Journal of Cerebral Blood Flow & Metabolism (2007) 27, 69–75 & 2007 ISCBFM All rights reserved 0271-678X/07 $30.00 www.jcbfm.com
Cerebral perfusion response to hyperoxia Daniel P Bulte, Peter A Chiarelli, Richard G Wise and Peter Jezzard FMRIB Centre, Department of Clinical Neurology, University of Oxford, Oxford, UK
Graded levels of supplemental inspired oxygen were investigated for their viability as a noninvasive method of obtaining intravascular magnetic resonance image contrast. Administered hyperoxia has been shown to be effective as a blood oxygenation level-dependent contrast agent for magnetic resonance imaging (MRI); however, it is known that high levels of inspired fraction of oxygen result in regionally decreased perfusion in the brain potentially confounding the possibility of using hyperoxia as a means of measuring blood flow and volume. Although the effects of hypoxia on blood flow have been extensively studied, the hyperoxic regime between normoxia and 100% inspired oxygen has been only intermittently studied. Subjects were studied at four levels of hyperoxia induced during a single session while perfusion was measured using arterial spin labelling MRI. Reductions in regional perfusion of grey matter were found to occur even at moderate levels of hyperoxia; however, perfusion changes at all oxygen levels were relatively mild (less than 10%) supporting the viability of hyperoxia-induced contrast. Journal of Cerebral Blood Flow & Metabolism (2007) 27, 69–75. doi:10.1038/sj.jcbfm.9600319; published online 3 May 2006 Keywords: arterial spin labelling; hyperoxia; MRI; perfusion
Introduction An increase in the inspired fraction of oxygen (FiO2) is known to be effective as a contrast agent for T2*weighted magnetic resonance imaging (MRI) (Losert et al, 2002; Rostrup et al, 1995). Unlike injected MRI contrast agents, oxygen has a much faster wash-out time and fewer contraindications. It is also readily available in most clinical settings and very economical. Oxygen is, however, not entirely passive when administered in this way. An increase in the fraction of inspired oxygen leads to an increase in the saturation level of arterial haemoglobin (SaO2), and in the partial pressure of oxygen dissolved in the arterial plasma (PaO2) (Berkowitz, 1997; Johnston et al, 2003b). This results in a multitude of physiologic and biochemical effects (Jensen, 2004; Watson et al, 2000), which alter the acidity of the blood, the binding of carbon dioxide and oxygen with haemoglobin, the partial pressures of oxygen and carbon dioxide in the tissue, plasma and expired gases, as well as changes in ventilation, metabolism, and cerebral blood flow (CBF) (Becker
Correspondence: Dr DP Bulte, FMRIB Centre, Department of Clinical Neurology, University of Oxford, John Radcliffe Hospital, Headington, Oxford OX3 9DU, UK. E-mail:
[email protected] Funding support provided by The Medical Research Council, UK, The Wellcome Trust, and The Rhodes Trust. Received 21 December 2005; revised 23 March 2006; accepted 24 March 2006; published online 3 May 2006
et al, 1996; Bohr et al, 1904; Christiansen et al, 1914; Johnston et al, 2003a, b; Poulin et al, 1998). The exact causes and relationships between these events are quite complicated and not directly essential to the use of hyperoxia as an MRI contrast mechanism. However, the magnitude of the change in CBF is of importance in determining the extent to which hyperoxic contrast can be considered to be a source of noninvasive, pure image contrast with little or no effect on metabolism or physiology. Ultimately, a decrease in the measured perfusion of grey matter regions in the brain will result. The magnitude and temporal dynamics of these changes to blood and tissue oxygenation, blood flow and volume, metabolism, and relaxation times for a given level of increased FiO2 must be determined for inspired oxygen to be a practical means of obtaining contrast for blood flow or volume studies. Ideally, a level of hyperoxygenation is sought that produces sufficient contrast for imaging purposes and yet does not significantly alter perfusion. If this can be achieved, then hyperoxia may be used as a slow infusion, intravascular contrast mechanism in boxcar protocols. Using dual BOLD/ASL acquisition sequences, such methods will open up a variety of techniques for obtaining more detailed, simultaneous information regarding changes in flow, volume, and metabolism. A characteristic response to hyperoxia is a reduction in end-tidal PCO2 (Becker et al, 1996; Floyd et al, 2003; Watson et al, 2000). A decrease in PETCO2 is accompanied by vasoconstriction in the
Cerebral perfusion response to hyperoxia DP Bulte et al 70
