Some achievements our pulse diagnosis group have made in HIT:
While paying paid more attention to the quality of pulse waveforms, increasing the pulse database, applying some new analysis methods, we have achieved many results in this interdiscipline. These encouraging results can be divided into the following five respects:
1.
Nonlinear Dynamics Analysis
The
dynamically nonlinear characters and the variability analysis is one of the
important aspects in the pulse image processing. Approximate entropy and Sample
entropy are the two methods on quantifying and comparing the irregularity and
the variability. Now, some algorithms on the Approximate Entropy and Sample
entropy are finished by us. What is more, we have improved these algorithms. At
first, we can apply the known methods to the pulse image research and publish
some papers, and then we can apply our improved methods into this research and publish some papers.
This
respect is a new topic, only some Japanese researchers made some experiments on
the pulse, but they only did some skin-deep research. This is the chance to
apply some nonlinear methods on pulse image analysis in-depth. There are also
some works need to do. And we have finished some algorithms on nonlinear
filter, state phase, recurrent plot, detrended fluctuation analysis (DFA) and
so on. Some results are being analyzed recently.
I
have contacted the experts who proposed those methods such as Approximate
Entropy, Sample entropy, DFA analysis and recurrent plot, and they gave me some
good advice and support on my research.
2.The
Relationship between Pulse Image and Pathology & Physiology
Currently, we have acquired 1500 pulse data related to the sport physiology. These pulse data are acquired from healthy graduate students in HIT. Everyone of these graduate students must be acquired for twice before running and after running in the supine position. We also have 1000 healthy undergraduate students pulse image during sitting; and more than 3000 patients pulse data (nearly 600 are the coronary heart disease pulses).
3.Baseline Drift Removal of Pulse Waveform
Pulse waveform can be easily influenced by many factors such as respiration, body temperature, muscle dithering, body movement and so on. These influence embodies in pulse baseline wander. The whole pulse goes down when exhaling and goes up when inhaling. Holding the breath may make pulse more stable. But these restricts not only make the patient uncomfortable and inconvenient, but also prevent us from acquiring the long period of stable pulse. Thus, we have developed an algorithm for removing the baseline wander. For more accurate analysis, the baseline wander will be further studied.
4.Monitoring of Pulse
Medical informatics has expanded rapidly over the past few years. Continuous electrocardiogram (ECG) monitoring in the patients has become a very common procedure during the last thirty years. The MIT-BIH database also consists some of long-term ECG data. The means of acquiring the pulse information and the performance of pulse sensor are satisfied. However, the research of pulse monitor is reported seldomly because of baseline drift and noise interferences. Having combined with some signal processing technologies, we extracted the baseline drift and attenuated noise interference. Thus, the monitoring of pulse can be realized. This does have the pathological and physiological meaning. During the process of monitoring, we can trace patient physical condition and study disease evolution. At present, we have made some progress in monitoring the pulses and acquired some significant long-term pulse data.
5.Pulse and Its Parameters Variability Analysis
Analyzing
the long-term pulse parameter variability is more significant for studying the
cardiovascular and nerve system. We have analyzed some of the features
extracted from the long-term pulse waveform and studied the chaos
characteristic of Pulse Rate Variability (PRV) of different persons.
The dynamical analysis
of pulse variability gives new insight into researches of cardiovascular system
dynamics. Firstly, long-term pulse variability analysis for the researches on
cardiovascular system was proposed. Secondly, approximate entropy was applied
to analyze three groups of long-term pulse waveform variability and we found
that the pulses approximate entropies of patients with cardiovascular disease
preferred to smaller value and less irregularity. What is more, the pulse
variability of the patient with pacemaker newly implanted was also studied.
Finally, the clinical value of pulse variability for cardiovascular system was
concluded. Future work needs to quantitatively analyze cardiovascular system
behavior on a larger dataset of pulse. Some of nonlinear and chaotic signal
processing methods will also be applied in pulse and its parameter analysis.
For the demonstration of pulse
waveform variability, please see [avi].
6.Pulse diagnosis device designed by us
This first generation system comprised a set of pulse sensors, an amplifier,
and a computer. The sensor, named HMX-4, was made by Shanghai Medical
Instrument Company. It is a strain cantilever beam transducer, which is not the
same as the previous sensors used for studying western medicine. At the center
of the main pulse sensor are placed an array of seven sensors for detecting
pulse width. What we can acquire is not only the pulse pressure waveform, but
also other TCPD information such as position, trend, width, shape, strength,
rhythm and so on. From the position, we can judge whether the pulse image is
floating or sinking; from the trend, we can judge whether the Pulse
Image is feeble or forceful, and so on.
