Clinical and Research MRI Techniques for Assessing Spinal Cord Integrity in Degenerative Cervical Myelopathy—A Scoping Review

Background: Degenerative cervical myelopathy (DCM) manifests as the primary cause of spinal cord dysfunction and is non-traumatic, chronic and progressive in nature. Decompressive surgery is typically utilised to halt further disability and neurological dysfunction. The limitations of current diagnostic options surrounding assessment and prognostic potential render DCM still largely a clinical diagnosis. Aims: To outline the limitations of current diagnostic techniques, present evidence behind novel quantitative MRI (qMRI) techniques for assessing spinal cord integrity in DCM and suggest future directions. Method: Articles published up to November 2021 were retrieved from Medline, EMBASE and EBM using key search terms: spinal cord, spine, neck, MRI, magnetic resonance imaging, qMRI, T1, T2, T2*, R2*, DTI, diffusion tensor imaging, MT, magnetisation transfer, SWI, susceptibility weighted imaging, BOLD, blood oxygen level dependent, fMRI, functional magnetic resonance imaging, functional MRI, MRS, magnetic resonance spectroscopy. Results: A total of 2057 articles were retrieved with 68 articles included for analysis. The search yielded 2 articles on Quantitative T1 mapping which suggested higher T1 values in spinal cord of moderate-severe DCM; 43 articles on DTI which indicated a strong correlation of fractional anisotropy and modified Japanese Orthopaedic Association scores; 15 articles on fMRI (BOLD) which demonstrated positive correlation of functional connectivity and volume of activation of various connections in the brain with post-surgical recovery; 6 articles on MRS which suggested that Choline/N-acetylaspartate (Cho/NAA) ratio presents the best correlation with DCM severity; and 4 articles on MT which revealed a preliminary negative correlation of magnetisation transfer ratio with DCM severity. Notably, most studies were of low sample size with short timeframes within 6 months. Conclusions: Further longitudinal studies with higher sample sizes and longer time horizons are necessary to determine the full prognostic capacity of qMRI in DCM.


Epidemiology
Degenerative cervical myelopathy (DCM), earlier referred to as cervical spondylotic myelopathy (CSM), is the most common cause of spinal cord dysfunction, affecting an estimated 5% of adults over 40 years of age [1]. It is a significant cause of disability and carries substantial economic burden to the affected individuals, including their families and their community [2].
A comprehensive review of the literature demonstrated that such degenerative conditions of the spine are estimated to encompass 59% of non-traumatic spinal cord in-jury in Japan [3], 54% in the United States [4], 18-26% in Australia [5][6][7] and 16-39% in Europe [8][9][10][11][12][13][14]. The incidence was also purported to be 76, 26 and 6 per million in North America, Europe and Australia, respectively [7]. Notably, these data exclude many patients with less severe symptoms. Very few studies have been conducted on DCM prevalence. In Victoria, Australia, the prevalence of all non-traumatic spinal cord injury was estimated to be 367.2 per million in 2010 [15]. On the basis of these figures, studies have estimated the incidence and prevalence of DCM in the United States to be 41 and 605 per million, respectively [16].
However, a constraining factor of these estimates are the poor quality of the data of which they are derived from, and it is likely that the figures severely underestimate the burden of disease. As one of the most common causes of non-traumatic spinal-cord injury it is reasonable to infer that DCM represents a formidable issue in the aging population [17].

