Transcranial Doppler / TCCD

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

Vignette A. Pediatric SAH day 6, suspected vasospasm#

A 12-year-old presents with a ruptured AVM and aneurysmal SAH. On day 6 the nurse notices a subtle right-arm drift. You bring the TCD cart, slide the probe over the right temporal window, and within seconds the familiar saw-tooth velocity envelope appears. PSV 220, EDV 90, MFV 130, PI 1.0. The team checks the extracranial ICA: MFV 30. Lindegaard ratio 4.3, mild-to-moderate vasospasm. You alert neurosurgery, escalate haemodynamics, and schedule angiography. 1 2

Vignette B. Severe TBI, the ventriculostomy is delayed#

A 4-year-old fell from a roof. GCS 6, blown pupil on the right, CT shows a large temporal contusion with effacement of the basal cisterns. Theatre is 30 minutes away and a parenchymal monitor is not on the unit. You scan the left MCA: MFV 28, EDV 7, PI 1.8. The diastolic notch is shallow; the Bellner regression returns ICP ≈ 28 mmHg. You mannitol, raise the head, and call again. By the time the team is in theatre, TCD PI has fallen to 1.2. 3 4 5

Vignette C. Post-arrest HIE, day 2#

A newborn cools after a tight nuchal cord and meconium aspiration. Day 2, rewarmed. NIRS rSO₂ has risen to 88%. TCD MCA: PSV 110, EDV 60, PI 0.5, MFV 80. The diastolic shoulder is high and the envelope looks "lush." This is luxury perfusion: relative hyperaemia in a brain whose metabolism has collapsed. A low PI with high EDV in HIE is not reassurance; coupled with isoelectric aEEG it is one of the worst prognostic signatures. 6

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

TCD is a handheld pulsed-wave ultrasound at ~2 MHz, low enough to penetrate the thin parts of the skull, that measures the velocity of red blood cells in the basal cerebral arteries. The Doppler shift formula is:

where Δf is the frequency shift, f₀ the transmitted frequency, v the blood velocity, θ the angle of insonation, and c the speed of sound in tissue.

Two things follow immediately.

TCD measures velocity, not flow. Flow is velocity × cross-sectional area. The cross-section of the M1-MCA changes with vasospasm, hypocapnia, fever, sedation, and anaesthetic. If a vessel narrows by 50%, velocity must roughly quadruple to preserve flow. So a rise in MFV may mean the same flow through a smaller pipe (vasospasm), or more flow through an unchanged pipe (hyperaemia), or both. The Lindegaard ratio exists to discriminate these. 7 1

TCD requires an angle assumption. Blind TCD assumes cos θ ≈ 1 (the operator hand-aims for the loudest signal). TCCD images the vessel in B-mode first, corrects the angle, and is preferred where available, especially in children, where the temporal bone is thin and the vessels are close to the probe. 8 9

3. The four insonation windows#

Fig. 1

4. The spectral waveform: anatomy of a TCD trace#

A normal MCA spectral envelope has four readable features.

1. A sharp systolic upstroke, driven by the cardiac systole reaching the cerebral circulation, propagated through compliant proximal arteries.
2. A clean systolic peak (PSV), the maximum velocity recorded in that cardiac cycle.
3. A dicrotic notch, the small inflection between PSV and the diastolic trough; it corresponds to aortic-valve closure. In a very high-resistance circulation (e.g., severe raised ICP) the notch deepens and may disappear into a true reverse-flow trough.
4. A diastolic shoulder falling to EDV, the floor of the cardiac cycle. EDV is the most sensitive single number to changes in downstream resistance: it falls when ICP rises, when small-vessel vasoconstriction occurs (hypocapnia), or when distal capacitance is lost.