arterioles reducing CBF; however, the increase in PaO2 also has a direct vasoconstrictive effect independently of the PETCO2 response (Floyd et al, 2003; Kolbitsch et al, 2002). The temporal responses of these two parallel effects are different, as are the magnitudes. Both responses are dose dependent, but have absolute boundaries beyond which they no longer cause a change in CBF. It has also been shown that CBF will tend to restore back towards a normal level during hyperventilation-induced reductions in PETCO2, except when hyperoxia is also present in which case the CBF will remain at a constant reduced level (Poulin et al, 1998). The exact means by which altered levels of CO2 and O2 in the blood, plasma, and tissues affect blood flow are not fully known. It is also unclear whether it is the tissue, arterial, or venous levels that dominate the response under any given set of conditions. The degree to which these effects may be of concern to MRI studies depends on what type of pathology, structure, or function is being imaged. In this study, we investigated the change in cerebral perfusion induced by increasing the FiO2, using an arterial spin labelling technique. The inspired oxygen content was varied between normal levels and pure oxygen at approximately 20% steps to fill in the details of the CBF versus oxygen content curve, which is well described for the hypoxic region but is relatively unknown in the hyperoxic region.
Materials and methods Appropriate ethical approval was obtained for the study from the Oxfordshire Clinical Research Ethics Committee. Normal, healthy volunteers (N = 9, three male subjects) with a mean age of 2573.6 years were studied using a Varian 3 T MRI scanner. A Magnex head-dedicated gradient insert coil was used in conjunction with a birdcage RF coil tuned to 127.4 MHz. Perfusion imaging was performed using the QUIPSS2 MRI sequence (Wong et al, 1998), TR = 2 secs, TE = 22 ms, tag-excitation time (TI2) 1.4 secs, tag-saturation time (TI1) 0.7 secs, and 10 cm inversion slab 1.5 cm from the imaging slab. A total of 1,804 volumes were acquired in five axial slices, 6 mm thick extending in a superior direction from the thalamus. An EPI whole-brain scan and a T1-weighted whole-brain structural scan were also acquired for registration. Respiratory composition was monitored using a carbon dioxide analyser (Model CD-3A) and an oxygen analyser (Model S-3A) (AEI Technologies, Pittsburgh, PA, USA). Composition was continuously monitored and then endtidal values were extracted from the data set using code created in Matlab (Mathworks, Natick, MA, USA). Subjects wore a close-fitting mask over the mouth and nose, which completely sealed them from either inspiring or expiring outside of the system. Monitoring was from a port in a disposable filter directly connected to the facemask. Inspired gases were delivered via a multitube system, which mixed humidified medical-grade oxygen and air Journal of Cerebral Blood Flow & Metabolism (2007) 27, 69–75
from a medically approved compressor in a small chamber 30 cm from the subject’s mouth at a total delivery rate of 30 L/min with excess gases passing into a flood chamber. Gases were exhaled into the large flood chamber (a 2 m long, 10 cm diameter, open-ended tube which, owing to the high delivery rate, allowed an accumulation of the supplied gases to build up for inhalation, and yet provided an exhaust path that did not produce significant re-breathing of exhaled gases) in combination with an open vacuum extraction system for safety. The continuous monitoring of both inspired and expired gases showed that no measurable re-breathing occurred. A protocol of graded steps of FiO2 was administered consisting of 12 mins at each level (FiO2 = 0.21 (normoxic), 0.4, 0.6, 0.8, and 1.0). This enabled the subject to reach an apparently steady state of blood flow, blood volume, and metabolism (after B3 mins) and provide sufficient sensitivity for a perfusion measurement. Both a monotonically increasing step protocol (N = 6) and a randomised protocol (N = 3) were tested for comparison. Perfusion imaging data were postprocessed using FSL (FMRIB’s Software Library, www.fmrib.ox.ac.uk/fsl; Smith et al, 2004) and avwmaths tools (locally written nonMatlab code for standard fMRI analysis) incorporating motion correction (MCFLIRT), brain extraction (BET), and temporal sinc-interpolation of both tag and control perfusion images, followed by pairwise subtraction. A GLM-based estimation of activation intensity was performed using FEAT (FMRI Expert Analysis Tool) Version 5.42, part of FSL. Time-series statistical analysis was performed using FILM (Woolrich et al, 2001). Z (gaussianised T/F) statistic