Download our first generation device
[JPG].
Based on the first generation system, we designed the second generation system: Type A and Type B. In Type A, we seek to acquire more information than simply pressure fluctuation in the radial artery. In this device, three photoelectric pulse sensors can exert the three grades of pressures, Fu, Zhong, and Chen. The contact pressure is regulated automatically by a computer system, which is connected to the pulse sensors through a USB interface. This system acquires more information than other systems because it uses a multi-sensor technique (pressure sensor, and photoelectric pulse sensor). The pressure sensor and the photoelectric sensor acquire the contact pressure and pulse wave respectively. The combination of several different kinds of pulse sensors reduces uncertainties and produces more accurate information than a single pulse sensor. The information acquired by these pulse sensors is complementary. For example, by combining information about the artery wall motion with the velocity of the blood flow produces more information about the Pulse Image.
Download our second generation
device Type A [JPG].
Type B device is a multipoint, three-position Pulse Image detection apparatus that can automatically apply contact pressure. The pulse sensor includes a main sensor and several subsidiary sensors. Together, these can imitate the pulse-taking process of a practitioner of TCPD. This device can detect and differentiate between twenty-seven (27) kinds of Pulse Images. The computer controls the mini stepping motors, the precision screw, and bearing to make the sensor probes move vertically in a range of 16mm. This mimics the Chinese practitioner taking the pulse using the contact pressures known as Fu, Zhong, and Chen. According to the theory of TCPD, the three positions, Cun (distal), Guan (middle) and Chi (proximal) and three kinds of contact pressures of Fu, Zhong, Chen consist of ˇ°three body parts and nine pulse-taking sitesˇ±.
Download our second generation
device Type B [JPG].
7.Releated codes
L-Z Complexity analysis [Matching] [Fast algorithm]
Generalized DTW [M file]
Fuzzy neural network [M files]
Pulse Waveform Segmentation [M
files]
Cluster [C++ files]
Members in our group:
2002:
Guo Ning; Xie Zhiguo; Tao Qingfeng;
2003:
Fan Peng; Zhou Bin; Wang Jiaxing; Li Yingzhe;
2004:
Wu Liangzhu; Li Zhengguo; Wei Lirong; Zhang Dongyu; Wang Lu;
2005:
Zhang Wenjie; Geng Bin;
2006:
Jiang Xiaorui; Chen Chaohai; Luo Guicun.
Picture of our
group in HIT:
My recent research output in CUHK:
1.
Wearable array system for the ambulatory location of a solid magnetic 
marker in human body
Figure 1.
Prototype of a wearable array
system for tracking magnetic signal source in gastrointestinal tract of
ambulatory human body.
(a) Front view. (Left) (b) Back view. (Right)
2. Effect of Subject Size on Electromagnetic Radiation
from Source in Human Body Following 2450MHz Radio Frequency Exposure.
For assessment of
compliance with the safety guidelines, the radiation effects and efficiencies
of ingested wireless devices in human body trunk models of five-year-old children,
ten-year-old children and adults are studied using several homogenous models
and a half-wavelength dipole at the frequency of 2450 MHz. The finite-difference time-domain (FDTD) method is used for
analyzing the radiation effects and efficiencies. Results show that radiation
efficiencies decrease with the increment of body size greatly, ranging from -53
dB to -116 dB. Considering the signal efficiency of wireless device and safety
guidelines of radiation exposure, we recommend setting the ingested wireless
device to different power levels for users with different body size such as
five-year-old children, ten-year-old children and adults.
The movie of the SAR distribution [avi]
The movie of virtual female [avi] and its
SAR distribution [avi]
3. Influence of Anesthetic and Dead Pig Bodies on
Ingested Wireless Device
We have studied the influence of anesthetic and dead pig bodies on the ingestible wireless devices (IWDs). Three monopole antennas were designed and two positions in the intestine of an adult female pig at the weight of 80 kg were selected. The results demonstrated that the operation frequencies of the three antennas increased 0.1-0.66 GHz after the euthanasia of the pig, which further validated the decrease of the dielectric properties of the animal tissues after death. These variations need to be taken into account in electromagnetic modeling before the design of ingested antennas.
