Functional near-infrared spectroscopy (fNIRS) was once confined to stationary, wired systems that limited studies to controlled laboratory environments. But today’s devices using this technology are lighter, wireless, and designed for true portability and clinical applicability. Ongoing research could change how the medical community views the utility of NIRS, expanding its use in neurodegenerative and neurodevelopmental disorders, rehabilitation, and psychiatry. Tracking dynamic signals in the brain is a primary purpose of functional near-infrared spectroscopy (fNIRS), which can be used in conjunction with other modalities. Courtesy of iStock.com/peterschreiber.media. With origins tracing back to the 1970s, early NIRS instruments detected only relative changes, were riddled with noise, sensitive to probe placement, and prone to misinterpretation. Unfortunately, the reputation has stuck, and some clinicians still consider fNIRS a research “toy.” Although the instrumentation has been upgraded significantly in the past decade, the use of fNIRS still lags behind more established modalities, such as MRI and electroencephalogram (EEG), as a trusted monitor of brain function. fNIRS measures changes in the properties of light as it shines through the skull and is refracted back to a specialized detector. It is most often used to monitor changes in the concentration of oxygenated and deoxygenated hemoglobin molecules in the blood that indicate brain activity, but it is also used for other biomarkers. Unlike MRI or EEG, fNIRS lacks decades of clinical evidence, so each new device must prove itself through large, peer-reviewed trials that correlate optical metrics with patient outcomes or gold-standard technologies. This evidence-building phase is expensive but essential to future acceptance in a clinical setting. Enhanced portability Advancements in probe design, signal analysis, hardware stability, and acquisition time are redefining where and how brain imaging can be performed. “The biggest breakthrough definitely lies in the achievement of complete portability and ease of use for fNIRS devices,” said Sophie Apprich, medical application specialist at Artinis Medical Systems. The Dutch company has developed the MediBrite clinical NIRS unit, which weighs just 200 g. The device attaches to a head cap via wireless streaming and uses short-channel separation to strip away scalp noise. A spectroscopy setup featuring the Ocean NR Near-Infrared Spectrometer by Ocean Optics. This system is commonly used in pharmaceutical quality control, material characterization, and chemical composition analysis, where NIR spectra provide insights into molecular structure, moisture content, and other physical or chemical properties. Courtesy of Ocean Optics. “You can put it on an infant, a Parkinson’s patient during rehab, or even outside a lab. That changes everything,” Apprich said. Allowing patients to move freely during monitoring extends the range of possible clinical applications. It also opens the door to measurements outside traditional clinical or laboratory environments, including potential home use. In recent years, the detection of neurological and psychiatric disorders with fNIRS has gained traction through the use of machine learning to extract salient features. For example, in a 2024 study, associate professor of biomedical engineering Edgar Guevara and his colleagues at the Autonomous University of San Luis Potosí in Mexico validated the use of Artinis’ wearable MediBrite device with machine learning to reliably detect Parkinson’s disease1. In the study, fNIRS data was collected during cognitive or motor tasks, preprocessed, and used to select features for high-accuracy classification. Short-separation channels helped to filter out superficial noise from the heart rate and breathing, and automated multipower gain control enabled flexibility in vivo. Artinis’ Brite Family devices, for example, used transmitters that delivered light at different output levels. This feature allows users to apply lower light in transparent regions, such as the prefrontal cortex, and higher output in areas with more hair or greater pigmentation. “Multipower gain control enables the possibility of measuring any [human] subject regardless of skin and hair color,” Apprich said. “It allows for measurements in various cortical regions simultaneously and decreases setup time.” While most fNIRS research focuses on oxygenated and deoxygenated hemoglobin, compact broadband systems can be adapted for tissue spectroscopy to support