Comprehensive Study Notes on Diffuse Correlation Spectroscopy (DCS) for Cerebral Blood Flow Monitoring

Significance of Cerebral Blood Flow and Introduction to Diffuse Correlation Spectroscopy

• Cerebral well-being is heavily dependent on adequate cerebral blood flow (CBF) to ensure oxygen delivery and the clearance of metabolic byproducts.

• The brain consumes approximately 20% of the body's total oxygen, primarily through aerobic metabolism process.

• CBF serves as a critical physiological liaison between metabolic demand, oxygen supply, metabolic consumption, and byproduct clearance.

• Impaired CBF is a direct cause of clinical conditions including ischemic stroke; its dynamic regulation is essential for brain auto-regulation.

• Cerebral Auto-Regulation (CAR) manages CBF by modulating vascular resistance amidst local pressure variations. Assessment of CAR often involves examining the correlation between cerebral perfusion pressure (CPP) and CBF, where $CPP = MAP - ICP$ (Mean Arterial Pressure minus Intracranial Pressure).

• Diffuse Correlation Spectroscopy (DCS) is a relatively new optical technique designed for non-invasive measurement of microvascular CBF through an intact skull.

• DCS utilizes the temporal fluctuations of near-infrared (NIR) light to access CBF information, offering advantages of deep tissue penetration ($\sim 1\,cm$), high time resolution, and suitability for continuous bedside monitoring.

Comparative Modalities for Measuring Cerebral Blood Flow

• Current non-invasive imaging methods for CBF include the following, as adapted from Wintermark et al. (2005):

  • $^{15}O-H_2O$ PET Scanning: Requires 5 to 9 minutes acquisition time; provides whole-brain CBF; requires contrast agents, ionizing radiation, and arterial blood sampling for quantification. High cost.

  • SPECT (e.g., $^{99m}Tc-HMPAO$): 10 to 15 minutes acquisition; providing whole-brain CBF; uses ionizing radiation and contrast agents. High cost.

  • Xenon-CT (XeCT): 10 minutes acquisition; provides CBF in 6 cm thick sections; quantitative; moderate cost.

  • Perfusion CT (CT-P): 40 seconds acquisition; provides Mean Transit Time (MTT); uses contrast and radiation.

  • Dynamic Susceptibility Weighted MRI (DSC-MRI): 1 minute acquisition; provides MTT; requires contrast; high cost.

  • Arterial Spin Labeled MRI (ASL-MRI): 5 to 10 minutes acquisition; provides CBF; non-contrast and non-ionizing. Quantitative ($>10\,ml/min/100\,g$). High cost.

  • Transcranial Doppler Ultrasound (TCD): 10 to 20 minutes acquisition; measures Blood Flow Velocity (BFV) in large proximal arteries (Macro-vasculature); non-invasive, no contrast, low cost. Limited by bone "windows" and arterial diameter precision.

  • Diffuse Correlation Spectroscopy (DCS): 0.5 to 6 seconds acquisition; measures microvascular CBF; bedside portable, non-invasive, no contrast, no radiation, low cost. Coverage is limited to a few sites per hemisphere with a spatial resolution of $\sim 10\,mm$.

Physical Principles of Diffuse Correlation Spectroscopy

• DCS is a variation of Dynamic Light Scattering (DLS) or Diffusing Wave Spectroscopy (DWS) applied to multiple-scattering media like tissue.

• It analyzes temporal fluctuations of the speckle or light electric field $E(\mathbf{r}, t)$ emerging from tissue to determine the motion of scatterers, specifically red blood cells.

• Single-Scattering (DLS): The autocorrelation function decays by $1/e$ when scattering particles move by an average distance of approximately the wavelength ($\lambda$).

• Multiple-Scattering (DCS): The autocorrelation function decays by $1/e$ when the typical scattering particle moves by a distance of approximately $\frac{\lambda}{\sqrt{n}}$, where $n$ is the number of single-scattering events in the photon trajectory.

• Correlation Diffusion Equation: The transport of the temporal electric field autocorrelation function $\langle E^*(\mathbf{r}, t) \cdot E(\mathbf{r}, t + \tau) \rangle$ is governed by:   \nabla \cdot D(\mathbf{r}) \nabla G_1(\mathbf{r}, \tau) - (v\text{\mu}_a(\mathbf{r}) + \frac{1}{3} v \text{\mu}'_s k_o^2 \alpha ⟨ Δr^2(τ) ⟩) G_1(\mathbf{r}, \tau) = -v S(\mathbf{r})

  • $G_1(\mathbf{r}, \tau)$ is the temporal field autocorrelation function.

  • $\tau$ is the correlation time.

  • $k_o$ is the wave-number of light in the sample.

  • $\alpha$ is the fraction of photon scattering events from moving scatterers (e.g., red blood cells).

  • $D(\mathbf{r})$ is the photon diffusion coefficient, defined as D \approx \frac{v}{3\text{\mu}'_s}.

