Mx, the TCD-based autoregulation index

1. Bedside vignettes: why this matters in the PICU#

Vignette A. Severe TBI, the bolt is delayed#

A 7-year-old severe TBI patient is in the resus bay 90 minutes after the injury. GCS 6T, anisocoria right > left. The CT shows diffuse axonal injury without a surgical lesion; the operating room is preparing for a parenchymal monitor placement in 30 minutes. The MAP is 75, the team has a robotic-frame TCD over the left MCA, and the bedside platform computes a rolling Mx every 5 minutes. Mx = +0.35. Autoregulation is impaired even without an ICP monitor. The team understands that passive CBF tracks MAP linearly here; they target MAP gently (no aggressive fluid or vasopressor escalation) until the bolt is in. 1 5

Vignette B. Adolescent SAH, MAPopt by Mx without ICP#

A 16-year-old SAH on day 4 after coiling. An ICP monitor was placed initially and removed at 48 h with stable ICP. The patient is awake but drowsy, MAP 95. A bedside robotic TCD on the right MCA collects continuous MFV; the bedside platform plots Mx against MAP across the last 4 hours; the U-curve fit gives MAPopt = 90. The team adjusts the noradrenaline to bring MAP from 95 to 88 to 92, sitting within ±5 of MAPopt. This Mx-derived MAPopt replaces the now-removed invasive CPPopt and lets the team continue individualised autoregulation-guided BP management non-invasively. 3 6 7

Vignette C. The patient with atrial fibrillation, Mx misbehaves#

A 14-year-old with congenital heart disease and chronic atrial fibrillation after a cardiac arrest, day 3 post-rewarming. Bedside TCD is in place. The MFV envelope is highly variable beat-to-beat (irregular R-R intervals); the slow-wave power is dominated by the cardiac chaotic rhythm rather than autoregulatory slow waves; the computed Mx fluctuates wildly between −0.4 and +0.6 within a single hour. Mx is not interpretable in this patient. The team falls back to NIRS-derived COx for non-invasive autoregulation and to clinical exam. This is a teaching pitfall: Mx requires a stable, low-frequency MFV signal to extract autoregulatory slow waves. 1 4

2. What Mx is, and what it is not#

Mx is a moving-window Pearson correlation between the time-averaged mean flow velocity (MFV) from a continuous TCD trace and the cerebral perfusion pressure (CPP), or mean arterial pressure (MAP) if CPP is not computable. The correlation is computed over the slow-wave band, typically 0.003 to 0.05 Hz, using 10-second averages over a 5-minute rolling window.

Three things follow.

Mx is the non-invasive cousin of PRx. PRx uses ICP and MAP; Mx uses MFV and CPP (or MAP); COx uses NIRS rSO2 and MAP. All three exploit the same physiology: in intact autoregulation, slow MAP waves do not produce slow CBF/MFV waves (the cerebrovascular bed adjusts CVR); in impaired autoregulation, slow MAP waves produce parallel slow MFV waves (passive flow). The Pearson correlation captures the parallelism. 2 8 9

Mx interpretation thresholds:

The Mx-vs-CPP U-curve gives Mx-CPPopt. Like PRx-CPPopt, plot Mx against CPP across 4 hours of data, fit a parabola, take the vertex. The U-curve allows estimation of LLA (where Mx crosses +0.3 on the left) and ULA (where Mx crosses +0.3 on the right). 3 7

3. The architecture: Mx in the autoregulation triad#

The three indices have different strengths and weaknesses:

The three indices typically agree (Pearson > 0.6) but discord meaningfully in subsets of patients. The discordance is itself informative: PRx-COx discord is a research topic in microvascular shunting and sepsis. 4 11

4. The signal: how Mx is computed#

The bedside platform (ICM+, Sickbay, Brain4Care, or custom) requires:

1. Continuous TCD trace with envelope detection running, sampling MFV every 1 to 10 seconds.
2. Continuous arterial line MAP sampled at the same cadence.
3. Continuous ICP if computing CPP (else use MAP).
4. Slow-wave extraction: a band-pass filter (0.003 to 0.05 Hz) on both signals.
5. Rolling Pearson correlation: typically 10 s averages over a 5 min window, updated every 1 minute.
6. Mx-vs-CPP binning: collect Mx and CPP pairs across 4 to 6 hours, bin CPP into 5-mmHg windows, compute mean Mx per bin, fit a parabola.
7. Display: continuous Mx trend strip, U-curve fit with vertex, ±5 mmHg target band.