bedside metabolic monitoring. For example, researchers at University College London paired an Ocean Optics QEPro spectrometer with a tungsten-halogen light source to track cytochrome c oxidase (CCO), a key enzyme in cellular respiration and mitochondrial activity. In one study, fNIRS was used to monitor brain activity in brain-injured patients instructed to imagine playing a game of tennis. The study provides evidence of preserved awareness despite no observable behavioral signs. Courtesy of Adrian Owen/University of Western Ontario. The team customized the setup by removing the QEPro’s entrance slit and coupling high-diameter fibers, optimizing for the faint signals reflected from brain tissue. In a pilot trial with traumatic brain injury patients, the system not only captured blood flow-related changes during controlled increases in carbon dioxide (hypercapnia) but also detected shifts in brain metabolism. Patients with severe injuries showed reduced or uncoordinated signals from CCO responses, suggesting impaired cellular energy production. “These spectrometers offer high sensitivity, low noise, and broad wavelength coverage (900 to 2500 nm and 220 to 1100 nm), enabling deeper tissue penetration and more accurate hemodynamic measurements,” said Joseph Bonvallet, director of product management at Ocean Optics. Despite measuring just 42 mm × 40 mm × 27 mm and weighing 70 g, the instrument delivers fast acquisition rates, strong UV sensitivity, and a high signal-to-noise ratio. “This study was one of the first to show that oxidized CCO could be measured in vivo using commercially available components,” Bonvallet said. “The Ocean ST series delivers full-spectrum performance in a device small enough to fit in the palm of your hand.” The same approach was used by a team led by the University of Cambridge’s Deepshikha Acharya to monitor oxidized CCO as a crucial biomarker for explaining cognitive differences in patients with dementia. In the study, the team used a customized broadband NIRS system with a filtered source and spectrometer2. Several concrete developments continue to boost NIRS performance and usability. For example, modern systems employ stable picosecond diode lasers and highly sensitive single-photon detectors with time-correlated single-photon counting, further improving accuracy and reproducibility of tissue oximetry measurements. Flexible probes have been developed with an integrated capacitive contact sensor to ensure consistent skin placement. “This seemingly simple improvement greatly reduces errors from probe repositioning — crucial for clinical reliability,” said Michele Lacerenza, cofounder and CTO of PIONIRS in Milan, Italy. Lacerenza and his team at PIONIRS, a spinoff company from the Polytechnic University of Milan, developed both flexible probes and photon travel-time analysis methods. These proved to be particularly invaluable in pediatrics because, unlike MRI, they enable infants to be monitored at the bedside with quantitative hemodynamic values. The team’s NIRSBOX platform uses subnanosecond laser pulses and single-photon detectors to capture photon time of flight, extracting absolute values of absorption and scattering. This means clinicians can now read baseline-referenced tissue oxygen saturation and hemoglobin concentrations — metrics comparable across sessions, operators, and patients. “This higher measurement precision lets us detect subtle physiological changes that earlier devices buried in noise — as demonstrated in our recent study on impaired cerebral saturation in children with mild pneumonia,” Lacerenza said3. The contrast with continuous-wave NIRS is stark. In a study of more than 350 children, the NIRSBOX delivered standard deviations of 4. “Five years ago, we could not imagine a portable time-resolved NIRS system with reproducibility under 1%,” Lacerenza said. “It’s like going from a thermometer that shows only whether it’s hotter or colder to one that tells you the actual temperature. Establishing normative curves, similar to pediatric growth charts, now looks achievable.” Currently, PIONIRS is a technology partner of the European Union-funded PROMETEUS Project (Preterm Brain-Oxygenation and Metabolic EU-Sensing, grant no. 10199093) aimed at reducing neurodisability in very low-birth-weight infants. The consortium is developing a neonatal intensive care unit (NICU)-ready time-domain NIRS (TD-NIRS) and speckle contrast optical spectroscopy (SCOS) cap to track cerebral oxygen saturation and blood-flow indices continuously while clinicians adjust