The mean (MFV) is the time-weighted average of the entire envelope across the cycle. A common bedside approximation, accurate to within ~5%, is:

PI (Gosling) captures the pulsatility, the height of the envelope relative to its mean. RI is similar but normalised to PSV. Both rise when distal resistance rises. PI ~ 0.5–1.1 is the healthy range across most ages. PI > 1.4 is "high"; PI > 2.0 with low or absent EDV is the territory of impending arrest. 3 11

5. The numbers: what to record, in what order#

For every patient, on every side, record this six-pack:

Record both sides and look for asymmetry > 30%. Asymmetric MFV is the bedside signature of unilateral vasospasm, embolic occlusion, or unilateral raised ICP from a mass lesion.

6. What is normal? Age-banded reference values#

Healthy MCA MFV is highest in early childhood, when cerebral metabolic rate per gram of brain peaks, and falls into adolescence and adulthood.

Sources: 12 13 14 15. LLA / ULA columns are bedside heuristics; autoregulation in children is narrower and lower than the adult Lassen plateau, and the lower limit is dangerously close to baseline MAP in infants. 16

7. What is abnormal? A small pattern library#

Decision tree: "what question am I asking?"#

flowchart TD
  Reading[TCD spectrum] --> Why{What is asked?}
  Why -->|SCD stroke screening| TAMV[TAMMV]
  TAMV --> SCDtier{Value}
  SCDtier -->|< 170| Normal[Normal, annual screen]
  SCDtier -->|170–199| Cond[Conditional, repeat sooner]
  SCDtier -->|≥ 200| Abn[Abnormal, chronic transfusion per STOP]
  Why -->|Vasospasm in SAH| LRcalc[Lindegaard ratio]
  LRcalc -->|LR > 3 + high MFV| Spasm[Vasospasm; treat]
  LRcalc -->|LR < 3 + high MFV| Hyper[Hyperaemia; look for cause]
  Why -->|Raised ICP suspected| PIcheck[PI + EDV]
  PIcheck -->|PI > 1.4, low EDV| Resist[Likely high ICP / low CPP]
  PIcheck -->|Oscillating EDV| Arrest[Pre-arrest; escalate]
  Why -->|Autoregulation| Mxcalc[Mx slow-wave correlation]
  Mxcalc -->|Mx < 0.3| Intact[Intact]
  Mxcalc -->|Mx > 0.3| Impaired[Impaired; re-target CPP]

8. Try it: interactive widgets#

9. TCD-guided blood-pressure and CPP management#

This is where TCD earns its place in the multimodal stack. Two TCD-derived indices, Mx (autoregulation index) and CCP (critical closing pressure), let you reason about this patient's upper and lower BP limits without an invasive ICP monitor.

9.1 Can TCD set BP targets on its own?#

Short answer: not by itself, but it can narrow the range.

• MFV trend across BP changes: If MFV moves linearly with MAP (passive flow), the patient is outside their autoregulatory plateau: either MAP is too low (below LLA) or too high (above ULA), or autoregulation is broken entirely. If MFV stays flat across MAP swings, autoregulation is intact and current MAP is on the plateau.
• PI as a CPP / ICP proxy: PI rises as CPP falls. The Bellner regression (ICP ≈ 10.93 · PI − 1.28) is the most cited bedside formula, with confidence intervals wide enough that it is a triage tool, not a measurement tool. 3 11
• Trend over absolute: A 50% rise in PI from baseline is informative; a single PI of 1.3 is not. 4

9.2 Mx, the TCD-based autoregulation index#

Mx is the moving-window Pearson correlation between MFV (TCD) and CPP (or MAP, if ICP is unavailable) sampled at slow-wave frequencies (~0.01–0.05 Hz, 5–10 second averages over a 5-minute window).

