Investigating Sepsis-Associated Delirium Through Optical Neuroimaging: A New Frontier in Critical Care Research

Abstract

Sepsis is a life-threatening syndrome consisting of physiological, pathological, and biochemical abnormalities induced by infection which continues to be a major public health burden. It remains one of the most common reasons for intensive care unit (ICU) admission. Delirium precipitated by sepsis in the intensive care setting is one of its most common neuropsychiatric complications that leads to prolonged hospitalization, increased mortality, and an increased risk of incident dementia. Understanding the pathophysiology and neurobiological mechanisms of sepsis-associated delirium is difficult; neuroimaging biomarkers are lacking due to difficulties with imaging critically ill patients. Optical imaging techniques, including near-infrared spectroscopy and diffuse optical tomography are potentially promising approaches for investigating this pathophysiology due to their portability and high spatiotemporal resolution. In this review, we examine the emergence of optical neuroimaging techniques for the study of sepsis-associated delirium in the ICU and how they can further advance our knowledge and lead to the development of improved preventative, predictive, and therapeutic strategies.

1. Introduction

Sepsis is a life-threatening syndrome consisting of physiological, pathologic, and biochemical abnormalities induced by infection which remains a major public health burden [1]. In the United States, sepsis is responsible for 6.2% of the aggregate costs for all hospitalizations, approximately USD 23.7 billion annually [2]. It is the leading cause of mortality and critical illness worldwide in the intensive care unit (ICU) setting, with an estimated worldwide prevalence of 29.5% and a mortality level of 25.8% [3,4,5]. Survivors of sepsis commonly suffer from various long-term physical, psychological, and cognitive disabilities with significant health care and psychosocial implications [6]. The main mechanism causing such disability and complications from sepsis is the presence of delirium. Delirium is a neuropsychiatric syndrome that typically manifests acutely or subacutely, with a fluctuating course. It involves disturbances in attention, alertness, various cognitive functions, psychomotor activity, emotional regulation, and the sleep–wake cycle. The three major subtypes of delirium are hyperactive (agitation), hypoactive (psychomotor slowing), and mixed [7]. It is always caused by a medical insult, with sepsis being one of its most common precipitants, particularly in the ICU [8]. Delirium is exceedingly common in the critically ill, with rates as high as 83% in mechanically ventilated patients [9]. Irrespective even of sepsis, delirium is independently associated with a higher risk of mortality, institutionalization after hospital discharge, and cognitive impairment [10,11,12]. At present, there is no definitive treatment for any type of delirium, and managing the underlying medical condition causing the delirium remains the priority. To make real progress in preventing and treating sepsis-associated delirium, it is essential to enhance our ability to investigate its underlying neural mechanisms and identify reliable biomarkers.

1.1. Sepsis-Associated Delirium Pathophysiology

Sepsis is defined as a life-threatening organ dysfunction caused by a dysregulated host response to infection [1]. The immune response to a pathogen starts with the activation of innate immune cells, such as macrophages, monocytes, neutrophils, and natural killer cells. This activation occurs when pathogen-associated molecular patterns (PAMPs), like bacterial endotoxins and fungal β-glucans, interact with pattern recognition receptors [13]. Additionally, damage-associated molecular patterns (DAMPs), which consist of intracellular substances or molecules released by damaged or dying cells (like ATP and mitochondrial DNA), can also engage specific receptors on monocytes, macrophages, and myeloid-derived suppressor cells (MDSCs) [14], including toll-like receptors (TLRs), C-type lectin receptors, nucleotide-binding oligomerization domain-like (NOD-like) receptors, and retinoic acid-inducible gene I (RIG-I-like) receptors. This interaction activates intracellular signaling pathways that lead to the transcription and release of proinflammatory cytokines, such as TNFα, IL-1, and IL-6 [15,16]. Some pattern recognition receptors, like the NOD-like receptor family, can form larger protein complexes called inflammasomes, which contribute to the production of cytokines like IL-1β and IL-18, as well as caspases involved in programmed cell death [17]. The resulting proinflammatory cytokines trigger leukocyte activation and proliferation and complement system activation, endothelial adhesion molecule expression, and chemokine production [18]. In sepsis, this immune response becomes dysregulated and exaggerated, leading to significant inflammatory damage to host cells and tissues.

