Biomedical Engineering Reference
In-Depth Information
one flux quantum to the SQUID; the feedback current sensitivity [
μ
A/
Φ
0
]is
the current at the feedback coil necessary to couple one flux quantum to the
SQUID; and the magnetometer sensitivity [nT/V] is the output voltage at
the FLL circuit when the pickup coil senses a unit magnetic field. We can
also determine the field gradient sensitivity [(nT/cm)/
Φ
0
] as the magnetic
field gradient at the pickup coil necessary to couple one flux quantum to the
SQUID.
The feedback current sensitivity is 49.5
μ
A/
Φ
0
. We adopt 20 k
Ω
as the
feedback resistor. This means 49
.
5
μ
A
20 k
Ω
=0
.
99 V is the output
voltage of the FLL circuit, while the feedback current balances one flux
quantum in the SQUID. The field sensitivity of the sensor is designed to
be 0.68 nT/
Φ
0
. In the sense of field-gradient sensitivity, it corresponds to
(0.68 nT/
Φ
0
)/(5 cm) = 0.136 (nT/cm)/
Φ
0
. The magnetometer sensitivity
and gradiometer sensitivity are calculated by (0.68 nT/
Φ
0
)/(0.99 V/
Φ
0
)=
0.69 nT/V and 0.138 (nT/cm)/V, respectively. These calculated values must
be confirmed by the calibration process described in Sect. 3.2.2 and the re-
sults are 0.59 nT/V and 0.118 (nT/cm)/V. The differences of about 17% are
due to field sensitivity and the feedback resistor. We adopt the calibrated
value for MEG measurement.
×
Noise and Flux Trapping.
It is important and practical to represent the noise
performance for the overall system, not just for the SQUID sensor. SQUID
sensors have low intrinsic noise; for example, 2.0 fT/rtHz in the white noise
region and 4.1 fT/rtHz at 1 Hz. However, additional noise sources which we
must include are the electronic noise of FLL circuits, Johnson noise from the
radiation shield foil in the vacuum chamber of the dewar, and electromagnetic
interference from the DAQ system/host PC/the stimulators. The measured
noise, shown in Fig. 3.54, is 2.3 fT/rtHz in the white noise region, above
50 Hz, and 7.0 fT/rtHz at 1 Hz with the overall system, which includes the
noise due to the SQUID, FLL, dewar, radiation shield, cables, MSR, and an
analog signal processor with a low-pass filter set to 2 kHz cutoff.
It is possible to assume that a signal is applied only to the sensing coil and
not to the reference coil. In such a case, if a signal of 2.7 fT/rtHz is applied to
the sensing coil, we can interpret this signal as being detectable with SNR =
1. Therefore, it is convenient to describe the noise performance in terms of
“field noise [fT/rtHz]”. However, in the actual MEG measurement, the signal
is detected by the sensing coil, and some decayed signal at the reference coil
is subtracted. ECD estimation is carried out by taking this subtraction into
account. Therefore, a notation of “field-gradient noise” is sometimes useful
for the expression of noise performance. The gradiometer baseline in this case
is 5 cm, which leads to another notation of 2
.
3
/
5=0
.
46 (fT/cm)/rtHz in the
white noise region and 1.4 (fT/cm)/rtHz at 1 Hz with overall system.
Incidentally, the range of white noise observed over all channels is spread
around the average value of 4.0 fT/rtHz, which corresponds to 0.8 (fT/cm)/
rtHz, with a standard deviation of 1.3 fT/rtHz.
