Hamamatsu Photonics	Pictures of the mothers and unknown women	Alteration or delay of the maturation of the occipitotemporal pathway may cause difficulties in face recognition in preterms
Frie et al. [59]	NIRO 300 or NIRO 200, Hamamatsu Photonics	Gray background, mother’s face, an unknown face	Different cortical face recognition processes than term-born infants

Abbreviations: HbO, oxyhemoglobin; tHb, total hemoglobin; crSO₂, cerebral regional oxygen saturation; GMA, general movement assessment; CGA, corrected gestational age.

Motor Performance

The fetus develops rapidly between 20 and 37 weeks GA, and many sensorimotor networks are established during the second half of pregnancy. Therefore, motor development is one of the most affected areas in preterm birth and may cause many restrictions in later life. An fNIRS study indicated that motor impairment might be associated with cerebral hypoxia during the transition immediately after birth in preterm infants [33]. A recent meta-analysis of 11 studies assessed the prevalence of motor impairments in school-aged children born prematurely after the 1990s, and reported it as ranging from 19.0% to 40.5% [34].

Neonatal motor performance is closely associated with the development of the sensorimotor cortex. fNIRS is practical for studying the role of cortical sensorimotor function in motor behavior. Isobe et al. [35] performed passive knee movements on healthy term infants and preterm infants (GA: 24–35 weeks; median, 29.6 weeks) on days 3–57 during sedated sleep. They found that oxyhemoglobin (HbO) and total hemoglobin (tHb) increased in the primary sensorimotor area of all neonates, and the local deoxyhemoglobin (HHb) decreased in 6 of 7 neonates in response to contralateral knee movement. However, during ipsilateral knee movement, HbO and tHb showed slighter changes. To study this further, Kusaka et al. [36] added the stimulation of passive elbow and knee movement in term infants and preterm infants (GA: 24–41 weeks; median, 29.6 weeks) on days 3–99. The fNIRS results showed a significant difference in the area and degree of response between contralateral and ipsilateral movements. They found that contralateral knee and elbow movement caused a marked increase in HbO in a larger area of the sensorimotor cortex, while ipsilateral knee and elbow movement caused smaller changes in HbO in a smaller area.

Apart from the passive motor behavior in a sleeping state, researchers have demonstrated that fNIRS can be used to identify the activation of sensorimotor areas in awake infants. Kashou et al. [37] compared the durations of HbO changes in the left and right hemispheres of preterm neonates (mean postmenstrual age (PMA): 41.6–47.0 weeks) between palmar, plantar, and oromotor stimuli, and found that the oromotor stimuli resulted in a 50% greater response than the palmar or plantar stimuli. de Oliveira et al. [38] compared motor performance between term and preterm infants (GA <34 weeks) at 6 months chronological age using fNIRS and the Bayley Scales of Infant Development, Third Edition (Bayley-III). Motor performance was similar in full-term and preterm infants, but the hemodynamic responses were different. During sensorimotor stimulation, the cerebral activation response in full-term infants was contralateral, whereas the response in preterm infants was predominantly bilateral. The preterm group also exhibited a longer latency for the hemodynamic response than the full-term group. In 2019, de Oliveira et al. [39] conducted a longitudinal study with preterm and full-term infants at 6 and 12 months of age. In response to sensorimotor stimulation, preterm infants showed wider bilateral activation at 6 months, and an exclusively contralateral and more local activation response at 12 months. However, the 12-month-old preterm group continued to show a larger activation area than the full-term group of the same age. This may indicate that the brains of preterm infants are less specialized. Allievi et al. [40] saw a maturational trend toward faster, higher amplitude, and more spatially dispersed functional responses in the brains of preterm neonates (PMA: 32–45 weeks). They also found that, in preterm infants at a PMA equivalent to full-term infants, there was a decrease in both response amplitude and interhemispheric functional connectivity, and an increase in spatial specificity, culminating in the establishment of a sensorimotor functional response similar to that seen in adults.

The above studies show that in both the awake and sedated sleep states, the sensorimotor functions in preterm infants can be imaged and evaluated with fNIRS. Moreover, fNIRS can detect real-time activation responses in cortex, and this can be complimentary to movement assessment scales such as Bayley-III. Therefore, fNIRS has the potential to supplement the developmental assessment of behavioral responses.

