Knee injuries and knee pain are common orthopaedic conditions that limit opportunities for athletes to engage in physical activity. One particularly problematic knee injury is a rupture to the anterior cruciate ligament (ACL). Affecting up to 350,000 children and young adults per year in the US alone, ACL injury has numerous short- and long-term ramifications (e.g., pain, kinesiophobia, reduced quality of life, and post-traumatic knee osteoarthritis), which warrant its consideration as a significant health problem [1,2,3]. As the ACL is a primary biomechanical stabilizer of the tibiofemoral joint, surgical reconstruction, and extensive rehabilitation (~6–12 months) are necessary to return an athlete to the field of play. In addition to ACL injury being one of the most significant injuries that impedes an athlete’s ability to participate in sport, additional knee-related conditions have similar detrimental effects. Patellofemoral pain (PFP), one form of chronic anterior knee pain, affects approximately 1 out of 4 physically active young adults and is one of the most common chronic pain conditions in the general population [4,5,6]. Interestingly, the biomechanics of PFP has been prospectively linked to ACL injury [7], with PFP and ACL injury both sharing similar negative biopsychosocial outcomes that underscore the poor long term prognoses associated with each condition. Notably, both PFP and ACL injuries in young adulthood are associated with the development of disabling knee osteoarthritis (OA) later in life [2, 8, 9].

Historically, orthopaedic knee injury and knee pain conditions have been investigated via biomechanical-, strength-, and/or peripheral nervous system-focused approaches like electromyography (EMG) [10]. However, over the past three decades, interdisciplinary pain and neuroscience research has established the central nervous system (CNS; brain & spinal cord) as a key modulator for pain processing, cognition, and sensorimotor control in musculoskeletal pain conditions (including but not limited to ACL injury [11], PFP [12], and knee OA [13, 14]). In this regard, various neuroimaging techniques have been adapted to characterize mechanisms of knee pain and injury. Primary techniques for the evaluation of central nervous system activity are functional magnetic resonance imaging (fMRI) and electroencephalography (EEG). Although these techniques have transformed our understanding of centrally-mediated processes contributing to musculoskeletal pain and injury, methodological constraints have limited their potential for generalizability (often low in ecological validity), thus precluding opportunities to directly translate EEG/fMRI findings into novel therapeutics. Namely, fMRI and EEG require individuals to keep their head still during data acquisition, thus limiting research paradigms to resting states or, at best, very limited movements (e.g., finger tapping). A primary concern with head movement during fMRI and EEG is “task-correlated motion artifact,” such that it is difficult to isolate signal generated from the manipulation of interest vs. signal that is generated from head movement itself [15, 16]. While in some fields this may not be particularly limiting, such as memory or general cognitive function whereby head motion associated with cognitive responses/fine motor control is minimal, healthy lower extremity function in its naturalistic form is integrally tied to gross motor control. Indeed, the pain neuroscience community has called for the evaluation of movement-evoked pain to better isolate neural mechanisms of musculoskeletal pathology from prior approaches (particularly painful movement of the lower extremity rather than pain self-recall, resting-state, experimentally induced pain, etc.) [17]. The purposes of this manuscript are 2-fold: (1) to summarize current approaches used to evaluate neural activity via fMRI and EEG concurrent with lower extremity movement and (2) introduce new methodologies, such as mobile, source-localized EEG to complement current technologies. The latter is expected to leverage traditional strengths associated with MRI (spatial localization) and EEG (temporal precision) to enhance characterization of neural activity during active movement (e.g., gait, squatting, upright balance).

To date, athletes who are at high risk for ACL injury, go on to experience an ACL injury, and those who undergo surgical reconstruction following ACL injury (ACLR) exhibit altered brain activity and connectivity in regions important for somatosensory, motor, cognitive, cross-modal, visual processing, among others [11, 18, 19, 20••, 21]. Prospective resting-state fMRI data in male football high school athletes who suffered a subsequent ACL injury revealed reduced connectivity between somatosensory, supplemental motor areas, and primary motor cortices compared to controls [22]. Similarly, in a cohort of female high school soccer athletes, reduced connectivity was seen in both somatosensory, and cerebellar regions [23] in athletes who went onto ACL injury. In both sexes at risk for ACL injury, decreased connectivity between sensorimotor regions may be indicative of CNS dysfunction that may elevate ACL injury risk. The prospective nature of these investigations has not only revealed disruptions within brain regions responsible for motor control, balance, and coordination, but expose potential neural biomarkers of ACL injury, which could be used to target CNS focused interventions for the prevention of future injury.