Common implementation details.

• Sampling rate: TCD at 100 to 200 Hz envelope; MAP at 100 Hz; downsample to common rate.
• Slow-wave band selection: most adult studies use 0.003 to 0.05 Hz; some pediatric studies use 0.01 to 0.05 Hz to avoid respiratory artefact from very slow neonatal respiration.
• Artefact handling: discard windows with > 20% missing data, electrode disconnection, ventilator manoeuvre, suction, posture change.
• Trend smoothing: 5 to 15 minute rolling mean of Mx is often what is displayed; the underlying minute-level Mx is more variable.

5. The numbers: what to record at the bedside#

Display Mx alongside PRx and COx where all three are available; the multi-index view is the modern standard at research-grade centres.

6. What is normal? Mx interpretation reference#

Pediatric thresholds: limited data suggest the adult thresholds (>+0.3 impaired) hold; pediatric reference ranges remain a research priority. 10 4

7. What is abnormal? Pattern library#

Decision tree: "what does Mx tell me?"#

flowchart TD
  Mx[Mx reading] --> Q1{Mx > +0.3 sustained?}
  Q1 -->|No| OK[Intact; continue]
  Q1 -->|Yes| Q2{Quality good?}
  Q2 -->|No| Pause[Probe / rhythm issue; pause interpretation]
  Q2 -->|Yes| Q3{PRx or COx available?}
  Q3 -->|Yes, concordant| Conc[Concordant impaired; narrow MAP range]
  Q3 -->|Yes, discordant| Disc[Discordant; investigate]
  Q3 -->|No| Act[Treat as impaired; tighter MAP control]

8. Try it: interactive widgets#

9. Mx-driven management decisions#

9.1 Pre-monitor-placement triage#

A severe TBI patient awaiting parenchymal monitor placement: Mx + TCD + arterial line provides a non-invasive autoregulation read in real time. Mx > +0.3 suggests broken autoregulation: do not push MAP aggressively; the volume goes to tissue (oedema) not to useful perfusion.

9.2 Post-monitor-removal CPPopt continuity#

A SAH patient whose ICP monitor has been removed at 48 to 72 h but who still needs autoregulation-guided BP management: Mx + TCD continues the personalised CPP / MAP target where PRx-CPPopt left off. 3 6

9.3 Centres without invasive monitoring#

Resource-limited centres or palliative-context patients may have no parenchymal monitor. Mx is the bedside autoregulation index of choice. The pediatric MMM consensus places Mx-driven autoregulation monitoring in tier-2 modalities for resource-stratified pediatric centres. 12 13

9.4 Pediatric ECMO#

VA-ECMO patients without invasive ICP. Mx is technically harder on full-flow ECMO (TCD PI falls toward zero with loss of pulsatility, but the slow-wave MFV component can still be extracted). Some centres run Mx on partial-flow ECMO; others use COx instead. 14 15

9.5 The mutual-validation use case#

When ICP is present and PRx is computed, adding Mx serves as a cross-check. PRx-Mx concordance raises confidence; discordance flags a technical issue (probe coupling, ICP artefact) or a real microvascular / shunting phenomenon worth investigating. 4

10. Clinical contexts: Mx across acute brain injuries#

10.1 Severe TBI#

The largest body of Mx evidence. Mx > +0.3 sustained is associated with worse outcome in adult TBI (Czosnyka 1996, Lang 2003, Aries 2012). Pediatric data: Tas 2022 in 30 children with severe TBI showed Mx-CPPopt feasibility; outcome correlation requires larger cohorts. 1 7 3 10