Speech Perception

Many studies have shown that preterm infants have speech perception abilities. Mahmoudzadeh et al. [41] recorded cortical responses in the temporal and frontal areas using fNIRS in preterm infants at ~30 weeks PMA. The results showed that these infants could discriminate two syllables (/ba/versus/ga/) and the gender of a human voice (male vs female). When preterm infants were exposed to auditory stimuli in the form of utterances made by their mothers and female nurses, the active response in the right frontal area was different. In addition, they could discriminate between their mothers’ utterances and those of female nurses [42].

Language disorder is a major concern among preterm infants. fNIRS studies are important for understanding the physiological mechanisms underlying brain functions in preterm infants. A study conducted by Naoi et al. [43] examined cerebral activation and functional connectivity in response to infant-directed speech and adult-directed speech in term and preterm infants (mean PMA: 275.4 days) using a 94-channel NIRS instrument. Cerebral activation in response to speech stimuli in preterm infants was higher in the right temporal region and inter-hemisphere connectivity was higher than in full-term controls. This was especially evident in regions known to be involved in speech processing, such as the temporal and temporo-parietal regions, suggesting that preterm infants follow different developmental trajectories due to differences in intrauterine and extrauterine development.

Studying the cerebral hemodynamics of speech perception in preterm infants is valuable for early diagnosis and intervention for speech development deficits. fNIRS research on adults and infants has indicated that speech stimulation typically causes an increase in HbO and a slight simultaneous decrease in HHb [44]. However, Arimitsu et al. [45] found that preterm infants (GA: 26–36 weeks) do not always have the same patterns of hemoglobin change as full-term infants in response to phonemic or prosodic contrasts. They further found that the proportion of preterm infants with atypical hemodynamic patterns decreased, and finally became the same as that of term infants when they reached a corrected age of 38 weeks. These results reflected that the cerebral cortices of preterm infants may not initially be fully functional and the higher brain functions of preterm infants develop gradually before the projected due date. Arimitsu et al. [46] tested 80 infants (GA: 26–41 weeks) from 33 to 41 weeks PMA, and found that as PMA increased, hemodynamic regulation and the functional system for phonemic and prosodic processing developed and became consistent with term infants by the projected due date. The maturity of the hemodynamic response may reflect the development of language functions in preterm infants. Results also suggested that hearing infant-directed speech is helpful for the development of speech function in preterm infants, and that early intervention has a positive effect on speech perception disorders [46].

Facial Recognition

The ability to distinguish and recognize faces is necessary for social interaction. At birth, infants are able to distinguish between facial and non-facial patterns and prefer face-like patterns [47]. This ability is likely associated with cortical circuits [48–50]. The infant is exposed to faces and the cortical circuits begin to develop at 2 months [51]. Between 6 and 8 months, an infant can identify a face [52, 53]. fMRI studies suggest that the right temporal lobe, prefrontal cortex, and fusiform gyrus are the most important areas for facial recognition processes [54, 55]. Injury of these areas may be associated with impaired facial recognition in preterm infants. Such an impairment could affect social interactions, and may lead to depression or anxiety.

fNIRS provides a way to measure cerebral activity during face processing by monitoring changes in HbO, HHb, and tHb concentrations [56, 57]. It also offers a method of exploring the neurodevelopmental mechanism of facial recognition. A study by Carlsson et al. [58] found that infants at 6–9 months exhibit a higher activation-related hemodynamic response in the right fronto-temporal cortex when exposed to an image of their mother’s face than to an unknown face. The results suggested that the right fronto-temporal cortex is involved in facial recognition processes at this age. They further hypothesized that an alteration or delay in the maturation of the occipitotemporal pathway may cause facial recognition impairments [59]. Frie et al. [60] compared cortical hemodynamic responses to known and unknown facial stimuli between extremely preterm infants (corrected age: 6–10 months) and full-term infants. They found that preterm infants had a significantly smaller hemodynamic response in the right frontotemporal areas when looking at their mother’s face compared to an unknown face, which was opposite to the response of term infants [60]. These results suggest that preterm infants have different cortical facial recognition processes than term infants, possibly due to disruption of the subcortical system associated with an innate ability to prioritize face-like patterns over non-face-like patterns.

