, Yu Su

Abstract:
Most of people studying and working in a sitting posture for long, spine disease became one of the
most important occupational health problems. The patients with condition worsen have to suffer harmful
CT scans or even painful serial lumbar puncture. We innovatively developed a biomedical optics based
detector for spine health monitoring and particularly improved the circuit design for feeble signal detection.
A series of experiments were performed to fully test its reliabilities, including scalability, noise and
crosstalk, sensitivity, stability etc. The results showed that this device was very reliable, sensitive, and
stabile, suggested the promising clinical use of our novel device in spine/spinal marrow health monitoring
and diagnosis.

Introduction:
For those people who suffering from spine diseases, there are two main clinical methods
for monitoring and diagnosis. One of them is medical biopsy. A special spine needle is used to
insert into the spine, then tissue fluid is extracted from the spine and analyzed. Accordingly, spine
disease is detected by the biopsy of the tissue fluid directly. Apparently this method is invasive
and it is extremely brutal and of unbearable pain for poor patients. Moreover, most patients with
serious spine diseases, such as multiple myeloma (MM), have to experience such painful test
many times after treatments for therapeutic effect evaluation (etc., to test if there is any cancer
cell left in the spine), until they are completely cured or died. The other method is imaging
apparatus, such as X-ray, computation tomography, magnetic resonance imaging (MRI). They
are noninvasive and patients can free from pain. However, these techniques are either ionizing
or contrast agent intake, of big instrumentation, which are expensive and non-portable. Some of
them are radioactive which are not really safe for multiple test, such as X-ray and MRI.
Meanwhile, the non-portable design makes these techniques impossible in continuous
monitoring or bed\home care.
In this study, we designed a novel detector for monitoring spine disease noninvasively and
nonionizing. Its principle is based on near infrared spectrum technology, which has been recently
applied to medical diagnosis by detecting hemodynamic parameters, such as breast imager,
 functional activity imager, monitors for muscle function, shock, and thrombosis [6]. This
technology is on the basis that the main substance water shows low absorption of near-infrared
light and that hemoglobin’s are the main absorbers of near-infrared light and are of different
absorption spectrum. With multiple well-selected wavelengths of near-infrared light, and the
light propagation theory, we were able to quantify different hemoglobin concentrations in the
measured spine site, which is key info for monitoring and diagnosing the health situation of the
spine and spinal marrow.
However, the measured light intensity variation measured out of the spine tissue is quite weak
compared to the background light noise and signal from other type of tissue without spine region.
These signals are also easily interfered with crosstalk from other channels/measured site or other
chromosphere, power-line interference, A/D conversion error etc. We improved the circuit design
to extract the weak signal and reduce the noise coming from photosensitive sensors in the whole
detector design. Finally, we implemented a sequence of performance tests to examine the
reliabilities, such as scalability, sensitivity, specificity, signal-to-noise ratio (SNR), stability, and
consistency, the data of which verified the reliability of our detector.

1. Method and experimental design

1.1. Device design

Our detector contains 7 modules shown in the Fig. 1, which are power, probe, A/D converter,
master control, probe driven, data transmission and storage. The power provides +5 V and +3.3 V
DC voltage for other modules to work well.

Fig. Structure diagram of detector

 Probes have a three-wavelength LED [12] and eight light-sensitive detectors. The three-
wavelength LED can respectively emit three near-infrared wavelength light (735 nm,805 nm, 850
nm). The light-sensitive detector is OPT101 [13] which can change different light signals into
electronic signals and it is with high responsibility 0.45A/W (650 nm) which make it possible for
us to detect weak signals. A/D converter with the resolution of 16 bits and speed of 250 ksps
changes the analog signals into digital signals precisely. Master control is an ARM chip which is
used to control the light source, detectors, data collection etc. Computer is used to display the result
by the upper computer.
When master control illumines one wavelength of the LED, the detectors are chosen sequentially
to work and receive the light intensity out of the tissue. Then repeat it with other wavelength. At
the same time, the A/D converter is changing the analog signals achieved by detectors into digital
signals. The master control sends the data to computer.

Reliability analysis

In the above-mentioned data collection process, because the signals collected out of the tissue
are very weak and many of the channel signals need to be obtained orderly, it is inevitable to
involve many noise into the signals. There are three main noises, crosstalk, power-line interference,
interference of A/D scanning module. To make our detector of good sensitivity, specificity, high
SNR and stability, some important measurements were taken to weaken those interference.