images (Figure 1) were thresholded using clusters determined by Z > 2.3 and a (corrected) cluster significance threshold of P = 0.05 (Worsley et al, 1992). Registration to high-resolution images was performed using FLIRT (Jenkinson et al, 2002; Jenkinson and Smith, 2001). The results of this latter analysis were only used to confirm that the majority of activation was confined to grey matter; the actual calculation of flow was performed using a dedicated Matlab script. Masks of segmented grey and white matter were calculated from coregistered, high-resolution T1-structural scans of each subject using the FMRIB automated segmentation tool (FAST) (Zhang et al, 2001). The pairwise subtracted images were analysed in Matlab to estimate the changes in flow. The four-dimensional data set was divided up into five blocks, one for each gas epoch. The first 3 mins of data from each block
Figure 1 Overlay of regions of significant apparent perfusion change of a representative subject at different FiO2 levels on a structural image, showing distinct alignment with grey matter regions.
Cerebral perfusion response to hyperoxia DP Bulte et al 71
were discarded and every volume was masked using a registered, matched-resolution binary mask of grey matter for that subject. The masked data were then temporally averaged over the 9 mins of the steady-state region of the epoch, to produce a three-dimensional data set of the average ASL signal in grey matter in each voxel in the volume. Each volume was then averaged to produce a grand mean signal in grey matter for each epoch. The percentage change in signal for each hyperoxic block relative to normoxia was then calculated to produce an estimate of the percentage change in the difference in magnetisation between tag and control images (DM(t)) for the subject during that block. The baseline value of DM(t) calculated by the kinetic model (Buxton et al, 1998) was then reduced by the same percentage as measured for each level of FiO2 to produce the expected values of DM(t) at those levels. The flow value was then adjusted within the calculation of the model until the expected value of DM(t) was produced. These calculations also took oxygenationdependent changes in NMR relaxation times into account, for which further details are provided in the next section.
Figure 2 Sampled oxygen trace for the randomised protocol. The data are very densely packed along the time axis, so individual breaths are not visible. However, as a result, the one side of the data represents the inspired levels and the other side the expired levels.
Results The mean perfusion estimate of the voxels in each segmented region was calculated for each slice and each subject at the five levels of inspired oxygen. Changes in perfusion under hyperoxic conditions were seen to be very regional in nature and from the segmentation interrogation were shown to occur predominately in grey matter (Figure 1), with little or no measurable change in white matter regions, in agreement with other studies (Kolbitsch et al, 2002). A comparison of the monotonic step protocol and the randomised protocol showed stark differences between the two approaches. The perfusion levels calculated during the randomised protocol were not comparable with the step protocol, but were also inconsistent relative to each other (the change in CBF at FiO2 = 0.8 was less than the change at 0.4). An analysis of the end-tidal measurements showed that although the PETO2 was reaching a steady state after B3 mins in each block (Figure 2), the time courses of the PETCO2 levels were strongly dependent on the degree of change in FiO2 (Figure 3). The discrepancy between the washout times of the two gases is thought to be owing to the different physiologic storage and transport of the molecules in vivo, combined with the Haldane effect altering the oxygen- and carbon dioxide-carrying ability of the blood. It was for this reason that the linearly increasing step protocol was chosen as the primary technique for investigation. The small and consistent step size between each level allows for mean relative changes in perfusion to be more reliably calculated so that the effects of O2 changes alone could be more easily assessed. To confirm that a relative steady state was reached during the step protocol after 3 mins, the PETCO2 respiratory data were analysed within single gas blocks. For exam-
Figure 3 End-tidal CO2 trace from a representative subject for the randomised protocol with empirical trend curves and blocks to represent FiO2. As can be seen, the time to plateau is strongly dependent on the size of the step, and very little stabilised data is available from the FiO2 = 0.8 and 0.4 periods.