In addition to proinflammatory pathway dysregulation, there is also the simultaneous dysregulation of hemostatic and coagulation cascades. The disruption of coagulation in sepsis arises from multiple contributing factors. One key driver of the hypercoagulable state in sepsis is the release of tissue factor, primarily from damaged endothelial cells, but also from monocytes and polymorphonuclear cells [19]. This tissue factor triggers widespread activation of the coagulation cascade, leading to thrombin production, platelet activation, and the formation of platelet–fibrin clots. These clots can cause microthrombosis, which impairs local blood flow, resulting in tissue hypoxia and organ dysfunction [20]. Additionally, sepsis is associated with a decrease in fibrinolysis. Elevated levels of TNFα and IL-1β prompt the release of tissue plasminogen activators from endothelial cells, increasing plasmin activation. However, this process is counteracted by a sustained rise in plasminogen activator inhibitor type 1 (PAI-1), leading to reduced fibrinolysis and the impaired removal of fibrin. This combination of hypercoagulability and reduced fibrinolysis contributes to ongoing microvascular thrombosis [21].

The blood–brain barrier (BBB) is composed of endothelial cells and is responsible for the regulation of nutrients and the prevention of neurotoxins from entering the central nervous system (CNS) [22]. Sepsis induces profound alterations in the BBB endothelium through the mechanisms described above, thus causing improper leukocyte adhesion, coagulation, vasodilation, and a loss of barrier function [23]. Macroscopically, this leads to hypoperfusion, impaired cerebral autoregulation of blood flow, and most critically, oxygenation deficits throughout various brain networks. Reduced cerebral oxygen levels have been closely linked to the disruption of ionic gradients, impaired cellular depolarization, and the accumulation of neurotoxic free radicals [24]. Furthermore, inadequate oxygenation disrupts normal neurotransmitter metabolism, leading to a significant increase in glutamate and dopamine levels, along with decreased acetylcholine production—all of which are strongly associated with the development of delirium, even in non-septic cases [25,26,27]. Additionally, poor oxygenation contributes to decreased cerebral perfusion and heightened brain inflammation, perpetuating a cycle of dysfunction [28]. In combination, immune, inflammatory, and coagulation dysregulation leads to hypoperfusion and oxygenation deficits in the brain which can ultimately cause delirium (Figure 1).

1.2. Delirium Neuroimaging Studies

To fully appreciate and investigate the underlying neural mechanisms of delirium, validating neuroimaging biomarkers is critically important. Previous structural neuroimaging studies using magnetic resonance imaging (MRI) and computer tomography (CT) have shown that delirium is linked to grey matter volume reduction and white matter hyperintensities, along with tract disruptions in various brain regions [29,30,31]. However, these results are not exclusive to delirium, and are largely due to non-specific findings, diverse patient populations, and small sample sizes. More recent functional imaging studies have identified significant hypoactivity in several cortical regions and disruptions in connectivity within the default mode and central executive networks [32,33]. The frontal and cingulate cortices are frequently implicated in delirium, along with subcortical structures like the thalamus and caudate, which are crucial for arousal regulation via the ascending reticular activating system [34,35]. Functional MRI (fMRI) and positron emission tomography (PET) have been used in these studies, though they often involve even smaller sample sizes than structural imaging research, given the operational difficulties in doing such scans. Moreover, PET imaging requires fasting and a long uptake period (approximately 45 min depending on the radioligand), making it impractical for many delirious patients. A common limitation across MRI, CT, and PET techniques is the lack of portability and low patient tolerance due to the patient’s acute illness and cognitive impairments. This is particularly cumbersome and non-feasible in the ICU population where patients are even more seriously ill and unable even to be transported readily. Moreover, in cases of hyperactive or mixed delirium, the severity of agitation often prevents the patient from being able to safely undergo imaging with standard techniques, as they are required to remain still in an enclosed space. While electroencephalography (EEG) is more portable and widely used, its low spatial resolution limits its ability to conduct anatomical localization of brain activity and advanced regional and network analyses (