Summary and Perspectives

The studies reviewed above analyzed the relationships between hemodynamic activity and motor and cognitive development in preterm infants. We focused on three aspects of brain development (motor performance, speech perception, and facial recognition), and found that preterm infants have different, even opposite responses to the same stimuli when compared to term infants. These differences may have a few different causes. First, during the last 3 months of pregnancy, the fetal auditory system undergoes functional development [61, 62]. Preterm infants have a shorter GA, so the hemodynamic response to speech stimuli may be affected by their immature brains after birth. Second, differences in intra- and extra-uterine developmental environments may cause differences in hemodynamic patterns between preterm and term infants. Compared to a term infant, a preterm infant at an equivalent PMA has less intrauterine experience and more extrauterine experience. In addition, preterm infants are generally treated in neonatal intensive care units and receive more facial and sound stimuli than their term counterparts. In the first few weeks after birth, over-stimulation in the neonatal intensive care unit may have long-term effects on brain size and function in preterm infants [63]. This experience could affect cerebral hemodynamic responses to speech or facial stimuli. Finally, these differences between pre- and full-term infants may be caused by immature neural networks and angiogenesis in preterm infants. After oxygen consumption accompanying neuronal activity, the supply of oxygen is not efficiently delivered to the area of active neurons in the cerebral cortex [64], leading to abnormal changes of HbO and HHb in certain brain regions of preterm infants.

Different hemodynamic responses in preterm infants suggest that their brain functions may not develop at first. However, these functions gradually develop before the projected due date and eventually become the same as those of term infants at an equivalent PMA. In motor performance, the differences in lateralization of evoked sensorimotor cortex responses in term and preterm infants indicate that a considerable degree of functional lateralization develops during the first months of postnatal life. Similar phenomena have been found in the auditory cortex. As for phonemic and prosodic perception, the phoneme is generally processed in the left-temporal area, while prosody activates the right-temporal area in term infants and adults [65, 66]. After 39 weeks PMA, preterm infants begin to show functional laterality in the auditory cortex and gradually resemble term infants [46]. The research reviewed above also suggests that early intervention facilitates the development of cognitive functions. fNIRS is useful for early screening and intervention in the developmental problems of preterm infants.

The preterm infant is at elevated risk of cognitive deficits and developmental disorders. The fNIRS studies in preterm infants described in this article mainly focused on the first year after birth, which is the critical development period. Over the past 40 years [67], fNIRS has continuously developed and much progress has been made [68]. This is mainly due to its advantages. fNIRS will continue to develop as a methodology because its potential is still far from fully utilized.

Although fNIRS has apparent advantages, there are several challenges that limit its applications. fNIRS has poor spatial resolution and limited penetration depth [69, 70]. When assessing cerebral oxygenation saturation, the distance between the optode and the detector is at least 2 cm, which limits the spatial resolution [71]. In addition, the signal-to-noise ratio (SNR) in fNIRS is lower than in MRI. Many research groups have tried novel methods to overcome these shortcomings, such as combining fNIRS with other imaging methods. Chiarelli et al. [72] integrated fNIRS with EEG to monitor the neonatal brain, and showed that this provides additional information about the electrical and metabolic hemodynamic activities of the cerebral cortex, and has higher sensitivity and specificity. In preterm infants, combined techniques can provide new insights into brain function and help us better understand impairments.

In addition to poor spatial resolution, limited penetration depth, and a low SNR, there are other challenges when using fNIRS to study brain functions in preterm infants. Although fNIRS does not have strict head-movement limitations like MRI, sliding of the probe or changes in the intensity of the near-infrared light can result in spurious signals (motion artifacts) at the corresponding detector [73]. Motion artifacts may affect study results by masking actual changes of HbO and HHb. These artifacts may also influence baseline data. Motion artifacts can be reduced with hardware and software optimization, as well as by calming the preterm infant during the experiment, thereby reducing unnecessary motion caused by anxiety. Another challenge is that the results of fNIRS studies of the same brain function in preterm infants may be different [74]. This may be because different experimental designs are used in different studies [75]. In addition, differences in chronological ages and developmental levels among different studies also influence the results. Future research needs to weigh the comfort of the subjects, spatial resolution of the hardware, and potential for motion artifacts. At the same time, larger sample sizes are needed to replicate experiments.

As an imaging technique, fNIRS establishes a correspondence between brain activity and brain areas by measuring HbO and HHb concentrations, and promotes our understanding of brain development in preterm infants. This will inevitably help us to identify critical periods for intervention in atypical development, thereby reducing the risk of motor and cognitive disorders and minimizing the impact of preterm birth on infants. However, the current fNIRS research in preterm infants still has many limitations. Apart from the three areas reviewed, the developmental mechanisms of other brain functions are still unclear. Further studies with fNIRS are needed to better understand the mechanisms of brain functions in preterm infants.

Conflict of interest

The authors claim that there are no conflicts of interest.