D-wave and multimodal monitoring: what the signals can tell us

Preserved conduction is not a guarantee of preserved function.

During intramedullary spinal cord surgery, the practical question is not which waveform is “best.” It is what a change tells the team about a particular pathway, at a particular moment, and how that relates to the patient’s later function. D-wave recordings and muscle motor evoked potentials (mMEPs) answer overlapping but different questions. Neither should be interpreted in isolation. [1] [2]

A preserved D-wave can offer useful information about corticospinal conduction even when muscle responses deteriorate. It does not rule out an immediate motor deficit or guarantee full recovery. The 2025 diagnostic meta-analysis also found that muscle MEPs identified persistent deficits missed by D-wave monitoring. That finding argues against treating an unchanged D-wave as permission to disregard a significant muscle MEP change. [2]

The 2024 AO Spine–PRAXIS guideline recommends IONM for high-risk spine surgery and suggests proactive risk identification, multidisciplinary discussion and an intraoperative protocol. Intramedullary spinal cord tumors are included among its high-risk conditions. This is support for a coordinated monitoring pathway, not endorsement of a single signal, alarm threshold or AI product. [3]

Three signals, different blind spots

This comparison describes their physiological scope; it is not a ranking of interchangeable tests. [1]

D-wave monitoring is relatively resistant to anesthetic effects, but that does not make it artifact-proof. Catheter displacement and waveform misidentification can mislead interpretation. An absent baseline is a monitorability problem—not a numerical value that can simply be entered as a normal result. [1]

Many reports discuss a 50% D-wave amplitude reduction as a warning criterion. We do not turn that literature convention into a universal stop-or-continue rule: baseline quality, persistence, associated signal changes, anatomy and the team’s protocol matter. The new observational studies below do not establish a universally safe threshold. [1] [4]

Evidence ledger: prediction, time and comparison

These are selected sources, not a systematic review. Outcome timing differs across studies; percentages should not be pooled informally or treated as a patient’s individual prognosis.

The studies support taking D-wave information seriously while preserving the contribution of muscle MEPs and the wider clinical picture. They do not establish that one modality consistently dominates every outcome at every follow-up point. [2] [5]

Three questions before using a percentage

Who was actually assessed?

A cohort size is not necessarily the denominator for a diagnostic estimate. Jiang’s abstract reports 125 matched patients and 88 with both monitoring modalities, but it does not state the evaluable denominator or confidence intervals for the six-month estimates. We have not reconstructed those missing data from rounded percentages. [5]

Monitorability also matters. In Tropeano and colleagues’ single-center series of 64 patients, D-waves were recordable in 57; the study reported a 53-patient comparison with all three modalities. Its prospectively maintained registry was analyzed retrospectively, and it had no unmonitored control group. Results from evaluable recordings cannot automatically describe all patients considered for monitoring. [7]

Which outcome, at what time?

A negative predictive value of 96.5% describes the proportion without the specified outcome among those with a negative test in the analyzed setting. It is not a 96.5% probability that an individual patient will recover, retain every neurological function or avoid all surgical complications. The six-month motor endpoint must remain attached to the number. Likewise, a median follow-up of 34 months does not mean every patient was assessed for 34 months. [5] [7]

Did the warning change what the team did?

Monitoring is part of care, not a passive test. Tropeano’s report describes pauses or changes in resection after signal changes. When the team responds to an alarm and no lasting deficit follows, the outcome alone cannot distinguish an unnecessary alarm from a successful protective response. This is one reason diagnostic accuracy and prevention of injury need separate evaluation. [7]

Where AI evidence starts—and where it stops

Automated signal classification, prediction of postoperative deficit and improved patient outcomes are three separate claims. The FDA summary for SafeOp 3 describes an SSEP baseline-classification algorithm and controlled future modifications. That is a different signal and task from D-wave interpretation in intramedullary tumor surgery. Its technical evidence must not be borrowed as proof that AI-guided resection improves neurological outcomes. [6] Read the existing Index record →

Our editorial standard for a future AI study in this setting is to ask:

• What is the unit of validation? Patient-level separation and independent centers matter; thousands of correlated waveform segments are not thousands of patients.
• What happens when the signal is unusable? Report baseline failures, artifacts and excluded cases, not performance only on clean recordings.
• What is the clinical endpoint? Specify immediate weakness, persistent motor deficit or later function, with follow-up completeness and uncertainty.
• What changes in real practice? Evaluate false alarms, missed injury, team response and prospective clinical outcomes before claiming benefit.

This checklist is our editorial assessment framework, not a claim that a particular AI system meets it. The sources selected here do not establish outcome benefit from AI-guided D-wave interpretation.

Connect the signal to the clinical question

The intramedullary teratoma Atlas examines the tumor–cord interface and the limits of resection. It remains a draft awaiting anatomy review. Its mechanism illustrations and this monitoring guide are complementary educational materials; ependymoma-dominant monitoring cohorts do not resolve teratoma-specific decisions.

For shorter research updates, browse Spinal Cord coverage in the Brief. For technical and regulatory claims, use the Index. Neither inclusion nor clearance establishes superior patient outcomes.