Table 1. Characteristics of functional imaging techniques. Abbreviations: DOT = diffuse optical tomography; fNIRS = functional near-infrared spectroscopy; fMRI = functional magnetic resonance imaging; EEG = electroencephalography; PET = positron emission tomography; SPECT = single-photon emission computed tomography.

Techniques	Spatial Resolution	Temporal Resolution	Bedside Capabilities	Patient and Safety Considerations
DOT	mm	ms	Yes	None
fNIRS	cm	ms	Yes	None
fMRI	mm	s	No	Ferromagnetic material
EEG	cm	ms	Yes	None
PET	mm	min	No	Fasting requirement
SPECT	cm	min	No	Ionizing radiation

) [36]. The emergence of optical imaging techniques offers a potential solution to the challenges of neuroimaging in delirium.

1.3. Optical Neuroimaging

In 1977, Jöbsis demonstrated the potential to measure blood and tissue oxygenation levels in the brain of a cat with near-infrared (NIR) light [37]. In the decades since, optical neuroimaging has expanded into a diverse research area, with many scientific and clinical studies utilizing the distinctive properties of light for brain imaging [38]. NIR light can provide valuable functional imaging information by capturing intrinsic changes in absorption, fluorescence, or the scattering of light within soft tissues. Additionally, various exogenous contrast agents can be used to obtain further insights [39,40,41]. Commonly measured chromophores include oxyhemoglobin, deoxyhemoglobin, cytochromes, and metabolites, which serve as indicators of brain activation, similar to traditional functional imaging techniques. A typical setup focused on portability will often include a cart that holds the light source (laser-emitting diodes or pulsed lasers), data acquisition system, computer, and timing electronics that connect to a flexible head interface via optical fibers. The interface itself contains the source–detector pairs and optodes that physically contact the hair and scalp, similar to how an EEG cap is designed.

Optical imaging offers several advantages over conventional functional imaging, including real-time data acquisition, greater portability, fewer movement restrictions, no exposure to ionizing radiation, a wider range of contrast agents, and lower costs [42]. These benefits are notable, particularly for use in ICU patients given their many restrictions due to their critically ill status. For applications in delirium, two key optical imaging techniques of interest are functional near-infrared spectroscopy (fNIRS) and diffuse optical tomography (DOT).

Functional near-infrared spectroscopy (fNIRS) is based on the core principles of NIR spectroscopy, with the term “functional” added after its initial use in human studies [43]. fNIRS allows for the simultaneous measurement of changes in the optical properties of the human cortex across multiple sites, presenting results as a topographical map or image of the region of interest [44]. The technique leverages the fact that human tissues are transparent to NIR light in the 650–1000 nm spectral range, where chromophores like hemoglobin in small blood vessels absorb or scatter the light [45]. By measuring increases in oxyhemoglobin and decreases in deoxyhemoglobin, fNIRS detects real-time markers of local arteriolar vasodilation, which corresponds to increased oxygenation, cerebral blood flow, and volume in specific brain areas. This hemodynamic response is linked to neurovascular coupling, serving as an indirect indicator of neuronal activity [46]. Many companies now offer multi-channel fNIRS devices at a much lower cost than fMRI machines. These devices are compact, easy to set up, and can be quickly transferred between patients, making them highly practical for clinical use [47].