Natural History
AO Spine RECODE-DCM has recently listed the diagnosis and pathogenesis, as some of the top priorities in DCM research [18]. Degenerative cervical myelopathy is a degenerative condition and as such, it is non-traumatic, chronic and progressive in nature, with surgery traditionally utilised to halt further disability and neurological dysfunction [19]. The pathogenesis of DCM is purported to involve a myriad of static and dynamic factors ( Figure 1). Static factors include spinal canal compression, spine deformity, disc herniation, osteophyte formation, ossification of the posterior longitudinal ligaments (OPLL) and ossification of the ligamentum flavum (OFL). Owing to its mobility, the vertebral column of the neck also suffers from dynamic stressors that include biomechanical changes, invagination of the ligamentum flavum and microstructural mechanical spinal cord damage from cervical instability. Such stressors, at a chronic magnitude, induce direct neuronal and glial cell damage as well as a secondary ischaemic cascade of neuronal excitotoxicity and apoptosis which contribute to the development of DCM [17].
Notably, very little is currently known with regards to the exact molecular mechanisms underlying the process of this condition. A 1963 retrospective study of the natural history of DCM found that a majority of patients had poor prognosis, 87% of which progressing to moderate or severe disability at the last follow up. Several historical and present day studies have indicated that the disease course of DCM is highly variable [20][21][22][23][24][25][26][27][28][29][30][31][32][33]. In particular, a 1956 study of 120 DCM patients and found 5% patients had a rapid onset of symptoms followed by long periods of quiescence, 20% had a slow, progressive deleteriousness of neurofunction, and 75% had a stepwise decline of neurofunction [21]. A further 2017 systematic review demonstrated that 20-67% of patients had experienced neurological deterioration after 3-6 years of follow-up [34]. It is not yet clear what manifests these differences in DCM pathogenesis between patients. Further research into DCM at a molecular level could result in promising diagnostic methods, enable detection at early stages and thus render timely intervention and treatment.
The current data associated with the natural history of DCM is largely derived from low-quality retrospective studies [19,35]. The limited existing prospective studies are markedly underpowered and have low level evidence with inconsistent results and risks of bias [36]. Accordingly, there persists a necessity for a large-scale prospective study focusing on natural history of DCM, specifically with the inclusion of novel multiparametric quantitative MRI that will be discussed further on.  Where novel qMRI would fit into the current diagnostic work-up and treatment of degenerative cervical myelopathy. The dashed lines represent pathways currently under investigation. The current diagnostic work-up starts with a detailed history, physical examination, and application of scoring systems. Electrophysiology and imaging can rule out differentials and provide additional information to generate a diagnosis of DCM or cervical spinal cord compression without myelopathy. Surgery is recommended for moderate to severe myelopathy according to current guidelines. Patients with spinal cord compression and evidence of radiculopathy may be offered surgical or non-surgical treatment. Patients with spinal cord compression and no evidence of radiculopathy should undergo clinical monitoring. Surgery may be offered to patients utilising non-operative treatment upon worsening of condition. There is potential for qMRI to play a role in monitoring this progression and provide prognostic value to the outcome of DCM. Abbreviations: CMS, cervical myelopathy score; CT, computed tomography; DCM, Degenerative cervical myelopathy; DTI, Diffusion tensor imaging; EMG, electromyography; EMS, European myelopathy score; ENG, electroneurography; fMRI, functional MRI; MEPs, motor evoked potentials; mJOA, modified Japanese orthopaedic association score; MRI, magnetic resonance imaging; MRS, magnetic resonance spectroscopy; MT, magnetization transfer; NCS, nerve conduction studies; NDI, neck disability index; NPRS, numeric pain rating scale; qMRI, quantitative magnetic resonance imaging; qT1, quantitative T1; qT2, quantitative T2; SEPs, somatosensory evoked potentials; SWI, susceptibility weighted imaging. Table 1. Typical presenting symptoms and physical signs in DCM [2,[37][38][39][40][41][42][43].

Presenting Symptoms Physical Signs
Neck -Pain and/or stiffness -Decreased cervical range of motion - The Lhermitte phenomenon [ Limitations: Although a useful element of diagnosis, physical tests are not always consistent in their ability to quantify the severity of DCM which is an important element in the consideration of surgical intervention [44]. Thus, correlations in further investigatory measures are required to arrive at the correct diagnosis.

Scoring Systems
Clinicians utilise scoring systems to categories the functional impairment of various conditions. Whilst different classifications may arise, there typically exists one standardised system for publications and treatments. This is not true for DCM; whereby different systems are utilised based upon preference. A 2016 systematic-review revealed that reported outcomes varied widely between studies of DCM [45]. Table 2 details the current most common classification systems, their benefits and limitations. The mJOA scale followed by the Nurick Grading system are the current most widely adopted measure for DCM patients [46]. However, the limitations covered in Table 2, are particularly problematic in mild DCM whereby strong floor and ceiling effects [e] in these scales inhibit ascertainment of more subtle neurological changes that provide information for decision-making in surgery18. As such, there exists the need to develop both a standardised scoring system and more sensitive and objective outcome instruments to enable more effective clinical assessment and efficient synthesis of research. Table 2. Common classification systems used for DCM [17,[47][48][49][50][51][52][53][54][55][56]. See Appendix A (Tables A1-A7.)