ple, a linear fit to the portion of the data shown in Figure 4 that corresponds to the FiO2 = 0.8 block yields a slope of 0.38 during the first 3 mins of the block, but a slope of only + 0.025 from 3 to 6 mins. Similarly, a linear fit of the data from 3 mins until the end of the block produces a slope of only 0.012, and from 6 mins until the end a slope of 0.011. Although imaging was performed continuously throughout the 60 mins gas protocol, the first 3 mins of data for each block were thus discarded before analysis. During the imaging sequence, inspired and expired levels of both oxygen and carbon dioxide were continuously monitored. This also provided information on the breathing rates of the subjects. Endtidal PCO2 values for a typical subject during the stepped protocol are shown in Figure 4. As can be clearly seen, the PETCO2, which is constant during the normoxic phase, decreases gradually but markedly throughout the experiment, and is even marginally lower than baseline at FiO2 = 0.4. The Journal of Cerebral Blood Flow & Metabolism (2007) 27, 69–75
Cerebral perfusion response to hyperoxia DP Bulte et al 72
Table 1 Values used in the kinetic model to calculate the per cent change in mean grey matter flow with increasing FiO2 FiO2 0.2 0.4 0.6 0.8 1
T1b (ms)
T2b (ms)
DM(t) ( %)
Std. Dev.
Flow (%)
1660 1550 1470 1410 1380
165 164 163 162 161
— 6.57* 11.01* 14.96* 17.69
— 5.79 5.60 2.27 6.07
100 97 96 95 93
The standard deviations are for the mean percentage change in DM(t) across all subjects. *Significant reductions from preceding block, P < 0.05.
Figure 4 End-tidal CO2 trace from a representative subject for the step protocol showing FiO2 levels during each block. The trend is for the trace to stabilise within 2 to 3 mins providing ample data during the plateau region of each block.
Figure 5 Ventilation rate during the entire experiment showing little change with FiO2 and PETCO2. The squares show the mean breath rate for each block, indicating that there was no trend to an increased ventilation rate with increased FiO2.
ventilation rates for the same subject are shown in Figure 5, where it is likewise clear that breath rate does not increase significantly over the course of the experiment. The supposition, therefore, is that if hyperventilation is contributing to the reduction in PETCO2, then tidal volume must be increasing with FiO2 leading to effective hyperventilation. As well as physiologic effects, the increased PO2 in blood changes the NMR relaxation properties (Kennan et al, 1997; Tadamura et al, 1997). The change in T* 2 is useful for BOLD contrast, although the T1 and T2 also change with FiO2 affecting the computation of flow using the QUIPSS2 sequence. Thus, the changes in relaxation times must be included in the assumed model (Buxton et al, 1998; Wong et al, 1998). At 3 T, the T1 of arterial blood under normoxic conditions (FiO2 = 0.21) is 1,660 ms (Lu et al, 2004). The T1 at FiO2 = 1 is approximately 1,380 ms. The T1 of tissue was taken to be 1,470 ms. The T2 of arterial blood was taken to be 165 ms at normoxia, and 161 ms at FiO2 = 1, the T2 of arterial blood being much less affected by Journal of Cerebral Blood Flow & Metabolism (2007) 27, 69–75
hyperoxia than the T1 (D’Othe´e et al, 2003; Noseworthy et al, 1999; Tadamura et al, 1997). For intermediate levels of FiO2, the values of T2 were linearly interpolated. The intermediate values of T1 were interpolated using an exponential function, the reason for this being that there are two competing mechanisms affecting the T1. As the FiO2 rises, the SaO2 increases from approximately 98% to 100%. The consequent decrease in deoxyhaemoglobin content results in a lengthening of T1, whereas the increase in PaO2 causes a shortening of T1. As the change in SaO2 slows down the rise in PaO2, the change in T1 has a slight concave-down trend with rising FiO2. All values used in the calculation of flow are shown in Table 1. When substituted into a MatLab implementation of the Buxton kinetic model (Buxton et al, 1998), the QUIPSS2 experiment is expected to depend strongly on the T1 of