Diffuse optical tomography (DOT) is a modern noninvasive imaging technique based on near-infrared (NIR) light, similar to how magnetic resonance spectroscopy relates to magnetic resonance imaging (MRI) [48]. In contrast to the two-dimensional capabilities of fNIRS, DOT utilizes multiple wavelengths and dense, overlapping channels with different source–detector distances to gather data at various depths. This allows for the measurement of hemodynamic responses at different brain depths and the generation of high-resolution three-dimensional images (Figure 2) [49]. As a result, DOT provides a strong alternative for 3D functional neuroimaging, even compared to fMRI. While both DOT and fMRI detect blood-oxygen-level dependent (BOLD) signals, fMRI primarily measures deoxyhemoglobin [50]. In contrast, DOT can simultaneously capture all hemoglobin signals, enabling better differentiation of the timing, location, and magnitude of neurovascular coupling, which provides a more detailed view of dynamic brain activity [51,52,53]. Like fNIRS, DOT is affordable, portable, and mobile, and the development of wearable DOT systems with flexible interfaces is progressing rapidly [54,55,56,57,58].

In addition to the technical parameters that highlight the potential strengths of optical imaging for sepsis-associated delirium, another critical advantage is the focus on measuring oxygenation changes through the evaluation of hemoglobin signals. This concurrently measures cerebral hemodynamic and perfusion states, thus providing a high-resolution map of various regions involving dysconnectivity. Based on sepsis pathophysiology, as described above, cerebral oxygenation changes are hypothesized to be the penultimate pathway in the development of delirium. Thus, the capability to measure this directly and dynamically at the bedside is highly important for advancing our understanding of delirium pathophysiology and how it aligns with sepsis pathophysiology.

2. Optical Neuroimaging in Sepsis-Associated Delirium

In 2008, Pfister and colleagues were the first to explore the potential of optical neuroimaging in studying the pathophysiology of sepsis-associated delirium (

Table 2. Summary of published sepsis-associated delirium optical imaging studies.

Reference	Number of Patients	Optical Imaging Technique	Imaging Site(s)	Results
Pfister et al., 2008 [59].	23	NIRS	Frontal lobe and frontoparietal	No difference in oxygenation levels during delirium; however, Doppler-based blood flow velocity was diminished in delirious patients.
Funk et al., 2016 [60].	15	NIRS	Frontal lobe	No difference in cerebral oxygenation in delirious patients.
Wood et al., 2016 [61].	10	NIRS	Single forehead optode	Higher saturated oxygenation levels were observed in delirium.
Vasko et al., 2014 [62].	25	NIRS	Frontal lobe	Lower cerebral oxygenation was reported in delirious patients consistently.
Wood et al., 2017 [63].	88	NIRS	Single forehead optode	Longer duration of delirium was correlated with worse cerebral oxygenation values.
Lee et al., 2019 [64].	40	NIRS	Single forehead optode	Duration of delirium and worse COx values associated with the development of delirium.
Rosenblatt et al., 2020 [65].	6	NIRS	Frontal lobe	Severity of delirium was associated with COx values.
Jiang et al., 2024 [66].	12	DOT	Dorsolateral and dorsomedial prefrontal cortex	During delirium and post-delirium exhibited lower oxygenation levels compared to controls. Higher severity of delirium correlated with lower oxygenation levels.

) [59]. They enrolled 23 septic patients, with 16 included in the final analysis, of whom 12 were diagnosed with delirium using the Confusion Assessment Method for the Intensive Care Unit (CAM-ICU) [9]. To assess cerebral oxygenation, the researchers used a tissue oxygenation index (TOI) as a proxy, measured via bilateral NIRS over the frontal to frontoparietal regions for one hour within 48 h of ICU admission. No significant differences in the TOI were observed between delirious and non-delirious patients. However, separate transcranial Doppler measurements revealed significant differences in blood flow velocity and regulation between the two groups. The authors suggested several factors that may have contributed to the inconclusive TOI results, such as a small sample size, a lack of simultaneous Doppler and NIRS recordings, and limited optode placement (two channels; an optode is similar to an EEG electrode).

