Why One MRI Image and a 10-Minute Survey Can Tell You So Much About Your Cervical Spine

Published by the eSpine Team | espine.com

When patients first encounter eSpine, a reasonable question often arises: “How can a single MRI image and a short questionnaire possibly be enough to say something meaningful about a condition as serious as degenerative cervical myelopathy (DCM)?”

It’s a fair challenge. And it deserves a rigorous answer grounded in the peer-reviewed literature — not marketing language.

The short answer is that the combination of a mid-sagittal T2-weighted cervical MRI and the modified Japanese Orthopaedic Association (mJOA) score is not a simplification. It is, in fact, the same information that spine surgeons have used for decades to make surgical decisions. eSpine didn’t invent this pairing — it made it accessible.

Here is the scientific case for each component.

The Sagittal T2 Image: A Compression-and-Cord-Injury Map in a Single Frame

The Gold Standard View

MRI is the definitive imaging modality for degenerative cervical myelopathy. Among its many sequences, the mid-sagittal T2-weighted image is the workhorse of clinical evaluation. It simultaneously captures:

• The degree of spinal canal stenosis at every cervical level
• The presence and location of cord compression
• Intramedullary T2 signal change — the imaging hallmark of cord injury

This is not a shortcut. Sagittal T2 is the view every spine surgeon looks at first, and in most clinical contexts, it is the view that drives the conversation about surgery.

What the Literature Says About Sagittal T2 Alone

A landmark 2025 deep learning study published in Skeletal Radiology trained ensemble neural networks (ResNet50, VGG16, MobileNetV3, and EfficientNetV2) to predict cervical canal stenosis exclusively from mid-sagittal T2-weighted MRI across a cohort of 7,645 patients. The ensemble model achieved an AUC of 0.95, with accuracy of 87.5%, demonstrating that this single view encodes enough structural information to match expert-level radiological grading at scale. Gradient-weighted class activation mapping confirmed the models were attending to clinically meaningful anatomic regions — cord indentation and subarachnoid space obliteration — not imaging artifacts.

A separate study using fully automated T2 signal intensity analysis of the spinal cord on sagittal sequences found that signal variability alone differentiated DCM patients from healthy volunteers with AUC values between 0.865 and 0.920 — performance sufficient to “diagnose radiological DCM more objectively to optimize treatment recommendation.” This approach required no axial imaging.

Does Adding Axial T2 Change Decisions Meaningfully?

In the context of a screening and decision-support tool — rather than surgical planning — the marginal gain from axial imaging is limited. Axial T2 views are most useful for distinguishing specific pathologies (e.g., OPLL from other compressive lesions, or assessing foraminal stenosis causing radiculopathy). For the core question eSpine addresses — does this person have significant cord compression with cord signal change, and how severe is their functional impairment? — the sagittal T2 provides the primary answer.

When a patient undergoes a cervical MRI, both sagittal and axial sequences are routinely acquired — that doesn’t change. eSpine simply requires only the sagittal T2 view from among those sequences. All the clinically relevant information for this tool’s purpose is already contained in that single image.

T2 Signal Change: The Most Clinically Significant Single Finding

The presence of intramedullary T2 hyperintensity — the bright signal within the cord itself — is one of the most consequential findings a spine surgeon evaluates. It reflects myelomalacia, edema, or gliosis from sustained cord compression, and its detection is a core reason the sagittal T2 view is indispensable. This is clinical context for why the sagittal sequence carries so much information; eSpine’s anomaly detection layer flags findings that may warrant further investigation by a clinician, but the tool makes no claims about prognosis or surgical timing based on signal change.

eSpine’s AI pipeline is specifically trained to detect this finding, quantify the degree of cord compression, and assess curvature — all from the sagittal frame. The model is conservatively calibrated not to flag incidental spondylosis without frank cord deformation, in alignment with the literature’s recognition that degenerative change is nearly universal in adults over 50 and only pathological when it causes cord injury.