Conventional MRI
Conventional MRI is the primary modality utilised for imaging in DCM as it enables high-resolution depiction of neural structures, bone and ligaments that are difficult to visualise in other scans [57]. Conventional MRI (such as T1-weighted and T2-weighted imaging) can characterise the degree and nature of degeneration (i.e., OPLL, spondylosis, disc herniation, hypertrophy of ligamentum flavum), identify spinal-cord compression, highlight changes in spinal-canal diameter, and detect changes in signal intensity [58][59][60]. MRI can also assist in ruling-out resembling differentials or other causes of myelopathy such as a tumour syringomyelia or demyelinating plaques [2,38,42]. CT myelography should be utilised in situations of MRI contraindication [61].
Identifying spinal-cord compression plays a pivotal role in treatment selection and outcome prediction and thus should be the foremost investigation. It is typically described based on the number of compression sites [30], appearance [32,[62][63][64][65] or ratio between the anteroposterior diameter and the transverse diameter (CR = Compression Ratio) [66,67]. A maximum spinal-cord compression (MSCC) index has also been developed by Fehlings et al. as a measurement of spinal-cord compression [68]. The primary object of these methods is to determine severity of spinal-cord compression.
Measurements of the anterior-posterior diameter at the region of interest (ROI) can be undertaken to evaluate the degree of spinal-stenosis [30]. Similar to MSCC, Fehlings et al. have developed a protocol to assess the maximum canal compromise (MCC) post-traumatic cervical spine-injury [69]. This has been additionally utilised for degenerative conditions and functions by calculating the canal size at the ROI and analysing it in conjunction to the average canal size for levels above and below. Multi-level signal-intensity changes are suggestive of necrosis or cavitation in the spinal cord and lend to poorer surgical outcomes [70][71][72][73]. T2-hyperintensity in conjunction with T1-hypointensity is associated with greater clinical deterioration when compared to T2-hyperintensity alone due to signal changes in T1-weighted images indicative of more permanent insult [70,[74][75][76][77].
Limitations: Findings on conventional MRI do not typically correlate well with the variable clinical presentations of DCM [42]. Although spinal-cord compression is a sensitive marker of myelopathy [78], approximately 5% of asymptomatic patients also present with it [42], thereby limiting its specificity. The supine patient positioning in conventional MRI hinders its utility in assessing alignment, providing only a superficial assessment for situations in which upright films are not available [79]. Conventional MRI is intrinsically limited in its capability to characterize tissue injury in the spinal-cord because of the lack of specificity in T1/T2WI signal-change and cannot highlight specific pathophysiological processes at a cellular level (demyelination, axonal loss, inflammation, oedema, gliosis and apoptosis) [57]. It also is not a good predictor of neurologic function before/after surgical intervention and has low sensitivity for structural spinal cord change in cervical myelopathy [57,70,[80][81][82][83]].

Plain Radiographs and Computed Tomography (CT)
Computed Tomography is useful for the study of bone anatomy and can aid in cases where spinal-fusion is being considered as a treatment. In cases where MRI is contraindicated (such as the presence of pacemakers or other internal metallic objects), CT is a valuable imaging alternative. Plain radiographs can provide useful information about spinal-canal stenosis, degenerating discs, degenerating joints, OPLL, vertebrae fusion, cervical-spine alignment and subluxation [2,38,84,85]. This can reveal scoliosis and loss of physiological cervical-lordosis and kyphosis. Lateral-films in cervical-flexion and extensions are utilised to evaluate instability of the cervical-spine. DCM patients frequently showcase increased C2-C7 Cobb angles, upper T1 slopes, lower C7 slopes and upper C7 slopes [86].
Limitations: Computed tomography suffers the same inability to characterise tissue injury that conventional MRI does [57]. In addition, a 2017 systematic-review found that the overall strength of evidence regarding the predictive value that CT parameters have for the clinical presentation or outcome of DCM is low [87]. There is also the issue of radiation exposure. Overall CT and plain radiographs play a more complementary role in DCM diagnosis, acting as an alternative to MRI and aiding in surgical-planning [88].

Electrophysiology
Several studies have indicated good correlation between electrophysiology and the severity of myelopathy, presenting it as a reliable predictor of surgical-outcomes [89]. Somatosensory evoked-potentials (SEPs) and motor evoked-potentials (MEPs) can be, respectively, utilised to detect central sensory conduction impairment and prolonged motor latency in DCM [2,89,90]. They are also useful in detecting subclinical degenerative spinalcord compression in asymptomatic patients and are thus useful in early identification of patients likely to develop myelopathy [91][92][93][94]. Feng et al. reported a correlation between the SEP and a declining mJOA (a more severe deficit) in an investigation of progressive myelopathy [95]. Needle electromyography (EMG) is a highly sensitive indicator of anterior horn cells damage, which occurs due to compression and ischemia in DCM [96]. Nerve-conduction studies can also be used to rule out peripheral neuropathy and nerveentrapment [2]. These techniques also allow other neuromuscular diseases that can mimic DCM to be ruled out (motor neurone disease, ALS) [97,98]. Apart from aiding in diagnosis and preoperative evaluation, electrophysiology facilitates longitudinal assessment. Capone et al. found that a decrease in central-motor conduction time for the tibialis-anterior muscle correlated with an increased mJOA score post-surgery. It therefore concluded that the beneficial effects of spinal-cord surgery could be detected with MEP, making it a useful tool in determining efficacy of post-operative rehabilitation [99].
Limitations: Electrophysiology provides no anatomical information and thus cannot determine the exact location of the lesion [100]. Although some evidence exists to justify the effectiveness of electrophysiology in predicting operative outcomes, the area remains to be better defined. A systematic review found a decrease in electrophysiology publications compared with other domains of DCM, suggesting a declining interest in this area [101]. Additional studies would be required before it can be universally recommended.