nutrition. Other than that, such hybrid devices have the potential to be used in many other applications, including adjustment of ventilation and transfusion protocols. By achieving quantifiable, baseline-referenced values, TD-NIRS may help to pave the way toward establishing normal reference ranges and clinically relevant thresholds. In other research, Adrian Owen, professor of cognitive neuroscience and imaging at the Western Institute for Neuroscience, University of Western Ontario, Canada, and cofounder and chief scientific officer of Creyos Inc., is applying fNIRS to detect consciousness in more than 100 brain-injured patients. For decades, behavioral assessments have dominated ICU practice, but Owen argues that their limitations — motor impairments, fluctuating arousal, and examiner bias — leave too much uncertainty. “As a result, they often fail to capture the full spectrum of responsiveness, leading to potential misdiagnoses,” he said. Perhaps the most striking impact of integrating the technology into research comes from direct patient studies. In a prospective analysis of 32 ICU patients, Owen’s team used fNIRS to detect willful brain activity during a motor imagery task. “Of 32 patients, eight were able to willfully modulate their brain activity when instructed to imagine playing a game of tennis — providing evidence of preserved awareness despite no observable behavioral signs that this was the case,” he said. Bringing fNIRS capability into new frontiers is professor Adrian Owen’s work in neuroimaging and the study of patients in a vegetative state. Images courtesy of Adrian Owen/University of Western Ontario. Artwork by Cassio Lynm. “We concluded that fNIRS can detect preserved consciousness in 25% of patients deemed behaviorally unresponsive in an ICU setting,” Owen said. “Such methods could prevent the inappropriate and premature withdrawal of life-sustaining treatments.” Owen believes there is clearly an opportunity for more widespread use that will be met when smaller and cheaper fNIRS systems become available. “We are in conversations with manufacturers about how to reach this goal because I think it will fundamentally change the assessment of serious brain injuries in ICUs worldwide,” he said. Bridging the credibility gap Two of the largest challenges companies face in commercializing fNIRS are raising awareness of its clinical uses and convincing clinicians and researchers of its potential. For all its promise, fNIRS remains unfamiliar to most clinicians. The reasons are as much cultural as technical. MRI has decades of validation, tens of thousands of published studies, and established reimbursement pathways in health care systems, while EEG has more than a century of clinical use. By contrast, fNIRS is still relatively unknown. “Market awareness is our biggest hurdle,” Artinis’ Apprich said. “Clinicians know fMRI; they don’t yet know fNIRS.” Without large randomized controlled trials and longitudinal studies, regulators hesitate, payers resist, and hospitals hesitate to invest. “To date, there is no real ready-to-use clinical application for fNIRS. There are no standard practices for fNIRS yet, and clinicians who do not perform research do not yet envy working with fNIRS,” Apprich said. “This is further complicated by the fact that, with the exception of Japan, most countries do not provide reimbursement for fNIRS measurements.” Hybrid systems Integration with other modalities may help to accelerate credibility: Early clinical data shows that decisions based on multiple parameters can be timelier and more individualized than with those made with any single modality alone. EEG-fNIRS caps are already commercially available, which produce data pairing millisecond-scale electrical resolution with hemoglobin-based spatial maps. Such multimodal approaches are particularly valuable in stroke rehabilitation and cognitive assessment, where neither modality alone is sufficient. fNIRS can also be combined with EEG to study noninvasive brain stimulation methods, such as transcranial electrical stimulation (tES). This approach is useful for assessing the effectiveness of tES as a therapeutic tool in various neurological and psychiatric disorders. Multimodal experiments introduce added complexity. For instance, coordinating the placement of fNIRS optodes and EEG electrodes requires careful planning, and precise synchronization of the recordings is essential for accurate data analysis. “Developments, especially in software solutions such as Lab Streaming Layer or stimulus presentation