Crosstalk

Crosstalk is a signal in the transmission channel on the transmission and is caused by the wires
which are parallel and close to each other. One of these wires will introduce electromagnetic
interference [14] to the others. Excessive crosstalk may cause false triggering of the circuit, causing
the system to fail to function properly. It mostly occurs in multi-core cable, bundled connector,
PCB parallel wires. The intensity of crosstalk [15] is decided by the circuit capacitance, mutual
inductance and circuit impedance. So crosstalk is usually classified into two kinds of noise:
capacitive coupling and inductive coupling [16]. The voltage change of aggressor will produce
induced current in victim in spite of electromagnetic interaction. The intensity of crosstalk depends
on a series of factors such as the distance between single lines, the length of wire, the structure of
signal lines etc. According to the previous research, the shorter of the distance between signal
wires, and the longer of the wire length, the greater noise will be introduced into the other signals.
So in this study, the class twisted-pair line [17] is used to reduce the noise caused by crosstalk. One
of the twist-pair is connected to ground and the other is the acquisition signal wire. In this way, we
can greatly reduce the coupling capacitance and the coupling inductance by isolating the signal
wires.
 Power-line interference

The other main noise is power-line interference [18]. The signals we collect are weak, it is easy
to be interfered by the power-line interference which is around in the air and interfere the electronic
device.
Power-line interference is caused by the mains voltage which is AC voltage of the frequency of
50 Hz. Therefore, the noise caused by power-line interference has the same frequency of 50 Hz.
To avoid this interference of our signals, a 50 Hz trap filter is used into our circuit before the A/D
converter.

Interference of A/D scanning module

It has been introduced before that our detector receives many channel signals each period.
Each channel signal is collected by scanning mode which let each channel work orderly. In this
mode, with the scanning frequency getting much higher, the sampling capacitor Cs of A/D
converter, which is connected to ground in the A/D scanning module, will not have enough time
to charge and discharge. Therefore, the successive channel signals will interfere with each other.
Therefore only to increase the sampling time or reduce the channels' resistances can it reduce the
sampling error voltage ΔU. So there are two methods to reduce this noise. One is to decrease the
value of input resistance. For this purpose, a voltage follower is designed between signals and
A/D converter. In this way, the equivalent resistance of signal source significantly decreases. The
other way is to increase the sampling time. However, the sample frequency will be reduced which
is not good. Therefore, in our device, voltage follower is introduced before A/D converter.

Reliability experiment design

To evaluate the reliability of our instrument, we conducted sensitivity experiment and

interference assessment test. In these experiments, we analyzed our instrument's sensitivity,
channel consistency, stability in detail.

Sensitivity test

In this experiment, we selected a polyethylene container as the experiment container. 500 ml

water was added into the container first. Then 0.5 ml of ink (diluted 1000 times) was injected
into the container each time, so the concentration of the solution changed over time. At the same
time, our instrument was detecting this high scattering solution. Because the different
concentration of the solution had different scattering and absorption coefficients [19,20], the
detected intensity of our detector changed together with the concentration of the solution.
Therefore, this experiment result can respect the sensitivity of our instrument. Meanwhile, our
instrument has 8 channels. When comparing all the channels' data of different wavelengths, we
found that the channel's consistency can be verified.
 Interference assessment test

Interference assessment test was designed to test the antiinterference performance. In this
experiment our detector was used to detect a uniform material which had stable scattering and
absorption coefficients. While a stable light source illuminated into this material, and the
background intensity was kept unchanged and the detector and material were left untouched, the
photons out of the material should be stable. So by detecting the result of the out photons, the
reliability and anti-interference of our detector can be certified.

Scalability test

An in-vivo physiological test was designed by this work to evaluate and test the clinical
performance of our detector. To achieve a reality effect, we choose an arm as the detecting part.
The probe was tied tightly on the subjects' forearm muscles with no sunlight detected by our
detector and a bandage was tied on the upper arm. The bandage was used to block blood flow to
let the forearm be lack of oxygen. Therefore, the oxygenated hemoglobin concentration and
deoxy-hemoglobin concentration of forearm changed in this process. Whether our instrument
can detected those change or not can easily be used to evaluate the performance.