Novel qMRI Modalities and Parameters
The limitations of current diagnostic options render DCM still largely a clinical diagnosis [17], making it necessary to develop and further research on novel diagnostic options with objective quantitative measures.
Advanced novel MRI protocols have been developed for the spinal cord that allow for acquisition within 45 min [102]. This involves direct measurement of spinal cord tissue changes, demyelination, axonal-injury and atrophy and thus renders the attainment of quantitative microstructural sequences now possible in the context of DCM. Such qMRI sequences and their derivable quantitative metrics are highlighted in Table 3. These derived metrics are highly sensitive to the myelopathic progression and can allow for the realisation of subclinical tissue-damage in patients with asymptomatic cervical-cord compression [103][104][105][106][107]. Quantitative metrics derived from DWI, such as DTT and DTI have been found to be more valuable when compared to conventional MRI scans in aiding diagnosis and outcome prediction in patients with DCM [108,109]. These qMRI sequences will be expanded upon later in this review. As an emerging field, the development of more advanced imaging techniques may potentiate in superior diagnostic tools, improved correlation with impairment and long-term predictions of DCM outcomes. SWI/QSM Sensitive to compounds that distort the magnetic field and alter phase of tissue and is therefore commonly used to detect blood products/haemorrhage and calcium Tissue susceptibility Abbreviations: ADC, apparent diffusion coefficient; BOLD, blood oxygen level dependent; DBSI, diffusion basis spectrum imaging; DCM, degenerative cervical myelopathy; DTI, diffusion tensor imaging; DTT, diffusion tensor tractography; DWI, diffusion weighted imaging; FA, fractional anisotropy; FC, functional connectivity; fMRI, functional magnetic resonance imaging; MD, mean diffusivity; MRS, magnetic resonance spectroscopy; MT, magnetisation transfer; MTR, magnetisation transfer ratio; QSM, quantitative susceptibility mapping; SWI, susceptibility weighted imaging; VOA, volume of activation.
[f] Fractional anisotropy (FA): Water molecules diffuse differently along tissues depending on its type, integrity, architecture, and presence of barriers. Fractional anisotropy is a value between 0 and 1 which indicates the degree to which diffusion of water is limited to one axis; [g] Apparent diffusion coefficient (ADC)/mean diffusivity (MD): measures of the average magnitude of water diffusion within a tissue.

Objective
A scoping review was conducted in order to systematically map the research done in this area, as well as to identify any existing gaps in knowledge. The following research question was formulated: 'What is known from the literature about existing clinical and novel research MRI techniques for assessing spinal cord integrity in patients with Degenerative Cervical Myelopathy (DCM)?'

Data Sources
Articles published up to November 2021 were retrieved from three main databases: Medline, EMBASE and EBM. Combinations and variations of keywords were used to conduct a comprehensive search: spinal cord, spine, neck, magnetic resonance imaging, MRI, qMRI, T1, T2, T2*, R2*, DTI, diffusion tensor imaging, MT, magnetisation transfer, SWI, susceptibility weighted imaging BOLD, blood oxygen level dependent, fMRI, functional MRI, functional magnetic resonance imaging, MRS, magnetic resonance spectroscopy.

Selection Criteria
Papers to be included required a focus on quantitative MRI techniques for assessing spinal cord integrity in patients with DCM. Exclusion criteria included any articles focusing on non-myelopathic diseases, flexion induced myelopathy, conventional MRI techniques only, non-degenerative causes of myelopathy, and studies on asymptomatic patients only. Reviews and non-English articles were also excluded from this study. A detailed flow diagram of this method can be seen in Figure 3.  Figure A1. for database search strategy). Adapted from PRISMA Scoping Review protocol [121]. Abbreviations: MRI, magnetic resonance imaging.

Synthesis of Results
Following the process detailed in Figure 3., after studies were included for qualitative review detailed were collected in a tabular format (see Appendix C- Table A8). Details were then summarised (see Table 4 in Section 3) and underwent discussion and critical appraisal in Section 4.

Results
A total of 2055 articles were identified using the search strategy outlined in Figure 3 from the three databases. There were 283 duplicates removed by automation and 99 removed manually. The remaining 1770 records were then screened by title and abstract according to the inclusion and exclusion criteria leaving 155 articles to be screened for full-text eligibility. A detailed evaluation rendered 68 articles included for analysis. The complete database search strategy can be found in Appendix B ( Figure A1). Quantitative T1/T2 mapping calculates the T1 (spin-lattice or longitudinal relaxation) time or T2 (spin-spin or transverse relaxation) time of a certain tissue and displays them on a parametric map. This is in contrast to conventional T1/T2-weighted MRI, which displays differences in T1/T2 relaxation times of tissues as a hyper/hypointense image with limited quantitative output [186,187]. T1 is influenced by the tissue's water, lipid, protein and iron content which explains its sensitivity to pathological microstructural changes in the spinal cord [125,186]. T2 is influenced primarily by the water content of the tissue and thus increases in T2 is associated with increased water content pertaining to increased disc water or glycosaminoglycan content in the spinal-cord [187].

