Advancement of On-Chip Mixing Capabilities for RAPID, a Microfluidic Radiotracer Synthesis Platform

ABSTRACT

Cancer is one of the leading causes of death, responsible for nearly 10 million deaths in 2020. One common method of cancer detection is positron emission tomography (PET), which uses radioactive tracers to create detailed maps of areas containing tumors. [¹⁸F]fluorodeoxyglucose ([¹⁸F]FDG) has long been the most commonly used radiotracer due to its ability to detect multiple varieties of cancer. Despite its success, many other radiotracers function to more accurately detect specific cancers and other diseases. As current production methods are designed around large batches of [¹⁸F]FDG, production of these more specialized radiotracers is prohibitively expensive. One solution to this is RAPID, a microfluidic device platform that synthesizes radiotracers on demand in single doses. In radiotracer production, mixing target molecules (antibodies, proteins, etc.) and radionuclides is crucial for successful synthesis. This study tests different microfluidic mixing channel designs, comparing flow rates and channel geometries to determine the most efficient design for radiotracer synthesis. Channel designs followed a zigzag pattern with varying widths, lengths, and pattern repetitions. Solutions of rhodamine b and fluorescein were mixed, and fluorescent microscopy was used to image the channels and calculate an overlap coefficient to determine mixing efficiency. The results of the study found that a 300 µm wide channel, with a pattern segment-to-channel width ratio of 10:1 with a flow rate of 400 µL/min was most optimal. This condition resulted in an overlap coefficient of 0.940, indicating near complete mixing

INTRODUCTION.

Cancer has proven itself to be a global issue, with over 600,000 deaths predicted for 2025 [1]. As cancer is often easier to treat when detected early, tools like radiotracers in combination with positron emission tomography (PET) scans can be very helpful [2]. Due to [¹⁸F]FDG’s success, most radiotracer production is performed in large batches, leading to prohibitive prices for alternatives made in smaller doses. The Radiopharmaceuticals As Precision Imaging Diagnostics (RAPID) project is a potential solution to this but requires a microfluidic mixing channel that is simple to fabricate and fits within the chip’s footprint. The goal of this research is to create a suitable and efficient microfluidic mixing channel design for use in these devices. PET scans have revolutionized cancer screening, allowing for more widespread imaging of tumors. To achieve this, a radioactive tracer, consisting of a target molecule and a radionuclide, is injected. For [¹⁸F]FDG, the target molecule is glucose, and the radionuclide is ¹⁸F. [¹⁸F]FDG is a glucose analog containing a single fluorine atom widely used for cancer detection (Figure 1). [¹⁸F]FDG’s status as a glucose analog is especially helpful as glucose is heavily used by cancer tumors which pick it up before it decays.

PET scans function on a basic level by detecting gamma rays emitted via radioactive decay of a radionuclide. As the tracer decays, positrons are released, which collide with electrons. When they collide, they annihilate one another, and a pair of gamma rays are released at 180° angles. The PET scanner then uses a ring of detectors to record where and when the rays are detected. Computer algorithms are then used to construct a map of where more tracer activity was recorded, allowing medical professionals to locate tumors and issue diagnoses.

Despite its popularity, [¹⁸F]FDG still suffers from one of its own strengths: lack of specificity. Due to glucose being a common molecule used for cellular respiration across the body, [¹⁸F]FDG cannot differentiate between cancer types and typically contains background noise [3]. This is where alternative radiotracers come in. Custom or alternative radiotracers solve the issues of lack of specificity and background noise by using specific molecules based off the intended target. A study conducted in 2025 found that [¹⁸F]FDG struggles detecting bone metastases, especially when compared to a novel radiotracer designed for the task, like [⁶⁸Ga]DOTA-IBA (Figure 2). Custom radiotracers also have found uses in diagnosing non-cancerous diseases like Alzheimer’s disease [4].

