Team:NCKU Tainan/Hardware


Hardware - CreSense

Fig. 1. CreSense final product.


Recent research has shown that the accumulation of p-Cresol in patient’s blood is highly associated with the deterioration of chronic kidney disease and cardiovascular diseases, which makes the monitoring of blood p-Cresol all the more necessary and important. Currently, blood p-Cresol levels is measured by high-performance liquid chromatography (HPLC) followed by mass spectrometry (MS). However, the cost of HPLC-MS instrument and the need for specialized training to operate it makes it harder for p-Cresol inspection even harder to gain its universality. Hence, we designed CreSense, to provide a simpler alternative to measuring blood p-Cresol levels. Without needing any special training to operate the device and costing just 1.6 USD for a single test, we hope to make blood p-Cresol inspection more accessible to the public.

Device Overview

Fig. 2A. CreSense prototype.
Fig. 2B. CreSense illustration.

CreSense, at its very core, is a user-friendly, affordable and customizable biosensing device. Unlike many other biosensors that only allows sensing, CreSense is also equipped with a centrifugal platform capable of separating blood plasma from whole blood. This allows the detection of biomarkers inside the blood without pre-treating the blood sample separately. Sensing is done via an engineered bacteria that is able to sense p-Cresol and induce expression of green fluorescent protein (GFP). The fluorescence emission can then be detected by a light-to-frequency converter (TSL 235R) inside the device and uploaded onto an online realtime database. The p-Cresol sensing bacteria can be replaced with other biomarker sensing bacteria, thus enabling CreSense to have a wide range of applications.

Device Budget

Table 1. CreSense: Overall device budget.
Component Quantity Cost per unit Cost
Arduino Uno 1 160 NTD (5.23 USD) 160 NTD (5.23 USD)
NodeMCU ESP8266 1 337 NTD (11 USD) 337 NTD (11 USD)
Electric speed controller 1 550 NTD (17.96 USD) 550 NTD (17.96 USD)
Brushless motor 1 450 NTD (14.70 USD) 450 NTD (14.70 USD)
TSL 235R 1 135 NTD (4.41 USD) 135 NTD (4.41 USD)
Filter 1 520 NTD (16.98 USD) 520 NTD (16.98 USD)
Casing 1 500 NTD (16.33 USD) 500 NTD (16.33 USD)
Screws 8 7.5 NTD (0.24 USD) 60NTD (1.92 USD)
Lithium-ion battery 1 300 NTD (9.80 USD) 300 NTD (9.80USD)
Sensor holder 1 200 NTD (6.53 USD) 200 NTD (6.53 USD)
Total 3212NTD (104.86 USD)

Blood Plasma Separation

There are components inside the human blood that could be lethal to the p-Cresol sensing bacteria. In order to allow survival of the p-Cresol sensing bacteria inside the blood and produce fluorescent protein, it is necessary to separate plasma from whole blood beforehand. We searched for a simple way to separate blood plasma while integrating all the required steps, before deciding on using a centrifugal microfluidic platform to achieve this goal[1].

In the microfluidic centrifugal platform, fluidic processing steps such as the mixing of reagents or the transfer of samples can be done simply by implementing different spinning profile[2]. After blood plasma separation, users only need to inject the sensing bacteria into the microfluidic chip and the device can begin detecting and quantifying the results. The simplicity of operating CreSense ensures that no specialized training is required and can thus be operated easily by healthcare professionals in diagnostic centers.

Fig. 3. Centrifugal platform.

CreSense is equipped with a brushless motor to reduce the operational and mechanical noises, and also to obtain higher torque to weight ratio.

Fig. 4. Exploded view for acrylic disc.

The microfluidic centrifugal platform disc is made by three layers of a laser-cut-machine-processed acrylic disc which is combined by two layers of adhesives.

Fig. 5. Microfluidic channel design.

For each slot, we designed two chambers. The chamber on the right (trapezoidal) is the blood separation chamber, the one on the left (square) is the reaction chamber. The blood separation chamber is where we inject whole blood for plasma separation and the reaction chamber is where the p-Cresol sensing bacteria reside and express GFP.

Fig. 6. Successful plasma separation.

By implementing different spinning profiles, plasma can be extracted from whole blood and directed into the reaction chamber. After injecting blood samples, the disc spins at 3000 RPM (600 RCF) for 10 minutes where the plasma and blood cells would be separated in the blood separation chamber. Then, we lowered the speed to 900 RPM (50 RCF) and capillary forces will direct plasma into the reaction chamber through the lower channel between two chambers.

At the beginning of designing the centrifugal microfluidic platform, we used polydimethylsiloxane (PDMS) to make microfluidic chips as sample holders and tested several versions. However, PDMS chip‘s test continues to give unsatisfactory results due to bonding issues. In the middle of August, we consulted Professor Ruey-Jen Yang, who suggested replacing PDMS with acrylic due to its strength and deformation resistance properties. In the end, we choose acrylic as the material used in the device. Both methods are listed below.

Fig. S1A. PDMS microfluidic chip.

We made acrylic molds for casting PDMS into the desired shape, then bonded another flat layer of PDMS onto it with the help of oxygen plasma treatment to cover up the channels, forming a three-dimensional structure. The chamber on the right (trapezoidal) is the blood separation chamber, the one on the left (square) is the reaction chamber. The blood separation chamber is where we inject whole blood to separate plasma. The reaction chamber is where the GFP production takes place. From June to July, we made three different versions of molds with minor changes solving the problems we faced.

Fig. S2A. Microfluidic chip version 1.
Fig. S3A. Microfluidic chip version 2.
Fig. S4A. Microfluidic chip version 3.