Application in DCM
Despite being a fundamental parameter in MR imaging, T1/T2 mapping is not traditionally frequented in clinical practice due to its lengthy scan times. However, recent advances to speed of acquisition have enabled T1/T2 to become another viable option in the analysis of the pathological spinal cord [120,[188][189][190][191]. Notably, T1 has been utilised in a multitude of brain studies to investigate changes in white and grey matter ratio in both a physiological and pathological context of which has demonstrated an interesting clinical correlation with cognitive decline [192,193]. Much of the literature surrounding its use in the spinal cord have highlighted reasonable scan times and reproducible methods to measure the T1 values in the healthy cervical spinal cord [120,194]. T2 relaxation times have been shown to correlate with lumbar disc degeneration [195][196][197], however our literature search reveals no studies conducted on the cervical spine. It thus pertains that T2 may potentiate as a useful biomarker for analysing spinal-cord integrity in DCM and so future research combining the two may prove beneficial.
From the conducted literature search, there exists only two studies applying T1 to the degenerative cervical spinal-cord which utilised 2D single section (inversion-recovery) radial-gradient echo [151] and 3D-MP2RAGE [125] sequences. The former study demonstrated that the difference in T1 relaxation times between stenotic segments and non-stenotic segments above and below were longer in higher grades of stenosis. It found that absolute T1 values were higher in high-grade stenosis but found inconclusive correlation between lower grades of stenosis [151]. The latter study utilising 3D-MP2RAGE found higher T1 values in the overall spinal-cord and the level of compression of patients with moderatesevere-DCM compared to healthy controls [125]. Overall, correlation findings of T1 and clinical severity are variable and inconclusive, particularly for patients with mild DCM which still maintain as the most difficult group to diagnose. Both studies are limited by their power and cross-sectional design. Further longitudinal studies with higher-sample sizes should be conducted particularly with respect to post-operative outcome to reach more conclusive results.

Diffusion Tensor Imaging (DTI) 4.2.1. Principles
Diffusion tensor imaging is a sensitive measure of tissue microstructure that works by measuring the diffusion of water molecules. Water in a glass of water for example would be considered isotropic, the diffusion would be the same in every direction. The diffusion of water in tissue however would be considered anisotropic, whereby the diffusion will vary with direction. This would depend on the tissue type, integrity, architecture and presence of barriers. Fractional anisotropy (FA) (a quantifiable parameter of DTI) is a value between 0 and 1 that indicates the degree to which the diffusion of water is limited to one axis. Notably in a healthy spinal cord, the axons largely run in in one direction-commonly analogised to a bundle of straws. In DCM, the axonal integrity is compromised and by first principles, fractional anisotropy would be reduced. Other quantifiable DTI parameters include mean diffusivity (MD) or apparent diffusion coefficient which are both measures of the average magnitude of water diffusion within a tissue [102,118].
Therefore, DTI can be considered as not only a complementary diagnostic evaluation, but as a vital tool in the diagnosis of DCM and an early identifier of the candidates best suited to surgery. It is important to note however that as an emerging field, many existing studies [198,199,[201][202][203] are of low-sample size and could suffer from potential bias from study design limits, patient selection bias or lack of blinding when examining new technologies. Studies utilising 1.5T-MRI as opposed to 3T-MRI are limited by poorer performance and resolution [158,201]. Additionally, some studies [134,158,177,183] are limited by their cross-sectional nature (limiting their ability to predict disease progression) and could benefit from more longitudinal components.
From our conducted literature search, only 3/43 articles were of a prospective longitudinal design with a DCM cohort size of greater than 50. A total of 26/43 of the studies were of a cross-sectional design, and a majority of the longitudinal design studies had a short follow-up period of 3-6 months. As such there exists a need for additional large-scale longitudinal prospective studies to be conducted on DCM utilising DTI methods with longer time horizons and larger sample sizes. DCM is a chronic progressive disease and as it stands there exists no study that correlates longer-term progression (of over 3 years) with DTI parameter scores. Such quantifiable long-term studies could assist in identifying the characteristics of groups at risk of progressing deleteriously, thus contributing to the identification of patients who may benefit from early intervention. Further, in a prospective case control study, that slowed down due to COVID-19, we have standardised protocols (Table 5.) and successfully extracted data from healthy controls ( Figure 4). This work will likely shed light on the spectrum of DCM when completed. Table 5. Preliminary data obtained from our initial incomplete study, showcasing FA and ADC ROI scores of a healthy recruit. Data obtained through a standardisation protocol of DTI using 3T MAGNETOM PRISMA MRI scanner with a rFOV ZOOMit sequence with 4 averages (dynamic excitation for selective centrally excited field of view). We expect FA to be lower in recruits with DCM.

