Team:BIT/Fluorescence Detection
Click on other areas to close the home directory
Fluorescence Detection
Abstract
[ Objective ] The fluorescence detection method based on the Crispr/Cas12
detection system was constructed in the biosensing module, and the miRNA content in the human
body was coupled with the intensity of the fluorescence signal. In addition, the modeling module
showed that the comprehensive risk score could be calculated by jointly detecting a variety of
miRNA content, which could effectively improve the detection accuracy. Therefore, the main
function of the fluorescence detection module in Hardware is to detect the fluorescence signal
strength in the SlipChip to realize the quantitative detection of miRNA signals. A
multi-channel, high detection accuracy, high repeatability and high linearity fluorescence
detection module for the SlipChip is constructed.
[ Methods ] The fluorescence detection module consists of two modules : confocal
light path and push rod scanning. The confocal light path is composed of a two-way color mirror,
a narrow band filter and a convergent lens. The light source is a blue LED driven by a constant
current source. The sensor adopts a high sensitivity photoelectric sensor. At the same time, the
fluorescence intensity is converted to voltage value by the Butterworth low-pass filter combined
with AD sampling chip. The whole system uses MCU as the main control, and configures Bluetooth
module to realize remote control and data acquisition of WeChat small program on the mobile
phone.
[ Results ] The fluorescence detection module had good repeatability for the
detection of fluorescent dyes. The coefficient of variation was not more than 3 %. The detection
precision was high and the fluorescence interference between different channels was small. At
the same time, the linearity of fluorescent dyes was good. The experimental results show that
for the fluorescence signal generated by Crispr/Cas12 system, the fluorescence detection
module developed by our project can achieve high precision and multi-channel detection. The
results are more accurate in the detection range, and different channels are less disturbed.
Fig1.Fluorescence detection module design
Fig2.Fluorescence detection module physical map
Fig3. Diagram of fluorescence detection module
Fig4. Excited effect diagram of fluorescence
Optical path design
The fluorescein is irradiated by the incident light ( i. e., excitation light ),
and according to the energy level transition principle, it can emit fluorescence with longer
wavelength than the absorption light. The fluorescence intensity is proportional to the incident
light intensity, quantum efficiency and sample concentration. To achieve high sensitivity
fluorescence detection, the key is to reduce the influence of background light, especially the
excitation light background. Therefore, the optical path structure design of the detection
system is very important. At present, there are orthogonal, non- confocal, confocal and parallel
types reported (Table1).
Table1. Comparison of different optical path structures
Due to the good separation effect of the confocal optical path structure on
fluorescence and excitation light, and the need for miniaturization of the equipment, we choose
the confocal optical path structure and consider the luminescence principle based on Crispr/
Cas12 constructed by the biological module and the corresponding optical device selected by the
FAM fluorescent dye.
In this fluorescence structure, since the objective is also a concentrator, the
focusing of the excitation light is avoided, and the number of optical elements is reduced. The
intensity of the fluorescence signal increases with the increase of the numerical aperture of
the objective. The excited light and reflected fluorescence propagate in the opposite direction
in the objective lens, thereby reducing crosstalk. In the confocal fluorescence structure, the
illumination surface and most of the fluorescence generation surface are on the same side of the
sample, which can effectively reduce the loss caused by fluorescence through the sample and
improve the signal to noise ratio.
Fig5. Design of fluorescence detection optical path module
Fig6. Design of fluorescence detection optical path fluorescence
Optical Component Design
Selection of excitation source
Since the fluorescence Stokes shift is small ( generally 40-60 nm ), a narrow-band
filter is needed to effectively filter the crosstalk between the light source and the result.
This project is based on the consideration of the passband characteristics and transmittance of
the filter and the excitation wavelength ( 488nm-532nm ) of the FAM fluorescent material used in
the biological module.
Table 2. Comparison of typical light sources
The performance comparison of various types of excitation sources is shown in the
table. Compared with LED, other excitation sources have the disadvantages of high price, short
life and high thermal radiation, which cannot meet the needs of low price and high reliability.
