Instrumentos de la IE

Perfusion is a fundam ental physiological process that is sensitive to tissue viability and function, and provides im portant inform ation regarding a broad range of pathologies. Traditionally, techniques of blood flow m easurem ent have made use o f the w ash-in and w ash-out kinetics of perfusion tracers. M R I in com bination with intravascular tracers such as G d-DTPA can be used in a sim ilar way to these techniques (see T able 2.2) to obtain inform ation concerning the cerebral vasculature (dynamic susceptibility contrast im aging (DSC-M RI)) (Rosen, 1989; Rosen, 1990; Rosen, 1991). M easurem ents of cerebral blood volum e (CBV) and the mean transit time (MTT) of the tracer through the capillary network are possible if the arterial input function (AIF) can be accurately determ ined. D irect quantification of flow is, how ever, difficult with these non- diffusable tracers due to factors such as tracer dispersion and recirculation (0 stergaard , 1996). A model of the vascular topology may be necessary for the m easurem ent of the M TT that is a prerequisite for the determ ination of CBF (W eiskoff, 1993). D econvolution o f the signal-tim e relationship with the arterial input function is required and although studies have reported absolute CBF values using such m ethods (Rempp, 1994; Schreiber, 1998), relative or cross-calibrated flow s may prove to be the only realistic m easurem ent with D SC -M R I (W eiskoff, 1993; 0 sterg aard, 1998). The technique is also ham pered by its invasive and dynam ic, character that necessitates a

C h a p te r 3 P e rfu sio n

trade-off between temporal resolution and SNR. The first pass of the tracer in the tissue volum e of interest (VOI) and the dynam ic changes associated with the A IF both require accurate characterisation with rapid sequential im aging during the w ash-in and w ash-out of the tracer.

The past 10 years has seen the developm ent of alternative, non-invasive N M R techniques for quantitative m easurem ents of cerebral perfusion that use tissue w ater as an endogenous, freely diffusable tracer. The techniques are based on the differentiation of flow ing arterial w ater spins and static tissue spins by the m agnetic labelling of one com partm ent with respect to the other. The blood w ater spins are delivered to the brain voxel where they are extracted from the capillary bed and join the larger pool o f tissue water. The exchange between the m agnetically differentiated w ater in blood and tissue leads to a change in tissue m agnetisation that can be detected by M RI. The goal of the so-called arterial spin-labelling (ASL) m ethods is to extract and analyse the flow -related m agnetisation change. The tagged images are alternated with control im ages in which the flow label is not applied. The signal difference between these tw o im ages directly reflects local quantitative perfusion since the static tissue signal is elim inated. The sensitivity of the ASL m ethod is an order of m agnitude sm aller than for D SC -M R I (signal changes of approxim ately 1-5% and 30-40% for A SL and D SC im aging respectively).

There are two distinct classifications of the A SL techniques: continuous and pulsed. Continuous arterial spin labelling (CASL) m ethods utilise the tagging o f arterial blood w ater at the level of the feeding arteries by continuous saturation (Detre, 1992) or, more com m only, by inversion using adiabatic fast passage (AFP) (see Section 1.4.1; Dixon, 1986). The blood arriving at the im aging plane in the brain contributes to a steady state level o f tissue m agnetisation that is low er than the equilibrium level by an am ount that is proportional to perfusion.

Pulsed arterial spin labelling (PASL) techniques utilise a single, relatively short RF pulse to create the m agnetic label. Variants of this non-steady-state classification of techniques have been developed that invert a thick slab of tissue proxim al to the im aging slice so that the inflow ing arterial blood is inverted with respect to the static tissue. During the subsequent delay period denoted the TI interval, the tagged blood flows into the im aging plane and the resultant, dynam ic change in m agnetisation is

C h a p te r 3 P e rfu sio n

related to the delay and to the level of perfusion. The PA SL techniques o f E P IS T A R (Edelm an, 1994), PICORE (W ong, 1997) and U N FA IR (Helpern, 1997) are based upon this approach and differ with respect to the spatial localisation of the inverted slab and the com bination of RF and gradient strength em ployed in the control im age. A conceptually similar approach to these PA SL techniques is the variant proposed by K w ong et al. and later, Kim who denoted it the flow -sensitive alternating inversion recovery (FAIR) technique (Kwong, 1992; Kwong, 1995; Kim, 1995), in w hich the tissue is inverted with respect to the blood during the tagging experim ent and the signal detection takes place within that zone of inverted tissue. The flow -w eighted signal is obtained by the acquisition of an inversion recovery (IR) image with a slice selective

inversion that creates the m agnetic label. This im age is subtracted from a non-selective

IR im age in which both the blood and tissue pools are inverted so that the difference signal reflects the m agnetisation delivered by flow to the tissue voxel during the TI interval.

The flow -induced signal is only a small proportion of the overall signal (approxim ately 1-5%) and the isolation of this signal places great dem ands on the M R I system . B oth continuous and pulsed A SL techniques potentially suffer from a variety o f system atic errors. These will be described in this chapter as they becam e apparent upon im plem entation of the techniques. The sources o f error differ in their significance betw een the various methods and include transit tim e effects, M TC and vascular contam ination.