Subsequent studies from other research groups helped shed more light on the potential mechanisms linking sepsis to delirium. Funk et al. conducted a similar study, enrolling 15 patients, 7 of whom were CAM-ICU positive for delirium [60]. By employing bilateral frontal NIRS, the researchers assessed tissue oxygen saturation for 24 to 48 h within 12 h of ICU admission. Their results indicated no definitive link between tissue oxygen saturation and the onset of delirium. However, the authors expressed concerns about the sensitivity of their chosen saturation threshold (65%) in detecting delirium and proposed that cerebral edema in their critically ill patients could have obfuscated the findings as well. In contrast, Wood et al. performed a smaller study using single-forehead NIRS on 10 septic patients, 5 of whom were delirious [61]. They found that the non-delirious patients had higher brain tissue oxygenation (BtO2) levels compared to those with delirium. Unlike previous studies, Wood et al. extended the monitoring period, using continuous NIRS over the first 72 h of sepsis onset. They proposed that the timing of the NIRS measurements—particularly during the early treatment phase—could be important due to fluctuations in oxygenation levels linked to delirium. These findings were later corroborated by Vasko et al., who compared cerebral oxygenation between 15 septic patients and 10 age- and sex-matched controls [62]. Their results showed significantly lower oxygen saturation levels in patients with sepsis-associated delirium, as measured by bilateral frontal NIRS.

The first major prospective cohort study using NIRS to track the development of delirium was carried out by Wood et al. as part of their subsequent trial, the Cerebral Oxygenation and Neurological Outcomes Following Critical Illness (CONFOCAL) study [63]. This trial included 88 adult ICU patients, who were monitored with a single forehead NIRS sensor during the first 24 h of their admission. Delirium screenings were performed daily, with a maximum enrollment period of 30 days. The study confirmed and expanded upon their earlier pilot findings, showing a significant inverse relationship between brain tissue oxygenation (BtO2) and the proportion of time patients spent in delirium while in the ICU. BtO2 was also identified as an independent predictor of delirium, regardless of other variables. A significant additional finding showed that BtO2 levels were not correlated with peripheral hemodynamic measures, suggesting that cerebral autoregulation operates through an independent mechanism. In a subsequent study using data from the CONFOCAL trial, researchers examined cerebral autoregulation by computing the cerebral oximetry index (COx). COx was defined as the time-varying correlation between the mean arterial pressure (MAP) and NIRS-derived BtO2, with positive values indicating impaired autoregulation. NIRS measurements were taken from 40 patients using a single forehead sensor, and COx values were calculated during the first 72 h of ICU admission. The findings showed that prolonged dysfunction of cerebral autoregulation during the early treatment phase was an independent risk factor for developing delirium. The authors suggested that future research should explore optimizing MAP to maintain healthy COx levels.

Building on the findings from the CONFOCAL study, Rosenblatt et al. published results examining the link between optimizing MAP based on cerebral oximetry index (COx) values and the severity of delirium [65]. In this study, six patients with sepsis-associated delirium underwent 12 h of bilateral frontal NIRS monitoring within the first 48 h of ICU admission. The severity of delirium was assessed using Glasgow Coma Scale (GCS) scores. The results showed that patients with lower GCS scores had consistently higher COx measurements, indicating greater impairment in cerebral autoregulation. Furthermore, there was significant variability in MAP values among patients, highlighting the need for personalized, continuous monitoring. The authors suggested that the standard MAP target of ≥65 mm Hg in sepsis may not be suitable for every patient and should be tailored based on individual COx data.