Despite the numerous advantages of boutique radiotracers, they still face one major issue: bulk manufacturing. [¹⁸F]FDG’s status as the primary PET radiotracer allows bulk manufacturing of it to be profitable based off a large supply of buyers. Additionally, ¹⁸F’s short half-life of 110 min can restrict its viability and, depending on transportation distance, significantly larger radiotracer batches may be needed to offput the decay [6]. Boutique radiotracers have found little widescale success due to the required small batch production, leading to prohibitive pricing compared to traditional [¹⁸F]FDG. [7].

Microfluidics and lab on chip (LOC) devices like RAPID propose a solution to the accessibility issues of these radiotracers [8-10]. RAPID is a microfluidic system that can produce radiotracers by combining the multiple steps of synthesis into a small chip. This allows users to create single or a few doses of virtually any radiotracer and circumvent many issues within typical bulk manufacturing. By creating custom radiotracers in single doses, the issues with bulk manufacturing like wasted material can be managed to allow for more access to custom radiotracers.

To produce these radiotracers in small scale doses, microfluidic mixing channels are required to properly combine radiolabels and vector molecules. For RAPID, there are multiple restrictions on mixing channel designs. Channels must be able to be fabricated using a resin printer, which limits feature resolution and certain designs such as overhangs. Additionally, any potential channel designs must fit within the limited space on the RAPID chip.

The objectives in this study are to find an optimal microfluidic mixing channel design for mixing radiolabels and vector molecules in a RAPID device chip. This fuels the future goal of lab-on-chip radiotracer synthesis via the RAPID system, expanding radiotracer access. While RAPID’s primary goal is radiotracer synthesis for human use, it will ideally be capable of synthesis for use in other animals such as rodents.

MATERIALS AND METHODS.

RAPID Fabrication Procedure.

Microfluidic devices were fabricated using a 3D printing-based molding process (Figure 3). All prints were modeled in Fusion 360 (Autodesk) and printed using a Sonic Mini 8K (Phrozen). Fusion 360 model files were converted to STL files for slicing which was performed in CHITUBOX. The Sonic Mini 8K has a z-axis layer resolution of 10 μm and an x- and y- axis resolution of 22 μm, supplying the needed precision for device fabrication. Only a single resin type was used for experiments, Aqua 8K (Phrozen). Finished prints were washed using a 70% ethanol solution by Wash and Cure 3 plus (Anycubic) for 3 min to remove uncured resin from the print. Next, the prints were sonicated in 70% ethanol solution for 30 seconds to remove micro-debris. After sonication, the prints were rinsed with DI water and 70% ethanol solution before drying with pressurized air to remove debris before the ethanol evaporated. Once dried, prints were cured for 30 min using a Wash and Cure 3 Plus. Prints were then stored for later modification.

Parylene coating was the next main step. Prints were moved to a PDS 2010 Labcoter 2 (Specialty Coating Systems) loaded with 2 g parylene-c starting material and coated in a thin layer of parylene to provide anti-adhesive properties to the molds. Liquid nitrogen was poured to the top of the cold trap every 15 min to prevent parylene-c from escaping in gaseous form. All other processes were carried out according to manufacturer directions. After parylene coating, a polydimethylsiloxane (PDMS, Sylgard 184; Ellsworth Adhesive Company) mixture of elastomer and curing agent was prepared at a weight ratio of 10:1 and poured over the print while in a petri dish. The PDMS and print were then degassed in a vacuum chamber for 30 min and cured in an oven at 80 °C. After parting the PDMS device from the printed molds, tubing holes were piloted and removed using a modified leather punch. Following hole punching, the PDMS devices were bonded to glass microscope slides using a PlasmaFlo PDC-FMG plasma cleaner (Harrick Plasma). Bonded devices were cured again in an oven at 80 °C for 30 min to ensure an effective bond.

Data Collection.