Measurements
Min  Functional-MRI (fMRI) is based on the BOLD contrast mechanism first introduced in 1990 and is a non-invasive technique that allows for the detection of neuronal activity. The fundamental principle behind the BOLD mechanism is that metabolic oxygen demand after neuronal activation causes a rise in blood flow and blood volume. This blood supply surpasses the actual oxygen needs which renders a transient rise of oxyhaemoglobin in the venous compartment and a relative decrease in the concentration of deoxyhaemoglobin. As deoxyhaemoglobin has paramagnetic properties, the change in deoxyhaemoglobin can be measured with MRI [112,204]. In DCM, fMRI has primarily been utilised to measure the functional connectivity (FC) and volume of activation (VOA) of regions of the brain before and after surgery to potentiate correlation [123,126,129,133,136,140,150,160,164,169,[172][173][174]180].

Application in DCM
The brain resting-state fMRI has been proven to be successful in differentiating DCM patients from healthy patients [123,126,150,173,174]. In contrast to the aforementioned MRI techniques which measures structural damage/integrity within the conduction pathways, BOLD fMRI measures the functional activation within the brain of which incorporates information collection, interpretation and distribution for all motor and cognitive functions. As a disease with a plethora of motor and sensory symptoms (see Table 1) DCM-associated information would be distributed to widespread areas of the brain [126,173]. As such, a multitude of studies have been performed to determine if fMRI can predict neurological recovery post-decompression surgery. Our literature search revealed that changes in FC strength between different areas of the brain appear to be associated with neurological improvement post-surgery in DCM. Numerous studies found an increase in VOA or FC strength of the pre/postcentral gyrus and SMA (supplementary motor area) following decompression surgery [123,126,133,140,164,174,180]. Functional connectivity alteration between the thalamus and cortex were also demonstrated [160]. Positive correlation of these findings with functional recovery assessed using various DCM grading questionnaires enables fMRI to indirectly assess spinal integrity in DCM patients.
Notably, Takenaka et al. found a positive correlation between post-operative improvement in the 10 s test ( [h] The 10 s test: the number of cycles the fingers can repeatedly grip and release in 10 s) and FC of three visual areas and the right superior-frontal gyrus in DCM patients, of which may enable the construction of a predictive formula for recovery potential [173]. Such a connection between visual cortices and DCM was also established in other studies [129,136]. Furthermore, an additional study by Takenaka et al. determined that resting-state amplitude of low-frequency fluctuation could function as a potential prognostic biomarker for DCM [172]. However, their two studies were limited by the use of mass univariate analyses which can only simply measure association. Given the multi-variable nature of fMRI, univariate analyses may miss information associated with DCM pathology and so studies using multivariate patten analysis should be conducted. Moreover, much of the research in the area is preliminary with studies of low sample sizes. Future external validation studies would be necessary for the proposed predictive formulas. None of the prospective studies have looked at the long-term use case of fMRI (over 6 months), thus development is also needed in this regard. The underlying principle behind MRS is that a proton experiences a slightly distinct magnetic field of which is dependent on its chemical environment. Reliable quantification of metabolites utilising 1.5T MRI scanners have been traditionally limited to N-acetyllaspartate (NAA), choline (Cho) and creatine (Cr). However, recent advances in imaging technology and 3T MRI scanners have enabled measurement of glumatate-glumatine (Glx) and myoinositols (Ins) [205]. N-acetylaspartate, despite not being a disease-specific marker is a sensitive indicator of axonal integrity due to its ability to be detected early in the disease course. Typically, NAA is expressed as an absolute value or as a ratio with Cho/Cr. The NAA/Cr ratio is generally viewed as a better ratio due to the more constant levels of Cr in the nervous system. However, changes in Cho are also believed to reflect increases in membrane phospholipids due to myelin breakdown from demyelinating diseases [113,118].

Application in DCM
Cross-sectional studies have determined that the Cho/NAA ratio is higher in patients with DCM compared to healthy controls [171] and is significantly correlated with mJOA score [135,163]. As such it provides a potentially clinically useful biomarker for the management of DCM. Ellingson et al. utilised both DTI and MRS data in a combined linear model. The results of this optimised model showcased a higher accuracy in predicting mJOA than DTI and MRS alone [135]. Thus, MRS could find utility in tandem with DTI as a predictive tool. Interestingly, Kowalczyk et al. found that cortical levels of NAA/Cr could also serve as a meaningful biomarker in DCM [147,148]. Nagashima et al. investigated alternative metabolites (lactate, alanine, acetate, glutamate, pyruvate and citrate) and found no significant differences between the myelopathic and control group [156]. The main limitations in the area of MRS is that MR spectroscopic data within the spinal cord is quite difficult to acquire reliably due to patient motion, spinal cord movement (due to the pulsatile flow of CSF) and the difficulties associated with magnetic shimming ( [i] Magnetic shimming: the process by which the main magnetic field is made more homogenous) [135,147,148,156,163,171].
Overall, research of MRS application in DCM is quite limited with no new research being conducted in the last six years. Of the articles identified from our literature search, all were cross-sectional and of low-sample size. Further longitudinal work should be done to assess the prognostic potential of MRS in DCM.