The first study to investigate the use of diffuse optical tomography (DOT) in delirium and sepsis-associated delirium was published in 2024 by Jiang et al. [66]. This was the largest functional neuroimaging study on delirium to date, involving 50 participants (12 with sepsis), with matched controls based on age, gender, handedness, and admission setting. A custom-designed 48-channel DOT system was developed, featuring a lighter mesh and lower-density optical fibers to enable bedside use in delirious patients, both in the ICU and in medical inpatient services. The DOT interface focused on the prefrontal cortex, particularly the bilateral dorsolateral and dorsomedial regions. To assess brain activation, the Months Backwards Test [67], validated for measuring attention, working memory, executive function, and processing speed in delirium and dementia, was administered during imaging. Patients were scanned both during their initial delirium episode and after delirium resolution, with Delirium Rating Scale-Revised-98 (DRS-R-98) [68] scores recorded at both timepoints to evaluate clinical severity. The study found significantly lower total and oxygenated hemoglobin levels during delirium, which remained low even after resolution when compared to matched controls. Additionally, lower hemoglobin levels were associated with higher, more severe DRS-R-98 scores. These findings suggest that DOT’s three-dimensional, bedside capabilities could play a key role in validating optical neuroimaging biomarkers for sepsis-associated delirium and potentially studying the long-term cognitive decline that can follow delirium.

3. Challenges and Limitations

The complexity of depth sensitivity and the penetration limits of near-infrared (NIR) light are challenging issues influenced by various factors in NIRS applications, including the technology, parameters, and neuroanatomical characteristics of the tissue being examined. These factors significantly affect how light is absorbed and scattered within the brain, with different depths exhibiting variable absorption and scattering coefficients that require specific calculations [45,46]. In commercial NIRS devices, the typical imaging depth is estimated to be around 2–3 cm beneath the scalp [69,70,71]. However, it has been reported that employing wavelengths of 808 nm or greater allows for imaging depths of 4 to 5 cm reliably [72,73,74]. Unlike standard fNIRS, DOT, with its advanced signal processing and sophisticated software, consistently shows the capability to image at these deeper depths [75,76]. Additionally, recent advancements in time- or frequency-domain detection modes offer potential improvements to the depth limitations of the conventional continuous-wave approach [77]. Despite these advances, the challenge of limited depth penetration remains a constraint for using NIR light to image deeper brain structures, which we postulate is important for fully understanding the pathophysiology of delirium. For example, subcortical regions like the thalamus and caudate are believed to play a role in disturbances of arousal in delirium, particularly the hypoactive subtype, due to their connection to the ascending reticular activating system [35].

A common challenge across all functional neuroimaging techniques is their vulnerability to motion artifacts. The subtle variations in blood oxygenation and flow associated with neural activity can be difficult to distinguish from motion-induced signal fluctuations, complicating the interpretation of hemodynamic responses in specific brain regions [78]. These artifacts can degrade image quality, skew statistical analyses, and lead to inaccurate conclusions [79]. In optical neuroimaging, movement between the optical fibers and the scalp can interfere with data collection, causing signal disruptions and fluctuations [80]. The most frequent motion artifact appears as a brief, high-amplitude intensity spike that quickly diminishes once the motion stops. In fMRI, such artifacts can render entire datasets unusable if the motion is significant during data acquisition. However, NIRS and DOT are generally more tolerant of head and scalp movement due to their flexible interface design, allowing for the capture of unaffected data segments despite brief motion [81]. This greater tolerance is likely due to differences in algorithms and the use of mesh-based headgear that adjusts to the scalp, as opposed to fMRI’s fixed setup. Traditional methods used to minimize motion artifacts include the careful design of the optode array, reducing subject movement through visual fixation, limiting stimuli, ensuring patient comfort, and applying advanced post-processing techniques [46,48]. New computational methods, extending beyond traditional filters and channel regression, include techniques such as temporal derivative distribution repair, transient artifact reduction algorithms, and dual-stage median filters. In the context of delirium research, optical imaging holds promise for examining hyperactive delirium or delirious agitation, owing to its lower sensitivity to motion artifacts. This would additionally allow for imaging such patients in a naturalistic manner (at the bedside), which would lead to a more realistic assessment of brain activity [82]. Previous research has primarily focused on hypoactive delirium, as patients with behavioral agitation often cannot endure prolonged neuroimaging sessions without inciting further agitation. The pathophysiology and neural mechanisms of different delirium subtypes likely differ considerably as they exhibit entirely different symptoms, making this a promising area for further investigation using techniques like NIRS and DOT.