arterial blood, but is not as affected by the estimated changes in tissue T1 or blood T2. For a pulsed ASL experiment, the model is given by DM ðtÞ 8 0otoDt < ¼ 0; ¼ 2M0B f ðt DtÞae1=T1b qp ðtÞ; Dtotot þ Dt : ¼ 2M0B f tae1=T1b qp ðtÞ; t þ Dtot where f is the blood flow, T1b is the T1 of arterial blood, and other factors are as described in Wong et al (1998). The change in magnetisation between tag and control images (DM(t)), predicted by the kinetic model, could then be compared with the acquired data allowing the relative changes in perfusion to be calculated independently from changes in relaxation times as FiO2 is altered. The relative change with increasing FiO2 in mean DM measured in all grey matter regions averaged across all subjects is shown in Figure 6. The fitted curve is an exponential with an R2-value of 0.9941. An exponential was chosen as the relationship between FiO2 and CBF, because this is known to be exponential in the hypoxic region. The data are also shown in Table 1, showing that the change in signal between the FiO2 steps 0.2 to 0.4, 0.4 to 0.6, and 0.6 to 0.8 were all statistically significant (P < 0.05). The step from 0.8 to 1.0 was not
Cerebral perfusion response to hyperoxia DP Bulte et al 73
Figure 6 Percentage change in magnetisation signal (DM) with increased fraction of inspired oxygen (exponential trend line), error bars are 71 standard deviation.
Figure 7 Relative change in grey matter perfusion from normoxia calculated from the kinetic model with increasing FiO2 (exponential trend line).
significant; however, with an exponential relationship and with a high standard deviation of the data, this would not be unexpected. ASL techniques have an inherently low signal-to-noise ratio and the flow changes being measured are relatively small compared with those caused by hypoxia or hypercapnia, and so quite large standard deviations are not surprising. Figure 7 shows the corresponding changes in regional grey matter perfusion calculated by the kinetic model at the mean data points shown in Figure 6, taking into account the changes in NMR relaxation times caused by increased PaO2. The exact values used to calculate the flow are shown in Table 1. Once again an exponential curve was fitted to the data; it suggests an exponential relationship between FiO2 and CBF at normobaric pressure that describes the behaviour from extreme hypoxia right up to inspiration of pure oxygen.
Discussion The decrease in average grey matter perfusion at FiO2 = 1.0 found in this study using the QUIPSS2 ASL imaging technique is comparable with that found in other studies using a range of different
measurement techniques (Johnston et al, 2003b). However, the apparent decrease in perfusion at even mild levels of hyperoxia in grey matter regions has not previously been reported. Studies that have measured middle cerebral artery flow velocity (MCAFV) or used magnetic resonance phase-contrast angiography are generally insensitive to changes in CBF beyond the large arteries and thus inappropriate for detecting regional changes in microvascular perfusion within grey matter. There are thus reasonable grounds to have confidence that the reduction in perfusion seen even at the moderate levels of hyperoxia is genuine. It is worth noting that owing to the considerably lower levels of perfusion in white matter relative to grey matter, there may in fact be a similar change in white matter perfusion; however, the signal to noise levels in the current study would not enable such changes to be detected. Thus, it would be possible to have a change in WM perfusion of relatively significant percentages without them reaching statistical significance in this study. The changes in relaxation times of blood and tissue with increased FiO2 are potential confounds of the technique. Failure to accurately account for these changes when calculating perfusion from the Journal of Cerebral Blood Flow & Metabolism (2007) 27, 69–75
Cerebral perfusion response to hyperoxia DP Bulte et al 74