Complementing resting-state paradigms, the creation of lower extremity movement paradigms for use in MR environments have allowed for investigation of movement-evoked brain activity in healthy athletes, those at high risk for ACL injury, and athletes with ACLR [24••, 25]. One study evaluated brain activity during isolated knee flexion/extension movements in a patient with ACLR (i.e., an open kinetic chain task), just days prior to a subsequent contralateral ACL injury, and found that brain activity was greater in motor planning, sensory processing, and visual motor regions compared to a matched control [26]. Furthermore, neural activity in athletes following ACLR, as quantified via the isolated knee flexion/extension paradigm in fMRI, indicated differences in brain activity to regulate the knee joint compared to uninjured controls. Specifically, athletes with ACLR demonstrated a visual-motor brain activation strategy to move the knee, whereas control athletes maintained a sensory-motor brain activation strategy to complete the same motor task [27]. To further extend the isolated knee flexion extension fMRI movement paradigms, multijoint ankle, knee, and hip flexion/extension devices were created to further simulate lower limb movement demands associated with physical activity (closed kinetic chain movements against resistance that simulate a unilateral or bilateral leg press) [28, 29]. Longitudinal fMRI testing of the unilateral leg press in uninjured athletes revealed high consistency for activating sensorimotor regions with limited head motion artifact (~7 weeks between testing sessions), thus demonstrating its potential to isolate neural activity associated with more complex movements in other at-risk populations. Indeed, a bilateral version of the fMRI leg press revealed neural activity differences between female athletes who were at greater risk for ACL injury [20••]. Specifically, greater high-risk landing mechanics measured via traditional laboratory testing (drop vertical jump; peak external knee abduction moment) was associated with greater activity in areas responsible for spatial awareness, attention for motor control, and visual-spatial coordination. fMRI findings highlight the presence of CNS alterations prior to ACL injury that may be causally related to high-risk landing mechanics, providing a mechanistic pathway for targeted prevention strategies that restore sensorimotor dysfunction.

Most common ACL injury-reduction training involve strategies to improve athlete’s motor coordination, with a focus on limiting knee frontal plane motion and load [30]. These strategies may be limited in their efficacy due to failure to consider underlying brain activity that can predispose athletes to injury. It is evident based on prior work that alterations within the CNS may be part of the mechanistic pathway underlying ACL injury risk, specifically activity related to visual coordination and cognitive processing during dynamic movements. Identification of these alterations may be addressed during athletic training, prior to or post-rehabilitation, by using an integrated framework of neurocognitive and visual-motor approaches with the addition of more traditional neuromuscular interventions that are known to decrease risk factors for future ACL injuries [31].

The investigation of ACL and ACL reinjury risk biomarkers via the use of fMRI have aided in the development of alternative pain-related fMRI investigations for additional knee conditions. Indeed, recent advancements in neuroimaging techniques have afforded the opportunity to discover neural mechanisms associated with PFP - a prevalent multifactorial musculoskeletal condition that has maintained an elusive etiology and has limited treatment options for long term pain reduction [5]. Compared to pain-free controls, patients with PFP exhibit altered resting-state functional connectivity between brain regions important for pain, psychological functioning, and sensorimotor control [12]. Importantly, brain connectivity in patients with PFP was associated with the degree of patients’ perceived disability, dysfunction, and kinesiophobia (i.e., patient reported outcomes). Similar to the evolution of ACL injury fMRI paradigms, altered resting-state connectivity in patients with PFP warranted consideration of more ecologically-relevant fMRI manipulations to elucidate CNS dysfunction during painful movements. Female adolescents with PFP underwent brain fMRI during a modified Clark’s test, which allowed for noxious pain induction via patella pressure and voluntary quadriceps contraction [32]. fMRI results revealed activation in the frontal, parietal, occipital lobes, and cerebellum during the modified Clarke test, with greater kinesiophobia positively associated with greater activity in the cerebellar and frontal regions during this manipulation. In a related investigation, young females with PFP that underwent a passive painful movement manipulation (medial patella displacement by the experimenter) during fMRI demonstrated greater activation in sensorimotor, cognitive, and pain-related brain regions compared to resting conditions. Notably, greater perceived pain unpleasantness was associated with increased activation in the posterior cerebellum during the passive painful movement, suggesting that CNS dysfunction may be underlying maladaptive cognitive and emotional appraisal of pain in patients with PFP. Building on passive/limited movement manipulations, preliminary fMRI studies that require active knee movements (isolated knee flexion/extension) have revealed less activation in sensorimotor cortices and sensory integration brain regions for patients with PFP compared to controls [33]. Interestingly, greater kinesiophobia was related to greater visual-related brain activity for isolated knee flexion/extension in patients with PFP [34].

