Introduction

Positron emission tomography (PET) provides regional, quantitative measurements of important parameters such as cerebral blood flow (CBF) and oxygen metabolism, as well as molecular imaging using specific physiologic and pathologic chemical compounds (molecular imaging), in living humans. The ability to quantitatively measure regional oxygen metabolism and oxygen extraction fraction (OEF) remains unique to PET imaging. OEF increases in response to a reduction in the delivery of oxygen to normal brain tissue. In patients with atherosclerotic occlusive disease, this occurs when collateral channels are insufficient to maintain normal blood flow. Increased OEF is a powerful and independent predictor of future stroke in patients with atherosclerotic carotid artery occlusion and there is an ongoing clinical trial of surgical bypass in this population.

PET imaging physics

PET imaging requires three components: a positron-emitting isotope (radiotracer), a tomographic imaging system to detect the location and to measure the quantity of radiation, and a mathematical model relating the physiological process under study to the detected radiation.^1,2 For example, the method used in our laboratory for the measurement of cerebral blood flow uses a bolus injection of O-15 labeled water (H₂¹⁵O, the radiotracer).³ The PET camera system records the location and number of counts during the circulation of the water through the brain. Finally, the tomographic PET images of raw counts are converted into maps of regional quantitative CBF using computer algorithms. This processing requires measurement of arterial blood counts and incorporates models and assumptions regarding the transit of water through the cerebral circulation.

Radiotracers are radioactive molecules administered in such small quantities that they do not affect the physiologic process under study. PET radiotracers decay by positron emission and may be separated into two broad categories: normal biological molecules, such as ¹⁵O-labeled water, or non-biologic elements attached to organic molecules as radiolabels, such as ¹⁸F-labeled deoxy-glucose (FDG). PET imaging detection systems use the phenomenon of annihilation radiation to both localize and to measure physiologic processes in the brain. In the body, the positron (a positively charged electron emitted by the radionuclide) travels up to a few millimeters before encountering an electron. This encounter results in the annihilation of both the positron and electron and the consequent generation of two gamma photons of equal energy. These two photons are emitted in characteristic 180-degree opposite directions. A pair of detectors positioned on either side of the source of the annihilation photons detects them simultaneously. This allows localization of the point source of the radiation.

The most important limitations of PET imaging of physiologic processes relate to the phenomenon of full-width, half-maximum (FWHM) and a related phenomenon of partial-volume averaging. Detected radiation is observed over a larger area than the actual source. The spread or distribution of activity is approximately Gaussian for a point source of radiation, with the maximum located at the original point. The FWHM describes the degree of smearing of radioactivity in a reconstructed image. The ability of a PET scanner to discriminate between two small adjacent structures or accurately measure the activity in a small region will depend on the FWHM of the system as well as the amount and distribution of activity within the region of interest and the surrounding areas. Because of the smearing or redistribution of detected radioactivity, any given region in the reconstructed image will not contain all the activity actually within the region. Some of the activity will spill over into adjacent areas. This phenomenon is known as the partial volume effect. An important consequence of this principle is that PET will always measure a gradual change in activity where an abrupt change actually exists, such as in an infarct or hemorrhage, or at the border of different structures like brain and CSF or gray and white matter.⁴

Finally, the externally measured tissue concentration of the positron emitting radiotracer (PET counts) is quantitatively related to the physiologic variable under study by a mathematical model. The PET scanner measures the total counts in a volume of tissue over time. The model then calculates how that measured activity reflects the physiologic parameter under study. These calculations account for several factors related to the tracer biomechanics and metabolism. These factors include the mode of tracer delivery to the tissue, the distribution and metabolism of the tracer within the tissue, the egress of the tracer and metabolites from the tissue, the recirculation of both the tracer and its labeled metabolites, and the amount of tracer and metabolites remaining in the blood.

Normal cerebral hemodynamics and metabolism

A brief introduction and definition of the common physiologic parameters measured with PET is useful prior to the discussion of normal hemodynamics and metabolism. Cerebral blood flow (CBF) the volume of blood delivered to a defined mass of tissue per unit time, generally in milliliters of blood per 100 g of brain per minute (ml/100g/min) (Figure 3–1). ¹⁵O-labeled water is the most commonly used tracer for measurements of CBF and the method used in our laboratory.³ Cerebral blood volume (CBV) is the volume of blood within a given mass of tissue and is expressed as milliliters of blood per 100 g of brain tissue. Regional CBV measurements may serve as an indicator of the degree of cerebrovascular vasodilatation, as discussed further in this chapter. CBV can be measured by PET with either trace amounts of ¹⁵O-labeled carbon monoxide or ¹¹CO.⁵ Both carbon monoxide tracers label the red blood cells. Blood volume is then calculated using a correction factor for the difference between peripheral vessel and cerebral vessel hematocrit. Mean transit time (MTT) is usually calculated as the ratio of CBV/CBF. By the central volume theorem, this ratio yields mean transit time, the hypothetical mean time for a particle to pass through the cerebral circulation. Increased MTT is used as an indicator of autoregulatory vasodilation. Some PET groups have advocated the use of the inverse of this ratio instead.⁶

Figure 3–1 Severe hemodynamic impairment. These images show a unilateral reduction in CBF distal to a complete atherosclerotic right carotid artery occlusion. This patient was a 70-year-old male presenting with TIA. Brain CT showed no evidence of stroke. CBV is elevated owing to autoregulatory vasodilation. CMRO₂ is relatively preserved owing to the increase in OEF.

Oxygen extraction fraction (OEF) is the proportion of oxygen delivered that is extracted by tissue for metabolism. In the brain, OEF normally varies between 0.25 and 0.5, with values over 0.5 signifying increased extraction. It is measured in our laboratory by an O¹⁵O inhalation scan and independent measurements of CBF and CBV⁷ (Figure 3–1). The CBF accounts for the amount of oxygen delivered to the brain. The CBV corrects for oxygen in the blood that is not extracted. An alternative count based method uses the ratio of the counts after an O¹⁵O inhalation scan to the counts from an O¹⁵O water scan, without CBV correction.^8–11 Other similar methods are also in common use. Cerebral metabolic rate of oxygen (CMRO₂) is the amount of oxygen consumed by tissue metabolism, measured in milliliters of oxygen per 100 g of brain tissue per minute⁷ (Figure 3–1). CMRO₂ is equal to the CBF multiplied by OEF and the CaO₂ (delivery of oxygen times the fraction extracted times the amount of available oxygen).

Whole-brain, mean CBF of the adult human brain is approximately 50 ml per 100 g per minute. Functional activation increases local or regional CBF, but global CBF generally remains unchanged. Under normal conditions any change in regional CBF must be caused by a change in regional vascular resistance. Vascular resistance is mediated by alterations in the diameter of small arteries or arterioles. In the resting brain with normal perfusion pressure, CBF is closely matched to the metabolic rate of the tissue. Regions with higher metabolic rates have higher levels of CBF. For example, gray matter has a higher CBF than white matter. While there is wide variation in levels of flow and metabolism, the ratio between regional CBF(rCBF) and metabolism is nearly constant in all areas of the brain. Consequently, the maps of OEF from the blood show little regional variation.¹² One exception to this is seen with physiological activation, where blood flow increases well beyond the metabolic needs of the tissue. This leads to a relative decrease of OEF and a reduction in local venous deoxyhemoglobin.¹³ This phenomenon is the basis for the use of magnetic resonance imaging (MRI) as a means to map brain function.