software solutions, can facilitate synchronization,” Apprich said. A recent study conducted by Li Yi, a professor at Foshan University in China, shows the potential of hybrid systems with software analysis in the diagnostics of depression5. The study demonstrates that combining EEG, fNIRS, and machine learning creates an effective method to detect and classify depression on an individual patient level. Accuracy of detection was increased when combining fNIRS and EEG, compared with EEG alone. Combining fMRI with fNIRS is more complex but offers the advantages of whole-brain coverage and deep-tissue validation of surface fNIRS signals. The PIONIRS team developed fMRI-compatible optical probes to enable simultaneous TD-NIRS and fMRI. “Typically, researchers perform one detailed mapping session with fMRI, then use fNIRS for repeated bedside monitoring or more ecological settings,” PIONIRS’ Lacerenza said. The integration of fNIRS has moved beyond its traditional pairing with EEG or fMRI. Today, TD-NIRS is increasingly combined with optical blood-flow monitoring techniques such as diffuse correlation spectroscopy (DCS) or its camera-based variant, SCOS. While fNIRS tracks changes in blood oxygen concentration, DCS and SCOS measure how fast red blood cells are moving. Together, these methods provide both concentration and flow, allowing researchers to calculate the brain’s metabolic rate of oxygen in real time — a powerful indicator of tissue health. The European Union-funded H2020 VASCOVID consortium was among the first to validate a fiber-based TD-NIRS and DCS system on peripheral microvasculature in sepsis and ICU patients, while also developing a streamlined, medical device regulation (MDR)-compliant, simplified quality management system to support future clinical translation. “We are now translating part of this architecture to the brain in PROMETEUS for a NICU cap that combines TD-NIRS with SCOS to monitor cerebral oxygen delivery and flow in preterm infants,” Lacerenza said. The road to clinical trust The trajectory of fNIRS resembles that of pulse oximetry in the 1980s: from bulky, unreliable prototypes to indispensable vital-sign monitors. The difference is that fNIRS must clear higher bars — measuring not the fingertip, but the brain. Success depends less on optics than on evidence. Projects such as PROMETEUS and VASCOVID, Owen’s ICU studies, and Artinis-led machine learning diagnostics all aim to provide the large, peer-reviewed data sets needed to persuade regulators and insurers. “If you don’t link optical metrics to outcomes,” Ocean Optics’ Bonvallet said, “the technology won’t leave the lab.” Beyond clinical validation lie regulatory hurdles. The international standards effort is playing catch-up. Clinical fNIRS systems are generally classified as Class II devices, with FDA or EU-MDR clearance currently taking 12 to 18 months and requiring extensive safety and efficacy data. The safety and performance benchmark for NIRS equipment, IEC 80601-2-71, was first published in 2015, with a major revision expected in 2025. Currently, the Society for functional Near-Infrared Spectroscopy is drafting best practices for device validation, probe design, and data analysis. These will be essential before fNIRS can shed its niche label. The next five years will be decisive. If multicenter trials confirm that fNIRS can guide neonatal care, predict ICU outcomes, or personalize neurorehabilitation, the credibility gap may close. If not, fNIRS risks remaining a boutique research tool. What fNIRS needs next is not another breakthrough in photonics — but the slow, painstaking accumulation of clinical trust. References 1. E. Guevara et al. (2024). Integrating fNIRS and machine learning: shedding light on Parkinson’s disease detection. EXCLI J, Vol. 23, pp. 763-771. 2. D. Acharya et al. (2025). Mapping functional hemodynamic and metabolic responses to dementia: a broadband spectroscopy pilot study. J Biomed Opt, Vol. 30, p. S23910. 3. M. Lacerenza et al. (2025). The role of cerebral oxygenation in pediatric lower respiratory tract infections based on insights from time domain near infrared spectroscopy tissue oximetry. Sci Rep, Vol. 15, p. 31171. 4. V. Calcaterra et al. (2025). Reference values for cerebral and peripheral tissue oximetry in children: a clinical TD-NIRS study. Acta Paediatr, Vol. 114, No. 3, pp. 515-525. 5. L. Yi et al. (2023). Automatic depression diagnosis through hybrid EEG and near-infrared spectroscopy features using support vector machine. Front Neurosci, Vol. 17, p. 1205931.