  • $\langle Δ r^2(τ) ⟩$ is the mean-square displacement of moving red blood cells, approximated as effective Brownian motion: $\langle Δ r^2(τ) ⟩ = 6 D_B τ$, where $D_B$ is the effective diffusion coefficient.

• The Siegert Relation: Relates the measured normalized intensity autocorrelation $g_2(τ)$ to the normalized electric field correlation $g_1(τ)$: $g_2(τ) = 1 + \beta |g_1(τ)|^2$.

  • $\beta$ is a constant determined by collection optics ($\beta \approx 0.5$ for unpolarized light/single mode fibers).

• Blood Flow Index (BFI): Obtained by fitting measured $G_2(\tau)$ data to solutions of the correlation diffusion equation. Relative BFI ($rBFI$) is defined as $BFI(t)/BFI(t_o)$.

Instrumentation and Hybrid Systems

• Light Sources: Continuous-wave (CW) narrow-bandwidth lasers, typically operating at $\sim 785\,nm$. A coherence length of at least $10\,m$ is required to exceed the spread of photon pathLengths.

• Fibers:

  • Source: Multi-mode fibers ($\sim 1\,mm$ diameter) to keep light below Maximum Permissible Exposure (MPE).

  • Detection: Single-mode fibers ($6.5\,\text{μ}m$ diameter) are traditional to optimize Signal-to-Noise Ratio (SNR) by collecting a single speckle.

• Detectors: Single-photon counting avalanche photo-diodes (SPADs), such as the SPCM-AQR(H) series. Electrical outputs are TTL pulses.

• Digital Autocorrelators: Hardware (e.g., Flex series) that tracks temporal separation between photons using a "multi-tau" delay-time scheme.

• Parallelization Progress: Dietsche et al. (2007) implemented a multi-speckle scheme using $N = 28$ single-mode fibers. This increases the DCS SNR by a factor of $\sqrt{N}$, facilitating deeper penetration or faster measurements.

• Hybrid DCS-NIRS/DOS: Combines DCS with Near-Infrared Spectroscopy (NIRS) or Diffuse Optical Spectroscopy (DOS). This allows for concurrent measurement of CBF and hemoglobin concentrations ($HbO_2$ and $Hb$).

Functional and Clinical Applications of DCS

• Functional DCS (fDCS): First demonstrated responses in human brain during motor stimuli (finger tapping), visual stimuli (checkerboard patterns), and verbal fluency tasks. Findings include a $\sim 40\%$ change in CBF compared to only $\sim 2\%$ change in hemoglobin concentration during identical stimulation.

• Steno-occlusive Lesions of the ICA: DCS identifies hemodynamic compromise by measuring Cerebrovascular Reactivity (CVR) during an acetazolamide (ACZ) challenge.

  • Case study (58-year-old): Impairment of CVR in left hemisphere shown by both TCD and DCS.

  • Case study (63-year-old): Preserved microvascular CVR (DCS) despite impaired macrovascular CVR (TCD), suggesting functioning collateral circulation.

• Carotid Endarterectomy (CEA): On-line monitoring during surgery. ICA clamping causes rapid decrease in ipsilateral rCBF and a "steal" effect (initial increase followed by decrease) on the contralateral side. DCS detects hypoperfusion faster and more reliably than EEG.

• Orthostatic Stress in Stroke Patients: Lowering the Head-of-Bed (HOB) angle reveals autoregulatory impairment. Per-infarct tissue typically shows greater rCBF change compared to the contralateral side. In $\sim 25\%$ of patients, a "paradoxical" decrease in CBF occurred when HOB was lowered, likely due to increased Intracranial Pressure (ICP).

• Absolute Calibration: Validation in piglets against indocyanine green (ICG) tracer methods showed excellent correlation (slope near unity), allowing the conversion of BFI ($cm^2/s$) into absolute flow units ($ml/min/100\,g$).

All-Optical Measurements of Cerebral Oxygen Metabolism (CMRO2)

• The hybrid combination of DCS and NIRS/DOS enables the calculation of relative changes in the cerebral metabolic rate of oxygen extraction ($rCMRO_2$).

• Equation for $rCMRO_2$:   $rCMRO_2 = rOEF \times rCBF \approx \left( \frac{SaO_2 - StO_2}{SaO_{2,bl} - StO_{2,bl}} \right) \times \left( \frac{SaO_{2,bl}}{SaO_2} \right) \times rCBF$

  • $StO_2$: Tissue blood oxygen saturation ($\frac{HbO_2}{THC}$).

  • $SaO_2$: Arterial oxygen saturation (via pulse-oximetry).

  • $bl$: Baseline values.

• Congenital Heart Defects (CHD): Studies in neonates with single-ventricle vs. two-ventricle defects. Post-cardiac surgery, single-ventricle neonates showed decreased $CMRO_2$ and $CBF$, whereas two-ventricle neonates showed increased $CBF$ and decreased $OEF$ with stable $CMRO_2$.

• Neo-natalogy: Research on premature and term brains demonstrates regional asymmetries in hemodynamic metabolism and changes in response to somatosensory functional stimuli.