ASL method will lead to gross overestimation of the changes induced. More precise estimates of the relaxation times in arterial blood and parenchyma (in particular, the T1 of arterial blood) at the different levels of FiO2 will lead to more accurate calculations of perfusion changes. The administration of hyperoxic epochs does appear to reduce regional CBF. However, the important issue is whether these reductions are significant enough to preclude the use of hyperoxia as an oxygenation-based intravascular contrast agent. It has been shown in hypercapnia studies that a change in flow of 30% produces approximately a 3% change in the BOLD response as measured by gradient echo EPI (Rostrup et al, 2000). Likewise, the BOLD technique can reliably detect at best B0.5% changes in signal intensity, thus an approximately 5% change in flow is the very smallest change that would affect gradient echo EPI sequences. Further, it has been shown that the signal changes during hyperoxia are dominated by changes in magnetic susceptibility from blood oxygenation rather than by blood volume changes (Kennan et al, 1997), and likewise that hypocapnia does not in itself cause significant changes in regional cerebral blood volume (Rostrup et al, 2005). The change in BOLD signal intensity produced by an FiO2 = 1.0 is between 3 and 3.5% (Losert et al, 2002). Thus at FiO2 = 0.5, for example, the flow changes observed during this study would be too small to directly influence the BOLD-weighted signal, and yet the oxygenation-based change in signal intensity would be sufficient to produce substantial contrast with no changes in regional cerebral blood volume. The experimental design used in this study did not control end-tidal levels of carbon dioxide during the hyperoxic epochs. As a result, there is no indication whether it was the change in PO2 or PCO2 that dominated the induced change in perfusion. This design was to investigate the viability of using very simple delivery systems as found in normal clinical settings to administer the oxygen. It may be that it is in fact the lowered levels of PCO2 that are causing the bulk of the change in perfusion. Further studies using end-tidal forcing methods would be needed to determine this. The slow uptake of oxygen in the bloodstream, the potential effect that uncontrolled breathing can have on the partial pressures of O2 and CO2, and the inherent changes induced in perfusion and relaxation times are all effects that must be accounted for to make supplemental oxygen an effective contrast agent. However, the benefits of using inspired oxygen for image contrast are substantial. The general lack of contraindications, the ready availability in clinical settings, the wash-out times, and the contrast to noise in T2*-weighted scans make oxygen a viable option in a variety of studies. By incorporating the changes in relaxation times caused by the increases in PaO2, hyperoxia can be used for obtaining image contrast in BOLD and ASL Journal of Cerebral Blood Flow & Metabolism (2007) 27, 69–75
sequences and may also be used in measurements of cerebral blood volume.
Acknowledgements We thank Professor Peter Robbins for assistance with interpreting physiologic responses.
References Becker HF, Polo O, Mcnamara SG, Berthon-Jones M, Sullivan CE (1996) Effect of different levels of hyperoxia on breathing in healthy subjects. J Appl Physiol 81:1683–90 Berkowitz BA (1997) Role of dissolved plasma oxygen in hyperoxia-induced contrast. Magn Reson Imag 15: 123–6 Bohr C, Hasselbalch K, Krogh A (1904) ber einen in biologischer Beziehung wichtigen Einfluss, den die Kohlensauerspannung des Blutes auf dessen Sauerstoffbindung u¨bt. Skand Arch Physiol 16:402–12 Buxton RB, Frank LR, Wong EC, Siewert B, Warach S, Edelman RR (1998) A general kinetic model for quantitative perfusion imaging with arterial spin labeling. Magn Reson Med 40:383–96 Christiansen J, Douglas C, Haldane J (1914) The absorption and dissociation of carbon dioxide by human blood. J Physiol 48:244–77 D’Othe´e BJ, Rachmuth G, Munasinghe J, Lang EV (2003) The effect of hyperoxygenation on T1 relaxation time in vitro. Acad Radiol 10:854–60 Floyd TF, Clark JM, Gelfand