The Role of CT in Cervical Spine Imaging for Suspected Cervical Myelopathy

Abstract

Introduction:

Cervical spinal canal stenosis is a degenerative condition that may lead to myelopathy. While MRI is the preferred modality for detecting changes in spinal cord signal, CT is superior in assessing bony structures. The diagnostic accuracy of these modalities in measuring canal narrowing remains a topic of debate.

Methods:

We conducted a retrospective study of 20 patients with cervical myelopathy who underwent both CT and MRI before surgical decompression. Thecal sac and spinal canal areas were measured at C2–T1 levels. Paired t-tests and Pearson correlations were used to compare CT and MRI measurements. Myelomalacia was assessed on MRI and correlated with cross-sectional areas.

Results:

CT consistently reported larger thecal sac and spinal canal areas compared with MRI, with statistically significant differences at all levels (e.g., C3–C4 spinal canal area: CT 1.76 ± 0.37 cm² vs. MRI 1.40 ± 0.60 cm², P = 0.01). Correlations between CT and MRI were moderate to strong for spinal canal area (r = 0.47 to 0.74), but weaker for thecal sac dimensions. At C6–C7, CT-derived canal area ≤1.32 cm² was significantly associated with myelomalacia (P = 0.007).

Conclusion:

This study suggests that CT offers superior anatomical assessment of canal narrowing and better predicts the occurrence of myelomalacia than MRI. Multimodal imaging is crucial in evaluating cervical myelopathy, and the use of CT should be considered, especially when MRI findings are inconclusive.

Cervical spondylotic spinal canal stenosis is a degenerative condition characterized by a progressive narrowing of the cervical spinal canal, resulting from a combination of osseous and soft-tissue changes. These changes typically include osteophyte formation, facet joint hypertrophy, thickening or calcification of the ligamentum flavum, and disk protrusions at one or more intervertebral levels.^1,2 The resultant chronic compression of the cervical spinal cord can lead to cervical myelopathy, including burning sensations, hand clumsiness, muscle weakness, and gait instability.^3,4 The natural history of cervical myelopathy is often one of gradual, unidirectional deterioration, and surgical decompression remains the definitive treatment to halt disease progression and improve neurological function.⁵

Magnetic resonance imaging (MRI) is the modality of choice for evaluating suspected cervical myelopathy because of its superior ability to visualize soft tissues, neural structures, and intramedullary signal changes such as myelomalacia, which is considered a poor prognostic indicator.^6,7 However, CT remains indispensable for assessing bony anatomy and is particularly valuable in detecting ossification of the posterior longitudinal ligament (OPLL), a condition with notable implications for surgical planning and risk stratification.^8,9 Nevertheless, CT has known limitations, particularly its inability to directly visualize the spinal cord and its susceptibility to beam-hardening artifacts in the lower cervical spine due to interference from the shoulder.¹⁰

Previous studies have evaluated the agreement between CT and MRI in measuring spinal canal dimensions, particularly in the lumbar spine. These studies consistently demonstrated that MRI tends to underestimate the degree of canal narrowing compared with CT, potentially leading to underappreciation of mechanical compression.^11,12 Similar comparative studies in the cervical spine are sparse and outdated. Early investigations produced conflicting results: while some reported that MRI was less reliable for differentiating between osseous and soft-tissue structures in the cervical region,¹³ others suggested that MRI might be more sensitive in detecting canal narrowing, albeit with the potential for overdiagnosis.¹³ Importantly, no recent studies have revisited this question in the context of modern high-resolution CT and advanced MRI technology.

Given the clinical relevance of accurate canal measurements for both diagnostic and surgical decision making, additional investigation into the agreement and discrepancies between CT and MRI in the cervical spine is warranted. The objective of this study was to compare measurements of the cervical spinal canal area, thecal sac area, and spinal cord cross-sectional area and signs of myelomalacia at disk levels along the cervical spine using CT and MRI in patients undergoing surgical treatment for clinically and radiographically diagnosed cervical myelopathy.

Methods

This retrospective observational study included 20 patients who underwent both cervical spine CT and MRI at a single tertiary medical center between January 2022 and December 2023. All patients had imaging performed within a three-month interval and were evaluated for suspected cervical spinal stenosis. Patients were included if both 1.5-mm-thick CT slices and 1.5-Tesla T2-weighted MRI scans covered cervical vertebral levels from C2 to T1 and were excluded if they had a history of cervical spinal surgery or other spinal pathologies, such as neoplasms, infections, or inflammatory diseases.

At each disk level from C2–3 through C7–T1, measurements were taken for the spinal canal area, thecal sac area, spinal cord area, and the AP diameters of both the spinal canal and thecal sac. Measurements were conducted using RadiAnt DICOM Viewer software by manually marking the circumferences of the spinal canal, thecal sac, and spinal cord. Area and perimeter were automatically calculated by software. Myelomalacia was defined radiologically as T2-weighted MRI hyperintensity and was assessed at each vertebral level. Radiographic measurements were conducted by an orthopaedic surgery resident and assessed by a North American fellowship-trained spine surgeon.