Given the prominent hypoperfusion, coagulation cascade dysregulation, and oxygenation deficits in sepsis, the occurrence of brain lesions is not uncommon. Longer durations and higher severity of sepsis have been correlated with the development of white matter, ischemic, and hemorrhagic lesions [83,84,85]. These tend to occur in areas of the brain sensitive to ischemia and hypoperfusion, such as the hippocampus, basal ganglia, cerebellum, and amygdala [86]. Large enough lesions (>5 cm) may obfuscate optical imaging results given serious structural changes that could occur during sepsis. There has been some success in using fNIRS and DOT for stroke detection [87,88,89,90]; however, these data remain preliminary and have not been validated well in large-scale trials yet, especially in comparison to CT or MRI. Pairing optical neuroimaging with CT or MRI may be a strategy to overcome this specific obstacle. Otherwise, smaller lesions caused as a complication of sepsis should not be a contraindication for optical neuroimaging. Furthermore, since the cortex is the predominant region of interest captured by fNIRS and DOT currently, the presence of lesions is likely not a hindrance as they likely occur in more subcortical locations susceptible to ischemia.

4. Future Directions

Most studies to date have concentrated on NIRS, rather than on fNIRS or DOT, and this distinction is significant. NIRS typically uses a limited number of channels, which restricts data collection to a small portion of the target region. In all published studies on sepsis-associated delirium, one or two-channel NIRS was applied to the forehead, meaning the data on cerebral oxygenation and functional activity were limited to a small section of the frontal cortex. In contrast, the denser optode arrays used in fNIRS and DOT allow for the simultaneous assessment of larger brain areas and facilitate connectivity analyses, similar to fMRI [91,92]. Network analyses are essential for progressing delirium research, as it is improbable that only the frontal lobe is impacted. Previous studies using EEG, fMRI, and PET have indicated that the parietal and temporal lobes may also contribute significantly given their roles in regulating attention, sensations, memory, and language [32,34,35]. For instance, in order to map attention as a cognitive domain, multiple areas of the brain would have to be analyzed, including the medial and lateral prefrontal cortices, precuneus, and inferior parietal lobule [93]. Such a network connectivity-based approach would allow for further studies characterizing different networks involved in various subtypes of delirium and would also lead to personalized interventions targeting abnormal neurocircuitry, depending on the type of delirium and individual patient. Recently developed whole cortex fNIRS and DOT devices, featuring over 100 channels, offer an ideal solution for studying these brain regions and their network connectivity in delirium. Despite their advanced capabilities, these devices remain lightweight and portable for use in human research [54,94,95].

In clinical research, neuroimaging studies on sepsis-associated delirium and other forms of delirium have been largely constrained by their focus on hypoactive delirium. This is likely because patients with hypoactive delirium experience marked psychomotor slowing, making them easier to image. In contrast, hyperactive delirium, characterized by severe agitation and an inability to tolerate routine medical care, creates significant challenges for imaging procedures like MRI or CT scans [7]. The mixed subtype of delirium poses additional questions regarding the underlying pathophysiology and neural circuits affected, as these patients exhibit periods of psychomotor slowing intermixed with episodes of agitation as well. The brain regions and underlying mechanisms of these different delirium subtypes are likely to vary significantly, and improving our understanding of the neurocircuitry may lead to improvements in the prevention and management of delirium.

