= Thermodilution
\b[Thermodilution] estimates <blood flow> from the temperature change produced by introducing fluid at a different <temperature>. It is a form of <indicator dilution>: the transported signal is a thermal deficit or excess rather than a dye concentration. An upstream <catheter> introduces the fluid and downstream <temperature measurement> records its dilution in flowing <blood>.
Two arrangements illustrate the principle. A finite injection produces a temperature-versus-time curve; a continuous infusion can produce an approximately steady downstream temperature change. A primary coronary-flow study used room-temperature saline, a 2.8-F infusion <catheter> and a 0.014-inch pressure/temperature guidewire, testing the method in five instrumented dogs and 35 patients.<Direct volumetric blood flow measurement in coronary arteries by thermodilution by Wilbert Aarnoudse and others>[Abstract, Methods]{ref} The quantities observed were infusion rate, infused-fluid temperature at the <catheter> tip and distal mixed-blood temperature.
For an idealized finite-injection model, let $H$ be the injected thermal deficit in joules, $\kappa$ the <volumetric heat capacity> of the moving fluid, and $\Delta T(t)$ the downstream temperature deficit. Assume steady volume <blood flow> $Q$, complete passage of the signal through the detector, representative mixing and negligible <heat transfer> to other structures. <Conservation of energy> then gives
$$
H=\kappa Q\int_0^\infty\Delta T(t)\,dt,\qquad
Q=\frac{H}{\kappa\int_0^\infty\Delta T(t)\,dt}.
$$
The <time integral>, rather than the maximum temperature change, determines this model's flow estimate. With stipulated $H=1$ J, $\kappa=4$ J/(mL K) and curve area $0.25$ K s, the inferred <blood flow> is $1$ mL/s, or $60$ mL/min. These values illustrate the balance, not a clinical protocol.
\Image[/thermodilution-equal-area-curves.png]
{title=Equal thermal-indicator area with different peak temperatures}
{description=Synthetic deficits $\Delta T(t)=At\exp(-t/\tau)/\tau^2$, with $A=1$ K s and $\tau=1$ or $2$ s. Both integrals over all time equal $A$; the plotted window ends at 16 s. Different peak heights therefore give the same ideal flow at fixed thermal input.}
This ideal calculation exposes measurement assumptions. If only a fraction $\eta$ of the thermal signal reaches the detector, the observed curve area is reduced by that factor and the uncorrected estimate becomes $\widehat Q=Q/\eta$. For example, $\eta=0.8$ would cause a $25\%$ overestimate in this model. <Coronary thermodilution mixing assumptions> similarly determine whether a local sensor reading represents the transported thermal signal. Accurate <temperature measurement> alone cannot establish those conditions.
Back to article page