Principles
Magnetisation transfer is a contrast mechanism that relies on the interaction between macromolecule bound hydrogen-protons, namely lipids and lipoproteins, and the freeprotons (in free water) normally imaged by MRI. As such MT is able to indirectly probe proteins/lipids. The derived parameter, the magnetisation transfer ratio (MTR) reflects the portion of bound protons. Thus, MTR can be utilised as an indirect marker of demyelination and axonal loss as the MT effect indicates the relative density of protein/lipid macromolecules. This ability to measure myelin and axonal loss in vivo allows for application of MT to demyelinating diseases and degenerative diseases like DCM [115,118].

Application in DCM
MTR has been well-established as a marker of myelin integrity in diseases such as multiple sclerosis [206] and has been shown to correlate with histopathological myelin loss [207]. MT imaging also presents an advantage over diffusion-based imaging in the form of higher signal-to-noise ratio and higher spatial resolution [170]. From our literature search, both Cloney et al. and Suleiman et al. found a negative correlation of MTR with severity of DCM (measured via the mJOA [130] and Nurick score [170], respectively), with pathological patients tending to have a decreased MTR compared with a healthy population. Such could be indicative of DCM associated myelin degradation [130,170]. However, Serbruyns et al. conducted a study that noted a decrease in MTR with aging [208]. The correlation of this with poorer functional tasks suggests that demyelination is associated with age-related decreases in functionality. As DCM is an elderly associated disease, this difficulty of determining causation means that MTR should be primarily interpreted as a quantitative measurement of demyelination from any cause, not just DCM. Paliwal et al. have also determined the prognostic potential of MTR for assessing response to surgery and recovery of DCM patients.
Perhaps the primary shortcoming of the current studies involves the small sample sizes and the limited number of prospective longitudinal studies. Future direction in this area could involve studies of higher sample sizes determining prognostic potential, utilisation of multivariate analysis as opposed to linear correlation, and longer follow up periods to track continued improvement beyond 6 months. 4.6. R2* or 1/T2*-A Promising Biomarker 4.6.1. Principles R2* MRI measures the 'observable' or 'effective' T2 (termed T2*) whereby R2 * = 1 T2 * . T2* primarily results from inhomogeneities in the main magnetic field as a result of susceptibility-induced field distortion produced by the tissue placed within the field. In the presence of tissue iron, T2* relaxation time shortens and thus R2* increases (as R2 * = 1 T2 * ). Thus R2* represents a quantifiable measure of tissue iron content, notably via deoxyhaemoglobin, hemosiderin or methemoglobin in tissues and lesions [209,210].

Role of Iron in Neurodegenerative Disorders
Homeostasis of heavy metals, such as iron and calcium are critical for cellular function. Imbalances in levels of iron and calcium have been implicated in various neurological disorders [211]. Iron plays an essential role in physiological functions during the ageing process. It is involved in DNA synthesis and repair, oxygen-transport, mitochondrial respiration, myelin synthesis, neurotransmitter synthesis and metabolism [212]. Abnormalities in homeostasis can induce oxidative damage through generation of reactive oxygen species and result in cellular death [212][213][214].
As a predominant neurodegenerative disorder of the ageing population; these changes in iron/calcium level could be implicated in DCM. As it stands; there is no current research in this area and studies looking to quantify these levels could assist in developing new diagnostic options and aid in understanding of the pathological processes of DCM at a molecular level.

Application in DCM
A review of the literature revealed a plethora of studies that were successful in utilising R2*-MRI to quantify iron levels in brain for conditions such as AD [223,[225][226][227], PD [228][229][230][231] and MS [232][233][234][235] in an effort to gauge correlation with the disease. A 2018 study utilised this R2*-MRI to quantify iron accumulation following acute traumatic spinalcord injury [236] and found an increase in brain and brainstem iron accumulation following progressive neurodegeneration of patients. This study however did not explore iron accumulation in the spinal-cord. A 2013 study did however look into iron accumulation in the spinal-cord of mice following traumatic spinal-cord injury and in chronic stages post-injury, using MRI and histological techniques [237]. They were able to detect these iron deposits at the lesion site with live MRI and confirmation with Prussian-blue stains. There has not yet been a study that has looked into spinal-cord iron accumulation in non-traumatic DCM. As the most common cause of spinal-cord dysfunction, a study conducted in this area would prove to be beneficial in developing a new potential MRI biomarker for use in diagnosis.
In Table 6 and Figure 5 we present preliminary unpublished data of R2* ROI scores of the spinal cord of a healthy recruit. These data were obtained through our R2* MRI standardisation protocol and demonstrates the feasibility of this technique to be utilised in the spinal cord of patients. Further work must now be done with regard to DCM patients. Table 6. Preliminary data obtained from our initial incomplete study, showcasing R2* region of interest (ROI) scores of a healthy recruit. Data obtained through a standardisation protocol of R2* using 3T MAGNETOM PRISMA MRI scanner. The values obtained from our R2* is equivalent to other soft tissues in the absence of pathological processes leading to iron deposition (based on R2* studies done in the brain and liver) which tends around 30. As such, we would expect higher R2* values in recruits with DCM.