LED has the advantages of long life, cold light source, low radiation and low price. Therefore,
LED is selected as the excitation source in this paper.
Fig7. LED excitation source
The specific design of LED should consider the following aspects.
( 1 ) Spectral power distribution characteristics, that is, according to the
spectral coverage of the excitation light required to determine the type of light source. In
practical applications, fluorescence is usually very weak. In order to obtain sufficient
signal-to-noise ratio, the fluorescence generation efficiency should be improved as much as
possible. The excitation light wavelength is one of the many factors affecting the fluorescence
intensity. LED light wavelength is optional, and blue LED can be selected as our excitation.
( 2 ) Power and conversion efficiency of light source. Since the fluorescence
intensity is proportional to the excitation intensity, it is required that the light source must
have sufficient luminous intensity to fully excite the measured object. However, the increase of
light source intensity will produce strong thermal effect and scattered light. The increase of
temperature weakens the fluorescence intensity. After testing, we choose 3W LED as excitation
source.
( 3 ) Light source stability. The reflected light produced by the fluorescent
material after irradiation is often weak, and the unstable excitation light will directly lead
to the instability of the instrument test results. LED is a special diode, belonging to the
current-type nonlinear element, with steep forward volt-ampere characteristics. The small
fluctuation of voltage may lead to the doubling of the current flowing through the LED, and the
temperature drift will also lead to a large change in the brightness of the LED. Therefore, our
project uses aluminum substrate to connect LED for effective heat dissipation, and uses a
constant current source for its power supply.
Selection of Filters
The role of the excitation filter is to enable the excitation light to pass
effectively, completely cut off the non-working band light into the optical system. The function
of the emission filter is to let the light emitted by the fluorescent substance pass through the
band at the imaging end, blocking all unexpected light outside the band, especially the
excitation light. Therefore, the filter should be selected according to the fluorescence
spectrum. The emission spectra of several commonly used fluorescent substances are shown in the
figure.
Fig8. Emission spectra of several common fluorescent substances
Since the fluorescence Stokes shift is small ( generally 40-60 nm ), a narrow-band
filter is needed to effectively filter the crosstalk between the light source and the result.
This project is based on the consideration of the passband characteristics and transmittance of
the filter and the excitation wavelength ( 488nm-532nm ) of the FAM fluorescent material used in
the biological part.
Fig9. 488m central wavelength transmittance spectrum curve.
Fig10. 532m central wavelength transmittance spectrum curve.
Two-way Colour Mirror Selection
Two-dimensional chromoscope is commonly used in enzyme labeling, fluorescence
microscopy and other biomedical testing instruments. Excited light is reflected into the
objective through a two-way mirror, and the emitted fluorescence ( longer than the wavelength of
the excitation light ) enters the optical sensor through the two-way mirror. The excitation
light reflected backward or scattered by the sample will be reflected by the dichroism again to
prevent it from entering the optical sensor to form stray signals.
According to the FAM fluorescence dye spectrum and the filter selected by our
project, we choose a dichroic mirror with a wavelength of 490 nm, and the spectral curve is
shown below.
Fig11. Spectral curve of two-dimensional chromoscope
TConsidering the need of instrument miniaturization, we choose the appropriate lens
to gather light and assemble the above optical devices with sleeve.
Device selection and assembly
Table3. Main selections
After assembling the optical devices, as shown in the figure.
Fig12. Optical module design appearance
Fig13. Optical module design perspective
Fig14. Assembly finished product of optical devices
Design of photoelectric sensor and signal processing circuit
Photoelectric sensor is a device that converts light signal into electrical signal.
Its working principle is based on the photoelectric effect : when light irradiates on certain
substances, the electron of the material absorbs the energy of the photon and the corresponding
electrical effect occurs. The commonly used photoelectric devices include photomultiplier tube,
CCD, photodiode, photocell, etc.