Cumulatively, the prior body of work using fMRI indicate that CNS dysfunction is related to psychological functioning (e.g., kinesiophobia, perceived pain) and sensorimotor control in patients with PFP. Specifically, alterations within frontal, parietal, occipital, and cerebellar brain regions are present in young females with PFP. Moreover, cognitive appraisal of pain and fear of movement appear to be modulated by cortico-cerebellar loops which may underlie a maladaptive downstream cascade that manifests as disrupted motor control. Movement-related investigations further highlight alterations within sensory systems and indicate that patients with PFP may use increased visual resources during movement to compensate for pain-disrupted sensory processing. By isolating sources of CNS dysfunction in patients with PFP via fMRI, targeted neurotherapeutic treatments may emerge for individualized interventions that can overcome the variable symptom presentation and unknown mechanistic etiology of PFP. Identification of distinct neural signatures associated with patient reported outcomes for patients with PFP may allow clinicians to tailor treatment strategies that addresses the psychobiological basis of the condition.

fMRI has provided valuable information about the neural alterations associated with ACL injury and PFP, but it possesses some inherent limitations. There is high cost associated with MRI acquisition and analyses, especially for the extended scanning periods that are required for fMRI paradigms. MRI machines are often shared among many clinicians and researchers, thus limiting necessary access to develop and optimize movement-based paradigms which can successfully acquire high quality data. Additionally, the supine positioning for fMRI movement-based paradigms limit its utility for mimicking more complex sporting movements. In summary, while fMRI with movement-based paradigms provide great utility for understanding the mechanisms of ACL injury and PFP, other neuroimaging modalities are useful to further characterize neural changes that can precipitate or follow ACL injury and knee pain.

The investigation of CNS biomarkers associated with ACL injury and knee pain with fMRI has motivated interest in complementary strategies that support enhanced temporal resolution during ecologically-valid measurement scenarios. The signal captured by EEG are thought to predominantly represent summed postsynaptic potentials generated by the intricate communication patterns of cerebral neurons [35]. EEG metrics enable the assessment of electrical brain activity and is most commonly used, in more clinically relevant/functional settings, to evaluate electrophysiological performance that can be associated with both ideal function or chronic pathology. EEG techniques have also been applied broadly to many domains where an assay of brain activity is required, such as sleep, cognition, attention, and other related areas [36, 37]. As previously mentioned, EEG allows for the tracking of CNS function with high temporal resolution. High temporal resolution of signal acquisition is particularly useful as specific patterns of time varying activity in the EEG (i.e., neural oscillations [38]) may coincide with elevated injury-risk, pain, or provide a useful indication of recovery within an athlete’s sensorimotor system [39]. Often, the oscillatory activity of the brain related to knee-related pain and movement dysfunction is considered based on the amount of energy within a specific frequency range (e.g., Alpha[8-14hz], Beta[14-30hz]). For example, EEG integrated with a knee positioning task showed that when athletes with ACLR attempted to gauge the position of their previously injured leg, they exhibited higher Theta band (4–8hz) activity in the frontal regions of their brain and higher Alpha-1 (8–8.9hz) activity in parietal regions relative to healthy controls [40]. In addition, the presence of aberrant frontal Theta activity in athletes with ACLR was further confirmed using a force production task [41]. Alterations to the amplitude of Theta and Alpha frequency components in these studies is thought to represent the increased attentional resources that are used by individuals when moving their previously injured leg.

