Phys.
Med. Bioi.,
1987, Vol. 32,
No
10,
13451353.
Printed
in
the
UK
Scientijic note
A
SAFE
(saturate after the echo) sequence to eliminate an
artefact in
measurement
of
T2
in
partially
saturated
NMR
systems
Paul
S
Tofts
and
G
Johnson
Institute of Neurology,
Queen
Square,
London
WClN 3BG,
Received
18
December
1986,
in
final form
14
April
1987
1. Introduction
'H
relaxation
times
tissues
are
influenced
by the
physicochemical
environment
of the
protons.
There
is, therefore,
potential
for
tissue
characterisation
times.
However,
most
pathological
show
elevated
(Beal
et al
1984)
simple analyses
which
assume
monoexponential
decay are
often
unable
to
demonstrate
any
differences
between
various
pathological
tissues. Many tissues,
however,
display
multiexponential
T2
(Barnes
et
a1
1986),
accurate
determination
of multiexponential
behaviours
may give the
specificity necessary for
characterisation.
To
study
multiexponential
transverse
magnetisation
decay with the
optimum sensi
tivity, it is important
to follow the decay up
to long
echo times
TE,
until
the signal
has
disappeared
into
noise
(i.e. TE=
5TJ.
The naive equation
often used
(Mills
et
1984)
to describe
image
intensity
in
a
singleecho
experiment
is (for
a single
T2
component
perfect
radiofrequency
(
RF) pulses)
IN
=
GN
exp(

TE/
T2)[
1


TR/
T1)]
(1)
where G is a
constant
describing
instrument
gain,
N
is the
spin density, TE is the
echo
time
TR
is the
repetition time. The
slope of
ln(Z,)
against
TE
gives
T2.
However,
fuller
analysis
reveals
that
the
exact
value
of the
signal
intensity is (Young
et
al
1982)
~E=GN~X~(TE/T~){~~~~~[(TRTE/~)/T']+~~~(TR/
T
(2)
in
which
case the
slope of
h(IE)
against TE is dependent
on
Tl
,
T2,
TE and TR, and
T2
cannot
be
determined
without making
a separate
determination of
and recourse
computer fitting techniques.
could
be left as
a free
parameter
the fit, but
this
increase
number of degrees of freedom
would
reduce
precision with which
T2
couldbedetermined.Measurements
of
T2
are usually made usingequation
(1).
The relative
error
in
intensity,
eI,
caused by assuming
data follow
when
in
fact
they are
described by
(2),
is given by
00319155/87/101345+09$02.50
@
1987
IOP Publishing Ltd
1345