• Mx ≈ −0.3 to 0.0: intact autoregulation (MFV does not track MAP)
• Mx > +0.3: impaired autoregulation (MFV passively tracks MAP)
• Mx vs CPP curve: U-shape with a minimum at CPPopt

The CPP at which Mx is minimised is the TCD-derived CPPopt: the perfusion pressure at which this patient's autoregulation is most efficient. The Cambridge CPPopt approach formalises this. 19 20 21 22

Practical CPPopt-by-TCD workflow (when an ICP monitor isn't available or you want a second opinion against PRx):

1. Place the TCD probe in a robotic / fixed headframe over the MCA.
2. Record continuous MFV synchronously with arterial MAP for ≥ 4 hours.
3. Bin MAP / CPP into 5-mmHg windows; compute Mx in each window.
4. Fit a parabola to (CPP, Mx); the vertex is CPPopt.
5. Re-target MAP so that CPP sits within ±5 mmHg of CPPopt.

Where ICP is available, PRx (ICP–MAP correlation) is preferred because it is closer to the mechanism; Mx is the non-invasive fallback. 23 22 24

9.3 Critical closing pressure (CCP)#

CCP is the MAP at which forward CBF ceases. Mathematically, the linear extrapolation of (instantaneous pressure, instantaneous velocity) back to v = 0 returns the closing pressure of the cerebrovascular bed. CCP is composed of ICP plus a wall-tension term reflecting smooth-muscle tone. 25

• CCP ~ 20–35 mmHg in healthy adults
• CCP rises with raised ICP and with sympathetic activation
• CCP > MAP means no forward flow, i.e., circulatory arrest

CCP from TCD is mostly a research tool, but it is the cleanest way to estimate MAP-opt = CCP + (CPPopt − ICP) when an arterial line and a TCD probe are co-located. 24

9.4 The bedside synthesis#

A pragmatic, PICU-friendly bedside synthesis when ICP is not available:

1. Establish the child's baseline MFV in the first 30 minutes after admission, in stable haemodynamics.
2. Define a PI ceiling (typically baseline PI + 0.3) above which you will assume ICP has risen.
3. If MAP falls and MFV falls proportionally, you are below LLA; give a fluid bolus or vasopressor to lift MAP back into the plateau.
4. If MAP rises and MFV rises proportionally, you are above ULA or autoregulation is lost; lower MAP gently (10–15%) and re-measure.
5. Document the MFV/MAP couple every nursing shift. A trending impairment (Mx drifting positive) is itself an alert.

10. Clinical contexts: TCD across acute brain injuries#

10.1 Aneurysmal SAH and post-haemorrhagic vasospasm#

The historical home of clinical TCD. Vasospasm typically develops between days 3 and 14 post-bleed, peaks around day 7, and resolves over the second-to-third week. TCD is used daily during this window.

• MCA MFV thresholds (adult): < 120 cm/s normal · 120–180 mild · 180–200 moderate · > 200 severe.
• Lindegaard ratio: > 3 confirms vasospasm; > 6 suggests severe.
• Rate of rise > 50 cm/s per 24 h is a stronger predictor of clinical DCI than the absolute number.
• Sensitivity vs DSA is highest for MCA vasospasm (~85%) and lower for ACA and posterior circulation. 26
• AHA/ASA SAH guidelines include TCD as a recommended monitoring modality. 27

Pediatric vasospasm is less common (AVM and trauma more frequent than aneurysm) and uses ratios plus within-child trend rather than adult absolute thresholds. 15

10.2 Severe TBI#

TCD in TBI is mostly about early ICP triage before an invasive monitor is placed, and about autoregulation monitoring afterwards.

• Pre-hospital and emergency-department TCD identifies abnormal cerebral haemodynamics within minutes of arrival. Tazarourte's pilot and Bouzat's 2016 cohort show that PI > 1.4 or EDV < 25 in the first 8 h predicts poor neurological outcome at 3 months. 28 5
• ICP screening: TCD can exclude intracranial hypertension with high negative predictive value when PI is low. Rasulo's 2022 IMPRESSIT-2 multicentre study found the TCD-derived ICP estimate excludes intracranial hypertension with a high NPV, though with poor quantitative concordance (~33%). 4
• CPPopt by Mx complements PRx and is the standard fallback when ICP is not yet placed.