EEG has also been used to quantify alterations to the CNS after ACL injury and reconstruction by implementing diverse analysis techniques and experimental paradigms. Important innovations include the synchronized recording of EEG and EMG, and advanced nonlinear analyses that go beyond standard power spectral analyses. Combined analysis of EEG and EMG signals represents a promising direction for the field because it allows for deficits in CNS function in those with prior ACL injuries to be directly linked to muscular function and, by extension, to potential deficits in real-world motor activity that would lead to re-injury. Synchronized EEG and EMG recording, during a maximum voluntary quadriceps contraction task, exposed deficient corticomuscular (i.e., brain-to-muscle) communication patterns in individuals with ACLR relative to uninjured controls [42••]. Specifically, corticomuscular coherence (CMC), a measure of shared frequency content in two signals [43], was calculated between a group of EEG electrodes and EMG electrodes placed on the vastus medialis and lateralis. Individuals with ACLR exhibited lower coherence between EEG and EMG in the Gamma frequency band (31–80hz) in both their involved and non-involved leg when compared to control participants. This study represents an important step towards understanding holistic neuromuscular consequences of ACL injury. However, CMC, like other frequency-based measures, only expose one aspect of neural activity. In an attempt to characterize non-linear CNS activity [44] related to ACL injury risk, recurrence quantification analysis [45] was implemented on EEG signals recorded during resting state and indicated that athletes who were at high risk for ACL injury, based on a drop vertical jump (DVJ) assessment, exhibited more deterministic brain activity [46]. This finding was interpreted as meaning that athletes who go on to injure their ACL may do so because their brain is unable to switch its behavior during high demand competition in a flexible manner and therefore is unable to adapt to anticipated perturbations to planned movements, putting them at risk of injury.

Although little EEG based research has yet been carried out concerning PFP, there have been a few recent studies that support the feasibility and potential impact of EEG to evaluate the physiology and treatment of PFP. One study combined EEG with a novel Virtual Reality (VR) based therapy to test the minimization of PFP as measured by standard clinical, and electrophysiological metrics [47]. Participants with PFP were randomly assigned into VR intervention and or control conditions. At baseline, no meaningful baseline differences were observed between the two groups. After 24 sessions of VR over 8 weeks, the intervention group increased Theta and Alpha EEG activity when compared to the control group. Another recent study explored the reliability and internal consistency of EEG measures in those who suffer from chronic knee pain across multiple sessions under resting state conditions and during a pain management activity [48]. Findings from this study indicate that alpha frequency band measurements exhibit high internal consistency and moderate to high re-test reliability between time points. Cumulatively, these data demonstrate that EEG has the capacity to capture pain related individual differences in CNS activity, at least within the Alpha frequency band. While further investigation is certainly required, this is a promising finding that supports EEG as useful technique for studying neurological factors of PFP.

Often, including the EEG studies described here, EEG signals recorded at the scalp are used as the basis of their final analysis. While this is the most straightforward method, the lack of anatomical localization in these studies create difficulties in relating results to prior fMRI research, which is capable of localizing findings with high spatial precision. To synergize with prior fMRI research and expand our understanding of the complex neurophysiological mechanisms involved with knee injury and pain, future EEG studies would need to implement source localization techniques [49]. Towards this end, one recent study did successfully localize EEG sources from individuals with ACLR and healthy controls during a single leg balance task, indicating decreased sensorimotor processing, increased demands on motor planning, and greater motor inhibition [50]. This was interpreted by the authors as supporting the view that after ACLR, individuals have a reduced ability to automatically control basic movements (e.g., in postural control) without visual or higher cognitive input. Independent component analysis (ICA) was used to spatially localize dipoles that were most likely to have produced the electrical activity recorded at the scalp. While this work is certainly moving the field in a valuable direction, our recommendation, as outlined in the next section, is that techniques like ICA should be instituted more broadly, and where possible, in conjunction with other modalities like MRI. Beneficially, adding MRI based modeling techniques allow the potential to target key cortical and subcortical structures with unprecedented detail [51].

Traditionally, fMRI can resolve hundreds of thousands of spatially unique measurements of brain activity but is limited to sampling slowly—about one measurement every half of a second [52]. By contrast, modern EEG can sample brain activity thousands of times per second but is limited to roughly 250 sensors, placed on the surface of the scalp [53]. While EEG enables the analysis of brain activity with very high temporal resolution, it shares similar constraints related to head motion artifact as fMRI. One significant problem that can arise is the addition of electrical activity from muscles in the neck, face, eyes, and heart that can mask or obscure the brain signals of interest. In addition, mechanical artifacts arising from movement of the electrodes themselves can make it difficult to obtain clean data from scalp EEG recordings during periods of movement. In the next section, potential ways to overcome these issues will be described such that MRI grounded EEG may be used successfully to localize sources of brain activity, with minimal noise and in conjunction with movement. Again, we believe that studying CNS activity during dynamic movement will be essential for the future of scientific inquiry and the development of strategies for clinical care. Therefore, we will outline a recommended experimental set up and processing pipeline (see also Fig. 1) that would yield a high-quality evaluation of the neural mechanisms of ACL injury and PFP.

A flow diagram that shows each step of data collection, processing, and analysis for a mobile, source localized EEG experiment meant to assess the neurological mechanisms of ACL injury and PFP