Table4. Comparison of various types of detectors
In practical applications, we choose optical diode sensors optimized for biomedical
applications. The sensor needs low power consumption during operation,
Fig15. photodiode application circuit
and supports power cycle to optimize battery life in portable applications. It is an ideal
choice to measure the signal with the highest fidelity under low light conditions to provide the
granular noise limitation performance.
At the same time, the operational amplifier chip is used to amplify and regulate
the signal, and the measured voltage is amplified to the appropriate range, which increases the
detection accuracy and outputs to the AD sampling chip.
The AD sampling chip converts the analog signal into digital signal, discretely
samples the continuous signal, and sends the obtained signal to the microcontroller for data
calculation and summary.
Fig16. Operational amplifier circuit diagram.
Fig17. Connection mode of AD conversion chip
The above circuit schematic drawing and PCB design, made of circuit board as
follows :
Fig18. Actual production circuit board
Bluetooth Control
In the Bluetooth control module, the Bluetooth module is connected by the minimum
system of microcontroller, and the sliding push rod is controlled to scan the SlipChip. Clicking
the button on the WeChat miniprogram can control the start and stop sliding，Specific use as
shown in the video.
Fig19. Bluetooth control WeChat applet
Video1. Bluetooth control video
Multi-channel scanning detection design
The self-sliding push rod is used to connect the optical component through the
connector. The self-sliding push rod is controlled by the minimum system of microcontroller to
make the optical component move along the direction of the main reaction chamber on the
SlipChip. The optical component continuously emits blue LED light to illuminate the FAM
fluorescent dye, and the fluorescence excited by the dye is transmitted to the position of the
sensor through the optical path. The optical sensor combined with the AD sampling circuit
converts the continuous fluorescence measurement value into a series of digital fluorescence
signals.
Fig20. Structure of optical module
Figure21. Scanning path
Video2.Design Scanning video
Video3. Real Scanning video
For the measured original fluorescence signal, because it may have burrs
and spikes, Kalman filtering method is used to smooth the signal. Kalman filter is an
algorithm that uses the state equation of linear system to optimally estimate the system
state through the input and output observation data of the system. The main process of
the algorithm is shown in the figure.
After Kalman filtering of the fluorescence signal, the peak detection
algorithm is used to extract the peaks of six different positions. The peak value is the
intensity of the fluorescence signal in the six chambers of the SlipChip. The specific
results of this module are shown later.
Fig22. Flow chart of Kalman filtering algorithm
Performance test of fluorescence detection module
Precision Test of Fluorescence Intensity Teste
Six detection chambers were selected within the measurement range of the
instrument. The calibration fluorescent dye solution of each chamber was prepared for detection,
and each calibration dye with concentrations of 0.1,1,10,100 μmol / L was detected once. The
data of the target channel were collected by the optical system. The average M and standard
deviation SD of the calibration dye measurement results of each concentration were calculated.
According to the formula, the coefficient of variation CV was not greater than 5 %.
CV = SD / M × 100 %
CV is coefficient of variation, SD is standard deviation, M is the average value of
the measurement results.
Fig23. Sodium fluorescein solutions of different concentrations.
Fig24. Average fluorescence intensity
Fluorescence Interference of Different Channels and Scanning Test
The six chambers of the SlipChip were placed with different concentrations of
fluorescent reagents of 0.2, 0.5, 0.8 μmol / L ) as shown in the table. The optical system was
controlled by a self-sliding push rod to obtain the fluorescence intensity value, and the
fluorescence value curve was plotted. The results are shown in the following figure.
It can be seen from the results that the fluorescence signal intensity between the
hole 1 and 2, 3 and 4, 5 and 6 is basically similar, indicating that the crosstalk problem can
be controlled to a certain extent. The fluorescence value of hole 1, 2 and 3, 4 and 5, 6 can be
clearly distinguished, indicating that the instrument has the ability to detect different
concentrations of fluorescent dyes.
Table5. Table of chamber number corresponding to fluorescein concentration
Fig25. schematic diagram of fluorescein placement.