10.2 Aneurysmal SAH#

Mx use in SAH covers two windows: peri-coiling (where Mx informs hemodynamic management) and post-monitor-removal (Mx-derived MAPopt continues autoregulation-guided care). 16 17

10.3 Pediatric AIS#

Limited Mx use in pediatric AIS; the bedside framework is BP target (often NIRS-driven) plus monitoring. Research interest in continuous TCD + Mx during recanalisation procedures. 18 19

10.4 HIE and post-cardiac arrest#

In neonatal HIE, NIRS-derived COx is more commonly used than TCD-derived Mx because NIRS pads are easier to maintain than TCD probes in neonates. In pediatric post-arrest, Mx (where TCD is sustained) plays a role in multimodal prognostication and BP target setting. 20 21 22

10.5 Pediatric ECMO#

Investigational. The non-pulsatile ECMO circulation alters TCD signal in ways that complicate Mx; some centres use it on partial-flow ECMO. 14 23

10.6 Meningitis and encephalitis with raised ICP#

Less common indication. Mx can supplement bedside management in fulminant meningitis with broken autoregulation; the primary management remains source control and ICP / CPP / fluid balance. 24 25

10.7 Brain-death determination#

Not a brain-death tool. As cerebral circulatory arrest approaches, the TCD signal degenerates (oscillating, pendular, systolic spikes), and Mx becomes uninterpretable. The TCD findings themselves are ancillary; Mx is not. 26 27

10.8 DKA cerebral oedema#

Limited published Mx use. In DKA-CO with broken autoregulation (theoretical concern), Mx could refine the BP / fluid management during rehydration; not standard practice. 28 29

10.9 Peri-arrest research contexts#

Mx has been used in cardiac surgery, post-CPR research, and CPR optimisation studies as a bedside autoregulation index. Pediatric and adult research data continue to accumulate. 4

11. Multimodal integration: Mx in the MMM/MNM stack#

12. Setup and technique#

12.1 Equipment#

• Continuous TCD probe (2 MHz pulsed-wave or TCCD), with a robotic / fixed headframe for hours-long monitoring (DWL Doppler-Box, Atys Robotic, Dolphin, Lucid M1).
• Synchronised arterial-line MAP at 100 Hz minimum.
• Synchronised ICP if CPP is to be computed (else use MAP).
• Bedside platform: ICM+, Sickbay, custom Python pipeline, or vendor-integrated Mx computation.
• Adequate ambient quiet: scanner / staff traffic moves the probe.

12.2 The setup workflow#

1. Identify the temporal window and obtain a clean MCA M1 envelope.
2. Position the robotic headframe so the probe is locked into the chosen angle and depth.
3. Verify the spectrum continuously: bright clean envelope, no overgain, no aliasing.
4. Confirm MAP recording (calibrated, square-wave test passed).
5. Confirm time-synchronisation between the TCD and the arterial line (most platforms autosync within ms).
6. Start the rolling Mx computation (typically 5 min window, 1 min update).
7. Display alongside MAP, ICP (if present), PRx, COx.

12.3 The Mx-CPPopt or Mx-MAPopt fit#

1. Collect ≥ 4 hours of data with Mx and CPP (or MAP) sampled continuously.
2. Bin CPP into 5 mmHg windows; compute mean Mx per bin.
3. Fit a parabola (least-squares, often with manual selection of fit region to exclude outliers).
4. Vertex = MAPopt (or CPPopt).
5. Target MAP (or CPP) within ±5 mmHg of the vertex.
6. Re-fit every 1 to 4 hours; the fit can shift over time as the patient evolves.