4. Radiation Characteristics of Ingestible Wireless
Devices in Human Intestine Following 430, 800, 1200 and 2400 MHz
In
order to assess the compliance of IWD within the safety guidelines, the
biological effects and the radiation efficiency of the IWD in two realistic
human body models were studied using FDTD method. Simulations were carried out
at 21 scenarios where the IWD was placed at seven positions in two human body
models with three orientations. Specific Absorption Rate (SAR), temperature rise,
near fields and far fields were analyzed in
these scenarios at the operation frequencies from 430 MHz to 2.4 GHz.
Results showed that the radiation intensity outside the human body decreased
with the increment of operation frequency. Maximum radiation occurred at the
operation frequency of 430 MHz. Furthermore, the orientation and position of the IWD, which had
maximum radiation, were frequency dependent. In our simulation, when the IWD
was placed at the anterior and posterior positions of the small intestine, the
intensity of electric field outside human body was maximum and minimum,
respectively. The radiation efficiency of the IWD was more affected by its
position in the gastrointestinal (GI) tract than its orientation. The electric
field polarization outside human body was similar to that of the IWD. Having
analyzed the worst cases of the radiation, we found that the frequency less
than 1.2 GHz could be used for stable communication in the IWD. As far
as the compliance of safety was concerned, the maxima of averaged-1g and
averaged-10g SARs could reach 3.71 and 1.37 W/kg at input power of 25 mW.
The IWD was safe to be used
in the capsule device at the input power less than 36 mW and 11 mW
according to the European and IEEE safety standards, respectively. For
the ingested and implanted wireless devices, the high local SAR peak needs to
be considered more carefully in comparison with the wireless devices used
outside human body due to higher local energy deposition of the radio frequency
source embedded in human body.
5. Effects of Dielectric Parameters of Human Body on
Radiation Characteristics of Ingestible Wireless Device at Operating Frequency
of 430 MHz
In order to assess the compliance of IWD within safety
guidelines, the SAR, near fields and far fields of IWD in two realistic human
body models whose all tissues dielectric values were increased from the original
by 5, 10 and 20% were studied using FDTD
method. The radiation characteristics of the IWD in the human body models with
increased dielectric values were compared to those of the IWD in models whose
dielectric values were unchanged. Simulations were carried out at 10 scenarios
where the IWD was placed at center position of abdomens in two human body
models at the operation frequency of 430 MHz. Results showed that variation of radiation intensity near the surface of
abdomen was around 1 dB within
20% variation of dielectric values.
Electric fields in the anterior of the human body models were higher than those
in the posterior for all scenarios. SAR values increased with the increment of
conductivities of human body tissues and usually decreased with the increment
of relative permittivities of human body tissues. An increment of up to 20% in
conductivities and relative permittivities alone or simultaneously always
caused a SAR variation of less than 10%. As far as the compliance of safety was
concerned, the maxima of averaged-1g and averaged-10g SARs could reach 2.21 and
0.63 W/kg at the input power of 25 mW. The
IWD was safe to be used at the input power less than 79 mW and 22 mW
according to ICNIRP and IEEE safety standards, respectively. For
IWDs, the high-local radio frequency energy deposition in human body needs to
be considered more carefully in comparison with the wireless devices used
outside human body.
6. Applicability of Homogeneous Human Trunk Phantom in
Estimating the Radiation Characteristics of Body-Worn Devices
In
this paper, the radiation characteristics with respect to the suitability of
using homogeneous phantom for testing the compliance of radiation frequency
devices are assessed. The Finite-Difference Time-Domain (FDTD) method is
applied to analyze the variations of a 900 MHz half-wavelength dipole
antenna biological effects and link performance depending on distance between
antenna and human body model. The distance between the surface of the model and
the outside exposure source is changed from 25 mm to 1 mm within
the range of wavelength/(2PI). The distributions of the specific
absorption rates (SARs) and the electric fields for various vertical slices of
a simplified homogeneous phantom and three anatomical human body trunk models
are calculated, respectively. The legs and head have little influence on the
radiation characteristics of body-worn, ingestible or implantable wireless
devices. The results demonstrate that a homogenous representation of human body
is suited for assessing the averaged SARs in human body and confirm that the
local energy absorption details and communication link performance need to be
analyzed by using the anatomical models or by combining with the worst-case
consideration.
7.Releated codes
FDTD1D [C++]
FDTD2D [C++]
FDTD3D [C++]
Picture of our group in CUHK:
J