10.3 Pediatric arterial ischaemic stroke (AIS)#

Continuous TCD in proximal MCA occlusion shows absent flow, low-velocity dampened flow, or post-recanalisation hyperaemia. In children, TCD also helps in:

• Confirming flow re-establishment after thrombolysis or thrombectomy.
• Monitoring for moyamoya progression (rising MFV, ipsilateral hyper-collateralisation).
• Tracking post-recanalisation hyperperfusion that needs BP lowering. 10

10.4 Hypoxic-ischaemic encephalopathy (HIE): neonatal and post-arrest#

In severe HIE, TCD evolves through a sequence:

1. Early (hours 0–12): low PSV and low MFV; true low perfusion.
2. Reperfusion / luxury phase (12–48 h): high EDV, low PI, high MFV; pathological hyperaemia in a brain whose metabolism has collapsed.
3. Recovery or progression: either normalisation, or rising PI signifying oedema and falling CBF.

A persistently low PI (< 0.6) with high diastolic flow at 24–48 h post-arrest is associated with poor neurodevelopmental outcome in pediatric cardiac-arrest survivors. 6

10.5 Pediatric ECMO#

VA-ECMO produces a non-pulsatile circulation; TCD signatures change accordingly. The systolic peak flattens; PI falls toward zero. TCD on ECMO is used to:

• Detect left-ventricular ejection recovery (returning pulsatility).
• Identify emboli (HITS, high-intensity transient signals).
• Monitor flow asymmetry that can predate stroke. 29

10.6 Meningitis and encephalitis#

TCD adds value in two situations: (a) acute intracranial hypertension from cerebral oedema, with high PI and low EDV; (b) vasculitic vasospasm from bacterial or tuberculous meningitis, with high MFV and LR > 3. Both signatures change the threshold for imaging and for ICP placement. 30

10.7 Brain-death determination#

TCD is an ancillary test in the World Brain Death Project framework. The two qualifying patterns are:

1. Oscillating / reverberating flow: small forward systolic spikes with reverse end-diastolic flow.
2. Systolic spikes only: short forward systolic peaks with no diastolic flow.

A single examination by an experienced operator with confirmation 30 minutes apart, in both anterior and posterior circulations, completes the test. Loss of previously-recorded flow is supportive but not sufficient alone. 18 17

10.8 DKA cerebral oedema#

A pediatric case where TCD's earliest signs (rising PI before clinical herniation) can change management. Cerebral oedema in DKA classically presents 4–12 h into rehydration. Bedside TCD documenting rising PI prompts mannitol / hypertonic saline before pupillary signs appear. 31 32

10.9 Sickle cell disease#

The STOP and STOP-2 trials made TCD the standard-of-care screening tool for stroke risk in children with SCD ages 2–16. TAMMV ≥ 200 cm/s triggers chronic transfusion to prevent first stroke; 170–199 is conditional and warrants re-screening. 33 34

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

12. Setup and technique: a step-by-step#

12.1 Equipment#

• 2 MHz pulsed-wave probe (blind TCD) or TCCD probe (~ 2–4 MHz with B-mode capability).
• Coupling gel.
• Robotic / fixation headframe (DWL Doppler-Box, Atys Robotic, or Dolphin) for continuous monitoring.
• Synchronised arterial-line pressure and (where present) ICP feeds.

12.2 Transtemporal MCA: the routine view#

1. Position the patient supine, head neutral, head-of-bed 30°.
2. Coupling gel above the zygomatic arch, just anterior to the tragus.
3. Probe: 2 MHz pulsed-wave. Initial depth 45 mm (adult) / 30 mm (toddler) / 40 mm (school-age).
4. Find the signal: angle anterior and slightly cephalad. Listen for the characteristic systolic "whoosh-thump"; the loudest signal is usually the MCA.
5. Confirm with carotid tap: ipsilateral carotid compression in the neck momentarily dampens the MCA envelope and confirms vessel identity.
6. Optimise: insonation angle < 30° (cosine error grows fast above 30°). Maximise spectral envelope brightness without overgaining.
7. Record PSV, EDV, MFV, PI over ≥ 5 cardiac cycles.
8. Repeat contralaterally.
9. Lindegaard denominator: switch to submandibular view; ICA at depth ~50–60 mm; compute MFV(MCA) ÷ MFV(ICA).

