Positron Emission Tomography

2.1.1.

General overview

P ositron E m ission T om ography (PET) is a technique that is used to ch aracterise the distribution o f radiolabelled probes in biological system s. Positron-em itting isotopes are produced by a cyclotron and used to label m olecules of a com pound o f interest w hose d istrib u tio n can then be exam ined directly. A tom s o f the p o sitro n -em ittin g isotope rad ioactively decay, em itting a positron and a neutrino. W hile neutrinos can only be detected by elab orate m ethods, positrons rapidly lose energy through co llision with electro n s in tissue and w ithin a very short distance are ann ih ilated through such a collision, giving rise to tw o high energy gam m a rays. The energy of these gam m a rays, that are em itted in opposite directions, is such that they escape the body and can be detected by a PET cam era. The cam era consists o f circum ferential arrays of scintillation detectors that identify coincidental em ission of gam m a rays on opposite sides of the head. By co m bining d ata from m any d ifferent detectors sim ultaneously and using a back projection algorithm the count density within the head can be determ ined.

Neural correlates of selective attention

The neuroscientific attraction of PET lies in the existence of positron em itting isotopes of carbon, nitrogen and oxygen, that are the m ajor elem ental constituents o f the hum an body. This gives PET considerable flexibility in labelling different biological substrates, including neurotransm itters and glucose. H ow ever, in the w ork presented here a single positron-em itting substance, H2'^0, was used. This tracer diffuses freely across the blood- brain barrier and so uptake in cerebral tissue is determ ined by the blood flow to that tissue rather than the diffusion rate of the tracer into tissue. A fter injection o f a bolus of

the tracer is extracted from plasm a into cerebral tissue on the first pass through the brain and the tracer uptake is highly correlated w ith cerebral blood flow . The PET im aging therefore takes less than two m inutes, and the half life o f is relatively short so repeated m easurem ents every 8-12 m inutes are possible.

2.1.2.

PET experimental design

T he studies described in this thesis w ere perform ed using the E C A T E X A C T H R + system (C T I S iem ens, K noxville, TN ) at the W ellco m e D ep artm en t o f C o g n itiv e N eu ro lo g y (Institute o f N eurology, U niversity C ollege L ondon). T he scan ner w as operated in 3D m ode with septa retracted (m eaning that coincidences are not detected only w ithin a 2D ring of detectors, but across all geom etric com binations o f detectors) w ith in creased sen sitiv ity to head counts p ro du ced by the use o f a N e u ro in se rt (G rootoonk et aL, 1996). A venous cannula to adm inister the tracer was inserted in the left antecubital fossa vein of the subject. A pproxim ately 350 M B q of Hj'^O in 3ml of norm al saline was loaded into intravenous tubing and flushed into subjects over 20s at a rate o f lOml/min by an autom atic pum p. A fter a delay o f approxim ately 35s, a rise in counts could be detected in the head that peaked 30-40s later (depending on individual circulation time). The interval between successive adm inistrations was Bmin. The data w ere acquired in one 90 second fram e, beginning 5s before the rising phase of the head curve. C orrection for attenuation was m ade using a transm ission scan collected at the beg inn in g o f each study. Im ages w ere reconstructed by filtered back projection (H anning filter, cut o ff frequency 0.5 cycles per pixel) into 63 im age planes (separation 2.4m m ) and into a 128x128 pixel im age m atrix (size 2.1mm)

Each study typically involved betw een five and nine subjects w ho w ere studied w ith tw elve separate m easurem ents of cerebral blood flow , every eight m inutes, over a two and a h alf hour period. All subjects gave inform ed consent for their participation in the

N eural correlates of selective attention

studies, that w ere approved by the N ational H ospital for N eurology & N eurosurgery ethics com m ittee. A t the end of each study, an anatom ical T1 w eighted M R I structural im age w as obtained for subsequent coregistration of the activation data. The radiation dose lim itations appropriate for the study of healthy norm al individuals with PET mean that each subject is usually scanned only tw elve times w ithin a single session. D uring an in d iv id u al scan, the ex p erim en tal p arad ig m is ty pically started m an u ally by the experim enter prior to data acquisition. The data are acquired in a single 30s fram e, so it is im portant that task perform ance is stable during this window. W ithin these constraints all potential types o f experim ental design can be accom m odated (Frackow iak et aL, 1998): in this thesis both sim ple 2x2 factorial designs and factorial param etric experim ents are presented.