Fig26. original data diagram of fluorescence detection signal
The data curves before and after Kalman filtering are plotted respectively as shown
in the following figure. The effect of the filtering algorithm on the fluorescence detection
data can be clearly seen, and the peak detection is performed on the filtered algorithm. The
results are shown in the following table, which is the fluorescence value of the material in the
6 chambers. List the peak detection results in the table and draw a histogram as follows.The
fluorescence intensity signal can be effectively obtained by peak detection algorithm,
indicating that the constructed multi-channel scanning system is basically correct in principle.
Fig27. Comparison of fluorescence intensity data before and after Kalman filtering
Table6. Fluorescence intensity peak detection results
Fig28. Fluorescence intensity peak detection results
Linear Test and Standard Curve Construction
The known concentration of standard fluorescent dye was diluted to 0.1, 0.2,...,
0.9 μmol / L fluorescein sodium solution, and different concentrations of fluorescein sodium
solution were detected. The linear correlation coefficient between the concentration and the
mean value of fluorescence measurement was 0.9976, indicating that the linearity was good. The
fitting curve formula was y = 0.1469x, which can be used as a standard curve for the conversion
of fluorescein sodium solution concentration and fluorescence measurement value.
Fig29. Operation process diagram.
Fig30. Different concentrations of fluorescein sodium solution
Fig31. Fitting curve of fluorescein sodium concentration and fluorescence detection value
Functional test of fluorescence detection module
After the completion of the previous fluorescence detection module of each module
of the construction work, we through the overall functional testing to prove whether our
fluorescence detection module can achieve the desired goal, achieve accurate fluorescence
detection and get the correct detection results. The function detection of the fluorescence
detection module includes the use of Crispr/Cas12 experiment on the SlipChip, the
corresponding fluorescence value is obtained by fluorescence detection, the content value of
miRNA is converted from the standard curve, and the prognostic risk score model is constructed
by the content value to obtain the risk classification. Our test steps and results are as
follows :
First, the slide chip was used for experiments to obtain the fluorescence results
corresponding to different pore positions. As shown in the following figure, the positive
samples detected at four pore positions of the slide chip were miR144 _ 5p _ 1, miR193a _ 3p _
1, miR126 _ 5p _ 1, and miR15b _ 3p _ 1, respectively. The other two pore positions were
negative samples. After the completion of the experiment, the device constructed by this module
was used for fluorescence detection to obtain the fluorescence content of the positive hole.
Fig32. The detection object graph corresponding to different hole positions of SlipChip
Fig33. Fluorescence detection fluorescein sodium solution
Secondly, in the ' linear test and standard curve construction ' section, we
obtained the relationship curve between fluorescein sodium concentration and fluorescence
detection value. Due to time reasons, we did not do the relationship curve between miRNA and
fluorescence value. Here, the relationship curve between fluorescein sodium and fluorescence
value is used as an alternative to illustrate the problem.
Take four of them as the fluorescence concentration of four miRNAs, and then complete the conversion of the fluorescence concentration of the four miRNAs to the fluorescence content through the standard curve. As shown in the following figure, the fluorescence concentrations of the four positive holes from top to bottom were 0.2,0.3,0.4 and 0.6, respectively, and the corresponding miR144 _ 5p _ 1, miR193a _ 3p _ 1, miR126 _ 5p _ 1 and miR15b _ 3p _ 1 contents were 0.088,0.059,0.044 and 0.029, respectively.
Fig34. Concentration-to-content conversion of four miRNAs
Thirdly, in the modeling module, we construct a formula for calculating the Risk
Score associated with patient survival :
Risk Score = -0.328 × miR144 _ 5p _ 1 + 0.496 × miR193a _ 3p _ 1 + 0.578 × miR126 _
5p _ 1-0.606 × miR15b _ 3p _ 1
The content of the four miRNAs was brought into the scoring model, and the final
risk score was obtained. The risk score of miRNA content in this group of samples was 0.008.
According to the scoring rules ( score ≤ 0.34 belongs to low risk group, 0.340.46 belongs to
high risk group ), this sample belongs to low risk group.