R, Detre JA, Ratcliffe S, Guvakov D, Lambertsen CJ, Eckenhoff RG (2003) Independent cerebral vasoconstrictive effects of hyperoxia and accompanying arterial hypocapnia at 1 ATA. J Appl Physiol 95:2453–61 Jenkinson M, Bannister P, Brady M, Smith S (2002) Improved optimization for the robust and accurate linear registration and motion correction of brain images. NeuroImage 17:825–41 Jenkinson M, Smith S (2001) A global optimisation method for robust affine registration of brain images. Med Image Anal 5:143–56 Jensen FB (2004) Red blood cell pH, the Bohr effect, and other oxygenation-linked phenomena in blood O2 and CO2 transport. Acta Physiol Scand 182:215–27 Johnston AJ, Steiner LA, Balestreri M, Gupta AK, Menon DK (2003a) Hyperoxia and the cerebral hemodynamic responses to moderate hyperventilation. Acta Anaesthesiol Scand 47:391–6 Johnston AJ, Steiner LA, Gupta AK, Menon DK (2003b) Cerebral oxygen vasoreactivity and cerebral tissue oxygen reactivity. Br J Anaesth 90:774–86 Kennan RP, Scanley BE, Gore JC (1997) Physiologic basis for BOLD MR signal changes due to hypoxia/hyperoxia: separation of blood volume and magnetic susceptibility effects. Magn Res Med 37:953–6 Kolbitsch C, Lorenz IH, Ho¨rmann C, Hinteregger M, Lo¨ckinger A, Moser PL, Kremser C, Schocke M, Felber S, Pfeiffer KP, Benzer A (2002) The influence of hyperoxia on regional cerebral blood flow (rCBF), regional cerebral blood volume (rCBV) and cerebral blood flow velocity in the middle cerebral artery (CBFVMCA) in human volunteers. Magn Reson Imag 20:535–41
Cerebral perfusion response to hyperoxia DP Bulte et al
Losert C, Peller M, Schneider P, Reiser M (2002) Oxygenenhanced MRI of the brain. Magn Reson Med 48:271–7 Lu H, Clingman C, Golay X, Van Zijl PCM (2004) Determining the longitudinal relaxation time (T1) of blood at 3.0 tesla. Magn Reson Med 52:679–82 Noseworthy MD, Kim JK, Stainsby JA, Stanisz GJ, Wright GA (1999) Tracking oxygen effects on MR signal in blood and skeletal muscle during hyperoxia exposure. J Magn Reson Imag 9:814–20 Poulin MJ, Liang P-J, Robbins PA (1998) Fast and slow components of cerebral blood flow response to step decreases in end-tidal PCO2 in humans. J Appl Physiol 85:388–97 Rostrup E, Knudsen GM, Law I, Holm S, Larsson HBW, Paulson OB (2005) The relationship between cerebral blood flow and volume in humans. NeuroImage 24:1–11 Rostrup E, Larsson HB, Toft PB, Garde K, Henriksen O (1995) Signal changes in gradient echo images of human brain induced by hypo- and hyperoxia. NMR Biomed 8:41–7 Rostrup E, Law I, Blinkenberg M, Larsson HBW, Born AP, Holm S, Paulson OB (2000) Regional differences in the CBF and BOLD responses to hypercapnia: a combined PET and fMRI study. NeuroImage 11:87–97 Smith SM, Jenkinson M, Woolrich MW, Beckmann CF, Behrens TEJ, Johansen-Berg H, Bannister PR, De Luca
M, Drobnjak I, Flitney DE, Niazy RK, Saunders J, Vickers J, Zhang Y, De Stefano N, Brady JM, Matthews PM (2004) Advances in functional and structural MR image analysis and implementation as FSL. NeuroImage 23:S208–19 Tadamura E, Hatabu H, Li W, Prasad PV, Edelman RR (1997) Effect of oxygen inhalation on relaxation times in various tissues. J Magn Reson Imag 7:220–5 Watson NA, Beards SC, Altaf N, Kassner A, Jackson A (2000) The effect of hyperoxia on cerebral blood flow: a study in healthy volunteers using magnetic resonance phase-contrast angiography. Eur J Anaesthesiol 17: 152–9 Wong EC, Buxton RB, Frank LR (1998) Quantitative imaging of perfusion using a single subtraction (QUIPSS and QUIPSS II). Magn Reson Med 39:702–8 Woolrich MW, Ripley BD, Brady M, Smith SM (2001) Temporal autocorrelation in univariate linear modeling of FMRI data. NeuroImage 14:1370–86 Worsley KJ, Evans AC, Marrett S, Neelin P (1992) A threedimensional statistical analysis for CBF activation studies in human brain. J Cerebr Blood Flow Metab 12:900–18 Zhang Y, Brady M, Smith S (2001) Segmentation of brain MR images through a hidden Markov random field model and the expectation-maximization algorithm. IEEE Trans Med Imag 20:45–57
75
Journal of Cerebral Blood Flow & Metabolism (2007) 27, 69–75