Ethical approval for this study was obtained from the Institutional Review Board (IRB), and informed consent was waived because of the retrospective design. The study was conducted according to the STROBE Statement (STrengthening the Reporting of OBservational studies in Epidemiology). Statistical analysis included descriptive statistics and two-tailed paired t-tests to compare CT and MRI-derived measurements at each cervical level. Pearson correlation coefficients were used to evaluate the strength of association between CT and MRI measurements. To explore the relationship between myelomalacia and anatomical parameters, spinal canal and thecal sac areas were divided into quintiles using visual binning, and associations with myelomalacia were tested using Pearson Chi-square and linear-by-linear association tests. A P value of less than 0.05 was considered statistically significant. All analyses were conducted using SPSS software (version 27, IBM Corp.).

Results

This retrospective observational study included 120 axial CT and 120 axial MRI scans that were analyzed. Figure 1 demonstrates the comparison of spinal canal area measurements obtained from CT and MRI across cervical levels C2–T1. The measured spinal canal area was consistently larger on CT scans compared with MRI scans at most levels, with the difference reaching statistical significance at the C3–C4 level (P = 0.01), where MRI-derived measurements were notably smaller. At the C2–C3 level, a similar trend was observed, although the difference did not reach statistical significance (P = 0.11). At lower cervical levels (C4–T1), spinal canal area measurements were comparable between CT and MRI, with no statistically significant differences observed.

Figure 1.

Box plot demonstrating spinal canal area by cervical level.

Correlational analyses comparing CT and MRI measurements demonstrated moderate-to-strong agreement for spinal canal area at most levels (Pearson r = 0.47 to 0.74, all P < 0.05), except for the C7–T1 level, where the correlation was weaker (r = 0.31, P = 0.096). For the thecal sac area, correlations between CT and MRI were generally weaker, particularly at lower cervical levels, suggesting reduced measurement agreement in this region. Notably, paired t-tests revealed significant systematic differences between CT and MRI measurements of thecal sac area at all cervical levels (all P < 0.05), with MRI consistently underestimating thecal sac dimensions compared with CT.

The comparison of thecal sac and spinal cord area measurements across cervical levels is illustrated in Figure 2. Regarding thecal sac area measurements, MRI consistently yielded lower values than CT, and these differences were statistically significant at every vertebral level from C2–C3 to C7–T1. For AP diameters, a significant difference in spinal canal parameters was detected only at the C3–4 level (P = 0.002), with CT measurements again being larger than those of MRI. AP diameters of the thecal sac differed significantly between modalities across all levels.

Figure 2.

Box plot demonstrating thecal sac and spinal cord areas by cervical level.

Thecal sac area was consistently larger on CT compared with MRI at all levels (C2–T1), with the difference reaching statistical significance at every level (P < 0.05). This discrepancy was most pronounced at the upper cervical levels (C2–C4), where CT-derived thecal sac area measurements were substantially higher than those obtained from MRI. Similarly, thecal sac circumference showed a significant reduction on MRI compared with CT across all levels, with P values ranging from <0.001 to 0.009, reflecting a consistent underestimation of thecal sac dimensions on MRI.

Spinal cord area and circumference were available for MRI only. These values demonstrated relatively stable distributions along the cervical spine, with a slight decrease in both area and circumference at the C7–T1 level.

Chi-square analyses assessed the predictive value of spinal canal area, measured by CT, for myelomalacia. At most cervical levels, no significant association was observed. However, at the C6–C7 level, a significant relationship emerged, whereby smaller canal areas were associated with a higher prevalence of myelomalacia (linear-by-linear association P = 0.013; likelihood ratio P = 0.007). This pattern was not observed at other levels, where the canal area did not markedly predict the presence of myelomalacia.

Additional statistical analysis explored the relationship between spinal canal dimensions and the presence of myelomalacia. A comparison of the spinal cord area on MRI between patients with and without myelomalacia revealed no significant differences at any cervical level (all P values > 0.1). Similarly, analysis of the spinal cord perimeter (cross-sectional shape ratio) showed no consistent association with myelomalacia, except at the C4–C5 level, where patients with myelomalacia demonstrated significantly higher perimeter values compared with those without (P = 0.017 1.76 ± 0.37 cm), suggesting localized cord deformation at this level.

Finally, when examining the canal area on MRI as a predictor of myelomalacia, results were inconsistent and generally weaker than those obtained with CT. Only at the C3–C4 and C5–C6 levels was a borderline significant ordinal association observed, suggesting that a narrower canal area on MRI may increase the odds of myelomalacia; however, these findings did not reach robust statistical significance. Overall, CT-derived canal dimensions, particularly at C6–C7, demonstrated superior predictive value for myelomalacia compared with MRI.

