Abstract :
[en] Detailed quantitative information about metabolic processes plays a crucial role in the
potential cure and for treatment of many diseases such as Alzheimer’s disease or brain
tumours. In the last decades, radioactive tracers such as 15O have been used to quantify
CMRO2 with PET imaging and this is regarded as the gold standard. However, such methods
are complicated and expensive as a consequence of the short half-life (2 min) of 15O
and inherently include radiation exposure and invasive measurements such as blood probes
to probe cerebral blood flow (CBF). Fick’s principle of arteriovenous oxygen difference [1]
connects CMRO2 and CBF via the measure of oxygen extraction fraction (OEF).
The main goal of this work is to achieve non-invasive measures of OEF based on magnetic
resonance imaging (MRI) to quantify CMRO2 allowing straightforward and comfortable
patient handling. MRI enables studies of large cohorts of healthy volunteers due to noninvasive
measurements and a lack of radioactivity. This can be achieved first by quantitative
relaxation time mapping of the transverse relaxation time (T2) of venous blood only in proton
(1H) MRI or by a measurement following inhalation of 17O gas and recording the signal
curve of directly detected 17O signal. Unfortunately, the most abundant isotope of oxygen
(16O) has a zero spin system, and cannot be detected with NMR experiments. In contrast,
17O, a stable isotope with a half-integer spin (I=5/2), can be detected by MR. Fortuitously,
however, in MRI it is only visible in the form of metabolically generated H17
2 O and not as a
gas. The low natural abundance of 17O, of only 0.037% (of the oxygen atoms) and the low
NMR sensitivity (2.9% that of 1H) gives rise to the need for ultra-high-field MRI to reach
a significant SNR per unit time.
Natural abundance images of a healthy male volunteer were acquired in vivo after having
gained written consent within a clinical trial of a 9.4T MRI system (Siemens AG, Erlangen,
Germany) [2, 3]. These natural abundance images, which reflect the 17O bound to
protons as H17
2 O and thus, the amount of water, are compared to 1H-based quantitative
water content imaging. For further studies, the voxelwise knowledge of the quantitative
water content is necessary to quantify CMRO2 based on the 17O signal behaviour.
To achieve that, methods which were originally used on 1.5T scanners had to be adapted
for the use at higher field strengths to overcome RF field inhomogeneities [4–11].
New correction methods were developed based on a well known correlation between tissue
T1 and proton density (PD) to estimate the receive bias field properly. These methods
were tested for quantitative water content determination. Averaged results in grey (GM)
and white matter (WM) respectively of 10 healthy volunteers are H2OWM=70.3 1.4 %,
H2OGM=84.7 1.5 %„ T1WM=918 24 ms and T1GM=1509 14 ms.
Further, 1H-based imaging methods called QUIXOTIC [12–14] and TRUST [15] appeared
in the literature. These methods are based on changes of the proton transverse relaxation
rate T2 with different oxygen saturation levels. Quantitative values of venous blood T2
were acquired using a so-called T2prep module or a multi-echo spin echo readout. While
the first method suffers from long acquisition times the latter one from large echo-spacing
of the spin echoes and stimulated echo effects. Both disadvantages were overcome using an
adiabatic multi-shot multi-echo spin echo sequence, which does not suffer from stimulated
echo effects and due to the multi-shot capabilities, the echo-spacing is reduced [16]. Mean
values in GM of four healthy volunteers are found to be venous oxygenation Yv=0.61 0.03,
T2=54 4 ms, CMRO2=174 13 mol/100g min and CBF=53 3 ml/100g min.