Optical imaging offers significant advantages for studying all delirium subtypes, especially hyperactive and mixed delirium, for several reasons. The ability to conduct bedside imaging allows for more naturalistic observations, avoiding the logistical challenges and tolerability issues associated with transporting critically ill patients to a designated scanner room or requiring the administration of extra sedation to facilitate imaging. This is especially important in ICU settings, where patients often have even stricter medical limitations and further administration of sedatives may prolong or worsen delirium [96]. However, it is important to acknowledge that in severe cases, standard optical imaging systems using fiber optics may still encounter significant motion artifacts. Advances in technology have introduced fiberless systems, incorporating cabled designs, flex-rigid printed circuit boards, and modular setups [94]. Although these systems may have fewer channels compared to fiber-based arrays, several wireless devices have been developed and validated for advanced functional data collection. These truly wireless systems are ideal for investigating delirium with agitation and can be used to assess the impact of interventions such as physical and occupational therapy on delirium outcomes, enabling functional analyses during activity. This could lead to more personalized and refined delirium treatment approaches.

The development of alternative optical imaging biomarkers could significantly contribute to understanding sepsis-associated delirium pathophysiology. Most optical neuroimaging studies on delirium have focused on absorption-based techniques to measure hemoglobin as the primary chromophore. However, other chromophores may also be useful in evaluating brain function. One such example is cytochrome c oxidase (COX), the final enzyme in the mitochondrial electron transport chain, which maintains the proton gradient essential for adenosine triphosphate (ATP) production, the cell’s main energy source [97]. Notably, mitochondrial dysfunction has been highly linked to sepsis-associated organ failure and has also been implicated in Alzheimer’s disease and other neurodegenerative conditions [98,99,100]. A 2019 study by Samuels et al. found that mitochondrial DNA variations were connected to both the risk of and protection from delirium in specific ethnic groups [101]. These findings suggest a new direction for using fNIRS and DOT to examine COX activity in relation to sepsis-associated delirium progression, especially considering the role of mitochondria in oxygen regulation.

Finally, resting-state and task-based approaches in optical neuroimaging require further exploration in sepsis-associated delirium. All previous NIRS studies essentially relied on a resting-state measurement, which offers the benefits of simplicity, easier baseline measurements for generalization, and the opportunity to study the default-mode network [102] (which is most active at rest and highly implicated in self-reference, social cognition, memory, and language). However, resting-state approaches involve less control over activated brain regions, a lack of behavioral correlations to specific cognitive processes, and interpretation difficulties given the presence of spontaneous physiological activity or background neural noise [103]. These weaknesses may be particularly problematic in the assessment of delirium, as patients are already unable to participate as readily in bedside neurocognitive evaluations (e.g., structured interviews and questionnaires). In the Jiang et al. DOT delirium study, a task-based approach using the Months Backwards Test was employed for brain activation. With this type of approach, brain activity can be more directly and confidently correlated to behaviors, further localized, and linked to specific deficits. Disadvantages include the requirement of patient cooperation, specific task-dependent results, a lack of generalizability, and the complexity of the design choice [104]. Both approaches have their merits, and utilizing them further in future studies would lead to the most consummate and expeditious improvements in our understanding of sepsis-associated delirium pathophysiology and neural mechanisms.

5. Conclusions

Sepsis-associated delirium remains exceedingly common yet poorly understood. Optical neuroimaging technologies hold significant promise for advancing the study of sepsis and delirium and enhancing our understanding of underlying mechanisms. Their key advantage lies in the distinctive combination of high-resolution imaging, portability, real-time data acquisition, and affordability. At present, no other imaging techniques provide all these benefits simultaneously, especially when applied in the ICU environment with patients experiencing delirium. Early research has demonstrated the effective use of NIRS and DOT in sepsis-associated delirium studies. With further investments, these methods are poised to play a crucial role in deepening our knowledge of sepsis and delirium pathophysiology and fostering the development of preventive, diagnostic, and therapeutic approaches.