In version 1, we found out that when we insert fluid into the chip, there would be bubbles stuck at the corner of the chamber. We solved it by filleting every corner in version 2. When testing version 1 & 2, the bottom of the whole blood chamber wasn’t strong enough to endure the force the sample creates when the spinner’s rotational speed reaches 2500 RPM. So, we tried to overcome this by reinforcing the bottom part of the blood separation chamber in version 3.

After we made the mold for version 3, we consulted Professor Ruey-Jen Yang. He suggested us to use glass slides instead of another layer of PDMS to cover the channels, so the bonding will be stronger compared to PDMS-PDMS bonding. We followed his advice and the result turned out to be unsatisfying - as several attempts at bonding failed. However, after a week of debugging, we obtained successful ones by adjusting the parameters of the oxygen plasma machine.

The final test for PDMS chip

In the final test, we put three microfluidic chips (PDMS-glass bond) on the chip holder and raise the spinning profile to 3000 RPM for 10 minutes and then lowered 900 RPM for 5 minutes[1]. The PDMS-glass bond chip is able to withstand the centrifugal force. With the adjusted parameters of the oxygen plasma machine, the bond between PDMS-PDMS layer was also strengthened and it too worked fine for blood plasma separation. However, we chose to use the acrylic disc instead of PDMS chip in the end because PDMS chip has a longer fabrication time.

Fig. S1B. Acrylic disc

As mentioned above, we replaced the acrylic chip holder and PDMS-glass bond microfluidic chip with an engraved acrylic disc to obtain better results. We continued to use the microfluidic channel that we designed in the PDMS chip, but with some minor changes.

A major problem we encountered when injecting blood into the PDMS chip is that the blood will flow into the reaction chamber instead of being confined in the blood separation chamber when before we started the separation procedure. So, we designed a passive valve in the connecting channel between the blood separation chamber and reaction chamber to prevent blood overflowing into the reaction chamber.

Fig. S2B. Passive valve

The final test for acrylic disc

In the final test for the acrylic disc, we repeated the same experiment we did for the PDMS chip final test. As seen in the picture below, the acrylic disc is able to successfully separate blood plasma. We can see a significant separation of plasma, and it also flowed into the reaction chamber as expected.

Fig. S3B. Successful acrylic disc plasma separation.

Blood p-Cresol Reader

For determining p-Cresol levels in patient’s blood, CreSense’s integrated fluorescent detector quantifies light intensity emitted by the p-Cresol sensing bacteria, which expresses GFP when in contact with blood plasma p-Cresol.

Fig. 7. CreSense: fluorescence detector.

In our BioBrick construct, we are using GFP as our reporter. To excite the emission of fluorescence by the GFP, we used a blue laser. The optimal fluorescence emission is at around 515 nm. A 510 nm filter is used to block the excitation light and allow only the emission light to pass through.

With the aid of a light-frequency converter - TSL 235R as our light sensor, we will be able to read different emission light intensity. Sensor readings will be converted to p-Cresol concentration and displayed on the LCD screen and uploaded onto our realtime database for further monitoring.

Our device is able to measure fluorescence intensity equivalent to 50 nM of fluorescein. We used the Fluorescein Sodium Salt provided in the measurement kit for our sensor calibration.

Fig. 8. Calibration test with Fluorescein Sodium Salt.

We also tested our fluorescence detector by adding autofluorescence E. coli (pSB1C3-j23100-sfGFP in DH5α) into blood plasma and we are now positive that our device can successfully sense the fluorescence emission from the bacteria in plasma.

Fig. 9. Detector test with autofluorescence E. coli.
Fig. S1C. Initial design of CreSense fluorescence detector.

When we first designed our fluorescence detector, we used LDR as our light sensor. However, we soon realized that without a filter, our detector reading could not reflect the actual fluorescence intensity. From Fig. S2C, we can see that the light sensor works fine for light transmittance measurement. The darker the surrounding is, the lower the LDR value.

Fig. S2C. First detector test.

This does not fit the purpose of our device which is fluorescence measurement. To measure only the fluorescence emission but not the total visible light transmission, we decided to add a narrow bandpass filter to block the excitation light (510 nm) from the blue laser reaching the light sensor. We also changed the position of the light sensor and avoid a collinear arrangement. Not only that, we replaced the light-dependent resistor with our TSL 235R light-frequency converter thus improving the accuracy of our device.

Online Database

CreSense is expected to be placed at all healthcare service providers. We connected the onboard Arduino Uno to a NodeMCU to upload detector readings onto Firebase - an online real-time database. These data can then be accessed by healthcare professionals. In the future, we are hoping that we can integrate this database with Taiwan’s Ministry of Health and Welfare database. By doing so, our device can make a greater impact by helping to monitor the blood p-Cresol level of our CKD patients.

Arduino code and Circuits

We use Arduino IDE to code for our Arduino UNO and a NodeMCU board for our whole device. The complete code can be found on Github.

Shown below is our entire wiring sketch:

Fig. 10. CreSense: Arduino sketch


CreSense is a user-friendly, affordable and customizable biosensing device, and is expected to be placed at all healthcare service providers as a simple and accessible alternative for measuring blood p-Cresol levels. To demonstrate how our engineered system works, we have made a video to explain the workings of CreSense and its operating processes.

Click me to see how it works!


  1. Gorkin, R., Park, J., Siegrist, J., Amasia, M., Lee, B. S., Park, J.-M., … Cho, Y.-K. (2010). Centrifugal microfluidics for biomedical applications. Lab on a Chip, 10(14), 1758. doi: 10.1039/b924109d
  2. Amasia, M., & Madou, M. (2010). Large-volume centrifugal microfluidic device for blood plasma separation. Bioanalysis, 2(10), 1701–1710. doi: 10.4155/bio.10.140