Stratification of CT spinal canal areas into quintiles (≤1.60 cm², 1.61 to 2.00 cm², 2.01 to 2.36 cm², 2.37 to 2.55 cm², >2.55 cm²) revealed a significant linear association between smaller CT-measured areas and the presence of myelomalacia at the C6–C7 level (P = 0.013), whereas no such relationship was found at other vertebral levels. A similar analysis using MRI-derived canal areas yielded weaker and less consistent associations with myelomalacia. Although linear-by-linear trends were observed at C3–C4 and C5–C6 (P = 0.010 and P = 0.007, respectively), the overall Pearson Chi-square values were not statistically significant, indicating that MRI had a limited predictive value in this context.

Discussion

This study assessed 20 patients with diagnosed cervical myelopathy, yielding a total of 120 axial CT and 120 axial MRI scans for analysis and providing a substantial data set. This study demonstrates that cervical MRI systematically underestimates spinal canal and thecal sac dimensions compared with CT, with the most pronounced discrepancies observed at upper cervical levels. At the C3–C4 level, for example, CT-derived spinal canal area measurements were significantly larger than those obtained from MRI (CT: 1.76 ± 0.37 cm² vs. MRI: 1.40 ± 0.60 cm², P = 0.01), underscoring MRI's known limitations in accurately depicting osseous boundaries. Similar, though less statistically robust, trends were observed at other levels, with the upper cervical spine showing greater discrepancies, while CT and MRI measurements became more comparable at lower levels.

Despite these significant differences, moderate-to-strong correlations were found between CT and MRI measurements of spinal canal area across most cervical levels (r = 0.47 to 0.74, P < 0.05), except for the C7–T1 level, where the correlation was weaker (r = 0.31, P = 0.096). However, thecal sac area demonstrated weaker correlations between modalities, particularly at lower cervical levels, and paired t-tests revealed consistent and statistically significant underestimation of thecal sac area by MRI at all levels (P values ranging from 0.001 to <0.001).

A key finding of this study is the superior performance of CT in predicting myelomalacia, particularly at the C6–C7 level. In this study, binning analysis revealed that smaller CT-derived spinal canal areas were significantly associated with the presence of myelomalacia (likelihood ratio P = 0.007; linear-by-linear association P = 0.013). Importantly, this association was not replicated with MRI-derived canal area measurements, which failed to show significant predictive value at this level (P = 0.143). Similar trends were observed at other levels, but statistical significance was only achieved at C6–C7 using CT.

Analysis of spinal cord area and shape perimeters further illustrates the limitations of relying solely on MRI measurements. Across all cervical levels, the spinal cord area on MRI did not differ significantly between patients with and without myelomalacia (all P values > 0.1). A thecal sac area on CT of ≤ 1.32 cm² at the C6–C7 level is highly suggestive of the presence of myelomalacia in the cohort, suggesting that this value may serve as a CT predictive cutoff for myelomalacia.

These findings collectively suggest that while MRI remains the benchmark for detecting intramedullary signal abnormalities such as myelomalacia, it may inadequately characterize the degree of mechanical canal narrowing that contributes to neural tissue injury. This underperformance of MRI may stem from its inherent limitations, including lower spatial resolution for bone and greater susceptibility to artifacts, which are particularly problematic in assessing narrow anatomical spaces in the cervical spine.

Our results align with those of previous studies in the lumbar spine, demonstrating that CT provides a more accurate structural assessment of canal narrowing than MRI. Extending these observations to the cervical spine—a region where even small dimensional reductions can have profound neurological consequences—highlights the critical value of a multimodal imaging approach. In particular, CT should be considered when MRI findings are borderline or discordant with the clinical presentation, to ensure that notable mechanical compression is not overlooked.

The study has limitations, including its retrospective design, small patient sample size, and reliance on radiological rather than histopathological confirmation of myelomalacia. In addition, interobserver variability in measurements was not formally assessed. Future research should evaluate how these imaging discrepancies influence surgical decision making and clinical outcomes in larger, multicenter prospective studies. Clinical assessment of cervical myelopathy, such as the JOA score or the Nurick scale, was not the focus of this study, which may be a limitation and warrants additional research. The study focused on the correlation between the most commonly used imaging modalities, CT and MRI.

In conclusion, CT provides a more reliable anatomical assessment of cervical spinal canal narrowing than MRI, particularly at the C6–C7 level, where it demonstrates superior predictive value for myelomalacia. These results support the routine integration of CT into the diagnostic workup for patients with suspected cervical myelopathy, particularly in complex or ambiguous cases.

Conclusion

MRI underestimates cervical spinal canal and thecal sac dimensions compared with CT, particularly in the upper cervical spine. CT measurements at the C6-C7 level demonstrated the strongest predictive association with spinal cord myelomalacia. Given these findings, CT may serve as a valuable adjunct to MRI in the preoperative evaluation of cervical stenosis, particularly in cases where MRI results do not correlate with neurological findings.