Fig.. Consistency and sensitivity experiment result

 Results

The consistency and sensitivity result is shown.. Each color curve represents a channel. As we
can see from the figure, the value of each channel fluctuates immediately while improving the
concentration of the solution every time and then it keeps to a steady value. Those results are
consistent that with improving the concentration, more photons will be scattered and absorbed,
and thus the photons out of the solution decrease. After a few seconds, the solution comes into
uniform mixing, the detected photons will keep stable. At the same time, the higher the
concentration is, the smaller the decrease will be. This because the base of the last data is already
very small. This result powerfully proves the high sensitivity of our detector to monitor the
change of solution. Meanwhile, we processed the data of each channel which is shown in the. In
the table, it shows each channel's amplitude of variation in detail.

At the same time, shows the change of the different wavelengths vary with the channel. The
six color parts differ in wavelength and source-detector separations, there are three wavelengths
and two source-detector separations (long separation 3.2 cm and short separation 2.5 cm,
channels is short separation, 5,6,7,8 represent long separation. For the short-separation channels,
the difference between the biggest value and smallest value of 735 nm is 0.0832 and the variance
is 0.0974. Similarly, the difference and variance of 805 nm are 0.0828 and 0.1138. For 850 nm,
the difference and variance are 0.074 and 0.0897. For long-distanced channels, the differences
of 735 nm, 805 nm. 850 nm are 0.0718, 0.0819 and 0.0826 and the variances are 0.0750, 0.1474
and 0.1312 accordingly. The largest difference and variance is 0.0832 and 0.1474. So, it's safe
to say that deviation of

Fig. 3. Curve of channels' amplitude variation.

the same wavelength and distance is comparably minute and our detector has a good channel
consistency.
As for the interference assessment test, we recorded the voltage values continuously and the
result is shown in the Fig. 4. The detail parameters are listed in the Table 2. As you can see in the
figure, the lines are almost straight and each color represents a channel. As the sampling numbers
grows, the measured voltage value nearly remains the same. We dare to say that the data acquired
by our detector is very stable because the biggest difference between maximum and minimum each
channel is only 0.05v which is very small compared with the detected value. Meanwhile, the
standard deviation of the data is tiny with 10−3. So we have every reason to believe that the small
fluctuations can be ignored. This result forcefully proves that our instrument is of little noise and
great reliability.
 In Fig. 5, the curve presents the relative variation of oxy-hemoglobin concentration, deoxy-
hemoglobin concentration. In the figure, with tightening the bandage at about 30 s, the deoxy-
hemoglobin concentration increases slowly. At the same time those of oxygenated hemoglobin
drops slowly. About 2 min later, loosening the bandage, the concentration of deoxy-hemoglobin
dropt quickly and those of oxygenated hemoglobin increased fast. This experimental result is very
consistent with the theory that when tightening the bandage the blood cannot reach forearm, so the
oxyhemoglobin is decreasing and deoxy-hemoglobin increasing with the physiological
consumption. This in-vivo physiological test powerfully verifies the excellent performance of our
instrument.
With the analysis of noise which may affect the reliability of our detector, a sequence of
measurements including class twisted-pair line, trap filter and voltage follower were taken in our
instrument to avoid those noise. Then we took two experiments to fully test the reliability of our
detector. According to the above results and analysis, it clearly shows that our detector is reliable.

Conclusion

In this work, we designed a novel device of spine disease monitoring based on near-infrared
spectroscopy technology. Three approaches were introduced into the circuit design of the detector
to improve the reliability and stability by reducing noise. Also, a serials of experiments were
implemented to evaluate the performance of this novel device.

Fig. 4. Interference assessment test of system.

Parameters of interference assessment test.

Channel Maximum(v) Minimum(v) Average(v) Standard
deviation
1 2.68 2.66 2.67 0.0025
2 2.31 2.26 2.28 0.0068
3 2.1 2.07 2.08 0.0056
4 1.98 1.94 1.95 0.0045
 Fig. 5. Result of simulation test.

The results of the first test are consistent with the theoretical prediction that the photons out of the
solution will decrease linearly with the increase in concentration, due to number of photons will
be scattered and absorbed, which verifies the scalability, reliabilities, and sensitivity of our
instrument. The other experiment was aimed at testing the reliability of anti-interference and
deviations. The results showed to be negligible, which verified the high reliability of our detector.
Taken together, the results of above reliabilities tests well demonstrated that our novel medical
device exhibited excellent performances and reliable in measurement and monitoring. At this stage,
the reliability analysis proved the design of spine detector and capability of this innovative medical
device in application in noninvasive, reliable monitoring of spine or spinal marrow health
monitoring and management. We are collaborating with the affiliated hospital of our university to
apply and test this device on noninvasive evaluation of multiple myeloma and low back pain
monitoring/diagnosis.