12.4 Quality control#

• Probe coupling: re-aim every 1 to 2 hours or use a robotic frame; probe drift is the leading source of Mx artefact.
• Slow-wave content: if the patient is in deep anaesthesia with very flat haemodynamics, slow-wave power may be inadequate for reliable Mx; flag and pause interpretation.
• Irregular rhythms: AFib, frequent ectopy, paroxysmal SVT all degrade Mx; use COx or pause.
• Artefact rejection: discard windows around suction, posture change, ventilator manoeuvres.
• Document quality with every Mx report: high / medium / low confidence based on probe coupling, slow-wave power, rhythm regularity.

12.5 Pediatric-specific considerations#

• Smaller temporal windows in neonates may favour TCCD over blind TCD.
• Higher respiratory rates in neonates shift the respiratory artefact band; the 0.01 to 0.05 Hz slow-wave window may need adjustment.
• Restless or unsedated children complicate sustained TCD; sedation status must be documented.
• Pediatric Mx normative data are sparse; threshold (>+0.3) extrapolated from adult; treat as research-grade.

12.6 When Mx is not the right tool#

• Cardiac rhythm chaos (AFib, frequent VT, post-arrest dysrhythmias): use COx or pause.
• Inadequate slow-wave power (deep anaesthesia, flat haemodynamics): pause until physiology recovers.
• Probe coupling unsustainable (toddler agitation, ECMO traffic, no robotic frame): use COx.
• Acute pre-arrest state: TCD signal degenerates; Mx uninterpretable.

13. Pitfalls#

• Probe drift during long recordings: envelope dims over minutes; Mx degrades; use a robotic / fixed frame.
• Irregular cardiac rhythms dominate slow-wave power and corrupt Mx; AFib, frequent ectopy, paroxysmal SVT all confound.
• Low slow-wave power (deep anaesthesia, flat haemodynamics): Mx unreliable.
• Single-snapshot Mx: trend over hours is the diagnostic signal.
• Pediatric normative data sparse: treat thresholds as adult-derived heuristics.
• Mx vs MAP without ICP: do not equate Mx-vs-MAP with Mx-vs-CPP if ICP is changing; ICP variation alters CPP for a constant MAP.
• Confusing Mx with PI: Mx is a slow-wave correlation; PI is a per-beat ratio; they measure different things.
• Mx-PRx discordance interpreted as one being "wrong": discordance is often physiology (microvascular shunting, mixed compartments), not error.
• Mx unreliable during post-arrest TCD degradation: TCD spectrum changes during cerebral circulatory arrest; the slow-wave correlation does not apply.
• Treating Mx as a treatment target itself: Mx is a guide to CPP / MAP target setting, not a direct treatment goal.

14. Combine with…#

• TCD: the parent modality; Mx is one of its derived indices.
• PRx: the invasive sibling; mutual validation when ICP is in place.
• COx: the NIRS-based non-invasive partner; COx is more practical in neonates.
• CPP: the upstream variable that Mx-CPPopt targets.
• CPPopt: the dedicated CPPopt page with workflow detail.
• Foundations: autoregulation: the physiology behind Mx, PRx, COx.

15. Evidence summary#

16. Recent literature (2022 to 2025)#

• Beqiri 2024 COGiTATE: randomised feasibility of CPPopt-targeted MAP management in adult TBI; provides class-A evidence for the CPPopt framework that Mx-CPPopt extends non-invasively. 34
• Tas 2022 pediatric Mx feasibility: 30 children with severe TBI; Mx computable and clinically actionable. 10
• Tas 2024 pediatric CPPopt: extended pediatric CPPopt feasibility; Mx-derived CPPopt as part of the toolkit. 33
• Rivera-Lara 2017 autoregulation review (still the standard primer): positions Mx in the triad with PRx and COx. 4
• Figaji 2025 pediatric MMM consensus: Mx and Mx-CPPopt recognised as tier-2 (specialist centre) modalities. 12
• Helbok 2024 pediatric MMM: bedside operationalisation, including pre-monitor / post-monitor Mx workflows. 13

17. Self-check#