12.3 Suboccipital basilar#

• Head flexed, chin to chest.
• Depth 75–95 mm for the mid-basilar.
• Flow is away from the probe (negative on the envelope); confirm direction.

12.4 Transorbital ophthalmic / ICA siphon#

• Use a low mechanical index (< 0.23) to protect the lens.
• Probe directly over the closed eyelid with a generous coupling cushion of gel.
• Depth 40–60 mm for ophthalmic, 60–80 mm for the siphon.
• Note flow direction: ophthalmic toward the probe is normal; reversal suggests proximal ICA stenosis.

12.5 Continuous monitoring#

A fixed headframe holds the probe over the MCA for hours. With synchronised arterial pressure recording, you can compute Mx continuously and run a TCD-CPPopt loop without invasive ICP. Pair with NIRS rSO₂ for a "non-invasive autoregulation bundle." 21 22

12.6 TCCD vs blind TCD#

• TCCD provides B-mode anatomical localisation plus colour Doppler plus angle correction. Preferred in children (thin temporal bone makes B-mode easy) and in adults with poor windows.
• Blind TCD is faster, lower in equipment cost, and the standard tool for STOP screening (which uses non-angle-corrected TAMMV). 9 8

13. Pitfalls: operator-dependent, physiology-dependent#

• Window quality varies by age, sex, ethnicity. Up to 10% of post-menopausal women have inadequate temporal windows; a contrast agent or a different window can rescue them.
• Hyperaemia mimics vasospasm without the Lindegaard ratio.
• Vessel-diameter change confounds velocity-flow proportionality; this is the entire reason for ratio-based interpretation.
• PI is not ICP. PI rises with anything that increases distal resistance. Use it as a triage, not a measurement. 11
• Operator dependence: inter-rater reliability is moderate even among trained operators.
• Sedation, anaesthesia, hypothermia lower MFV via CMRO₂ effect (15–25%).
• Hypocapnia raises PI and lowers MFV through hypocapnic vasoconstriction.
• Anaemia raises baseline MFV; every TCD value should be paired with the haemoglobin of the day.
• TCCD angle-correction systematically over-reads TAMMV in SCD; turn it off for STOP screens.
• Probe coupling drift during continuous monitoring shows as gradual envelope dimming; re-aim every 1–2 hours or use a robotic frame.

14. Combine with…#

• Lindegaard calculator: distinguishes spasm from hyperaemia.
• Mx, the TCD-based autoregulation index
• Non-invasive ICP: Bellner regression uses PI.
• Integration: TCD vs ICP vasospasm: worked scenario.
• PRx vs COx discordance: non-invasive autoregulation pairings.
• Foundations: autoregulation: Mx, PRx, ORx mechanism.

15. Evidence summary#

16. Recent literature (2022–2025)#

• Brasil 2022: non-invasive ICP from TCD waveform analysis (Brain4Care-style devices); moving from research into bedside. 41
• Cardim 2023, Rasulo 2024: multicentre validation of non-invasive ICP estimators against parenchymal probes. 42 43
• Figaji 2025 pediatric MMM consensus: formalises TCD as a tier-2 modality in resource-stratified pediatric centres. 35
• TCCD adoption: colour-coded duplex is replacing blind TCD for routine ICU bedside use, particularly in children.
• Robotic TCD platforms: fixed-frame continuous TCD now standard in many neuro-ICUs; enables Mx, CPPopt-by-TCD, and embolic-signal counting on ECMO.
• Non-invasive ICP exclusion: Rasulo 2022 (IMPRESSIT-2) shows the TCD-derived ICP estimate is a high-NPV rule-out for intracranial hypertension, but with poor quantitative concordance; a low PI is reassuring, while PI > 1.4 alone is not diagnostic. 4

17. Self-check#