Underlying Principle
Compounds that have paramagnetic, ferromagnetic and diamagnetic properties all interact with the local magnetic field created by MRI. These compounds distort the local magnetic field and alter the phase of the tissue, which ultimately results in a change in signal. SWI is an MRI sequence that is particularly sensitive to such compounds and is therefore commonly used to detect blood products/haemorrhage and calcium. SWI utilises both the effect on phase and the magnitude, unlike conventional MRI sequences. After acquisition, post-processing involves the application of a high-pass filter that removes background inhomogeneity of the magnetic field and the employment of a phase mask which is used to accentuate the change in signal. This culminates in a susceptibilityweighted image which simultaneously incorporates magnitude and phase information for clinical use [238][239][240][241].

Role of Calcium in Neurodegenerative Disorders
Calcium also plays an essential role in the ageing process. Physiological Ca 2+ fluxes across plasma membranes and between intracellular compartments play vital roles in neuronal function such as in synaptic-transmission and plasticity, regulating neurite-growth and synaptogenesis, and cell survival. In neurodegenerative disorders these systems are compromised resulting in neuronal degeneration and dysfunction [242][243][244].
Interestingly, studies revealed the role of cellular iron in the stimulation of calcium signalling [245][246][247]. Whilst physiologically, this relationship assists in enhancing calciumdependent signalling-pathways, an excessive iron accumulation promotes oxidative stress and a pathological upsurge in calcium-signals, of which results in mitochondrial damage. Moreover, this mitochondrial dysfunction renders a loss of iron homeostasis. If uncontrolled, this manifests a deleterious self-perpetuating cycle which eventuates in neuronal death.

Application in DCM
The important self-inducive relationship between iron and calcium renders calcium an important area of research in the scope of neurodegenerative disorders such as DCM. Multiple studies have indicated calcium overload (calcification) at the impact site of acute traumatic spinal-cord injuries [248][249][250]. It has been yet to be determined whether calcium accumulation occurs during the course of DCM. Modern SWI is a MRI sequence that is particularly sensitive to compounds that distort local magnetic-fields and has been successfully utilised to measure calcium accumulation in the brain in vivo [238,[251][252][253][254][255]. Of these studies includes a 2010 prospective study [255] with high-sample sizes and a varied population (age and gender). Extending protocols to image the spinal-cord could also prove beneficial in biomarker development.

Conclusions and Future Directions
For qMRI to attain clinical significance in DCM it must satisfy three overarching pillars of improvement. Firstly, the necessary advances must be made to minimise issues associated with artifacts and distortions whilst simultaneously improving on image quality, signal-noise ratio and spatial resolution. Such improvements will render qMRI both accurate and able to obtain repeatable results. Secondly, such advances must be utilised to further the literature on DCM, taking in account the limitations of current studies and inadequate areas of research as pointed out in this review. Finally, being both a novel and complex area of study, education is a priority, whereby researchers and clinicians must be updated on these novel quantitative techniques to enable more widespread and effective usage. This in turn will garner further research into this area. Notably, further longitudinal studies with higher sample sizes and longer time horizons are necessary to determine the full prognostic capacity of qMRI in DCM.

Acknowledgments:
We wish to thank the staff at Research Imaging NSW, a core facility located at UNSW for their contribution in preliminary data acquisition (Table 5 and Figure 4).

Conflicts of Interest:
No conflict of interest to declare. None of the funding source have a direct or indirect pecuniary interest in the current work.   Table A3. Neck Disability Index (NDI) [258].

Neck Disability Index
Please answer every section and mark in each section only the one box that applies to you.    Table A4. EQ-5D [259].

EQ-5D
By placing a checkmark in one box in each group below, please indicate which statements best describe your own health state today.  Table A5. Nurick Grading System [260].

Nurick Grading System
Grade. Definition 0 Signs or symptoms of root involvement, but without evidence of spinal cord disease.
I Signs of spinal cord disease, but no walking difficulty.

II
Slight difficulty in walking that does not prevent full-time employment.

III
Walking difficulty that prevents full-time employment or the ability to do all housework but is not so severe as to require help from another person to ambulate.

IV
Able to walk only with help from another person or with the aid of a frame.