The mJOA Score: The Field’s Best Clinical Severity Measure

What the mJOA Captures

The modified Japanese Orthopaedic Association (mJOA) score assesses five functional domains relevant to DCM: upper extremity motor function, lower extremity motor function, upper extremity sensation, lower extremity sensation, and bladder function. The total score ranges from 0 (maximum impairment) to 18 (normal function).

Severity thresholds established across 757 patients in the landmark AO Spine North America and International prospective studies define:

• Mild DCM: mJOA ≥ 14
• Moderate DCM: mJOA 11–13
• Severe DCM: mJOA ≤ 10

These thresholds are validated against the Nurick score and multiple quality-of-life measures, with strong convergent validity. They are the same thresholds used in AO Spine’s international surgical timing guidelines, and they inform the Fehlings criteria that eSpine uses to contextualize its output.

Preoperative mJOA Is the Most Important Predictor of Surgical Outcome

Multiple systematic reviews and prospective multicenter trials confirm that preoperative mJOA score is the single most important predictor of surgical outcome in DCM. A 2023 analysis of 2,156 patients in the Quality Outcomes Database confirmed that patients improved significantly from baseline regardless of baseline severity, but trajectory modeling showed that preoperative mJOA category was a dominant predictor of recovery trajectory.

A 2024 systematic review noted that across multiple RCTs, the severity of preoperative dysfunction measured by mJOA score was highly predictive of both the magnitude of postoperative improvement and the floor below which recovery could not reach. Clinical parameters — especially mJOA — were found more important than most radiological parameters alone in predicting outcomes.

This matters for eSpine’s design: the mJOA is not a triage curiosity. It is the metric the entire field of spine surgery uses to counsel patients, select surgical candidates, and measure success.

The mJOA as a Screening Tool

A 2024 study published in The Spine Journal specifically evaluated the mJOA as a clinical screening tool for DCM in a tertiary care population. A cut-score of ≤16 was 62% sensitive and 90% specific for a subsequent diagnosis of DCM before any imaging review. The authors concluded that the mJOA could appropriately expedite MRI referral in patients with possible myelopathy.

eSpine inverts this logic productively: patients who already have an MRI can use the mJOA to contextualize their imaging within a functional framework — and receive an integrated interpretation that mirrors what their surgeon will consider.

Why This Combination Is Greater Than the Sum of Its Parts

Imaging and clinical assessment answer different questions:

Neither alone is sufficient. MRI can show severe cord compression in a patient who is functionally intact — often prompting watchful waiting. The mJOA can reflect significant dysfunction in a patient whose imaging appears less impressive — raising concern for early or overlooked pathology. The pairing produces a richer picture than either input alone.

This is precisely why the Fehlings AO Spine guidelines — the most widely adopted international framework for DCM surgical timing — incorporate both imaging severity (cord signal change, compression) and clinical severity (mJOA score) in their recommendations. eSpine operationalizes this framework for the patient encounter.

What eSpine Does Not Do — And Why That’s Also By Design

eSpine is not a diagnostic device. It does not render a diagnosis of DCM, prescribe surgery, or replace the clinical encounter. A single sagittal image cannot capture dynamic stenosis, foraminal disease, or the nuanced neurological examination a spine surgeon performs in person.

What eSpine does is translate a widely available imaging study and a validated clinical instrument into an accessible, structured summary — the same information your surgeon will discuss with you, organized in a way that empowers the pre-visit conversation.

The goal is not to replace expertise. It is to reduce the information asymmetry between patient and provider, so that when the surgical consultation happens, it starts further along.

Conclusion

The choice to build eSpine around one sagittal T2 image and the mJOA score is not a limitation of ambition. It is a reflection of where the evidence is. These two inputs are the same two inputs at the heart of international surgical guidelines, multicenter prospective trials, and the most advanced AI diagnostic models published in 2024 and 2025.

Simplicity, in this case, is rigor.

References available upon request. eSpine is a decision support tool and is not intended to diagnose, treat, or replace the clinical judgment of a licensed healthcare provider.