All testing was performed by flowing serially diluted fluorescein and rhodamine b solutions through a microfluidic mixing channel. The final concentrations of rhodamine b and fluorescein were 24 µM and 30 µM, respectively. Luer-to-microbore tubing connectors were attached to two 3 mL syringes and filled with the prepared concentrations of fluorescein and rhodamine b and inserted into a SyringeEIGHT (New Era Pump Systems) (Figure 4). The microbore tubing connected to the syringes was inserted into the two Y-ports of the microfluidic device. A third piece of microbore tubing ran from the exit port of the device to a glass waste beaker. The device was placed onto a widefield fluorescent microscope for imaging. Flow rate started at 50 µL/min per syringe and was increased by 50 µL/min until reaching 400 µL/min. This resulted in a total flow of between 100 and 800 µL/min, increasing in increments of 100 µL/min. After waiting 1 min for the fluid flow to reach steady state, the start and end of the channels were imaged and saved as CZI files. Pressure testing was also performed, using a fluid pressure sensor and reader.

Data Analysis.

CZI files were ported into FIJI (Fiji Is Just ImageJ) for data analysis. Channels were split into fluorescein and rhodamine b. The brightness of each image was scaled by using the built-in “reset” button, and a line was drawn perpendicular to the channel. The line was then copied onto the other image, and a data series was created from the line across each image. To numerically identify mixing efficiency, an overlap coefficient was calculated. This was done by first converting each fluorescent series into a probability distribution function (PDF). The sum of the minimum of the two PDF’s at each point results in a coefficient ranging from 0 (no overlap at all) to 1 (perfect overlap).

To properly portray the fluorescent intensities, the gray values recorded by the data capturing software were normalized to their respective peak. Additionally, microscope calibration was performed to convert pixels to µm so graphs could be constructed in Excel.

RESULTS.

The objective of this study has been to develop an optimal microfluidic mixing channel design for mixing radiolabels and vector molecules in a RAPID or a similar LOC device. After reviewing previous literature and consulting my mentor, we decided a “Y” junction channel would serve best for mixing purposes, as many previous microfluidic mixing studies used it. One primary variable manipulated in this study was the segment length to pattern width ratio (S:W) (Figure 5). This provided a method to manipulate channel designs in a reliable way and was used in previous microfluidic mixing studies [11].

After testing multiple channel S:W ratios, the most optimal configuration was determined to be a 300 µm wide channel with an S:W ratio of 10:1 (Figure 6). Testing showed that higher pressure yielded a higher overlap coefficient compared to lower pressures (Figure 7). This design proved most suitable for use in an LOC device due to a high overlap coefficient value of 0.940 at 400 µL/min per syringe, and low pressure of 1.55 psi (Table 1). It demonstrated a comparatively high overlap coefficient compared to other tests. Testing wider channels yielded less mixed channels and could also pose difficulties with integration into the RAPID chip due to size.

 

 

Table 1. A table displaying mixing levels of microfluidic channels with an S:W ratio of 1:10.
Syringe Flow Rate (μL/min)	Overlap Coefficient		Pressure (psi)
400		0.940	1.55
350		0.921	1.30
300		0.888	1.07
250		0.902	0.83
200		0.895	0.67
150		0.788	0.48
100		0.675	0.28
50		0.596	0.11

DISCUSSION.

Overall, significant progress was made towards finding a suitable channel design for RAPID’s goal of small-scale production of radiotracers. Although there are preexisting studies into geometry for microfluidic mixing, their channel designs were not suitable for use in RAPID due to limitations like size and required fabrication tools. This study focuses more on creating a mixing channel to satisfy RAPID’s needs specifically. As the first target for synthesis is a radiolabeled antibody, the use of aqueous solutions like fluorescein and rhodamine b serve as an effective model.

Implications and Applications.

These findings help further the process towards the RAPID project and the subsequent goal of increasing accessibility to custom radiotracers that are currently prohibitively priced. This can help more diseases like cancer be diagnosed and mapped for treatment while reducing the total potential costs of undergoing those treatment procedures.

Future Directions.

Further research into pressure testing is important as it has the potential to greatly increase mixing efficiency, and lead to a more optimized mixing channel. Trends show that increasing pressure increases the overlap coefficient, achieving better mixing. Although no pressure failures occurred during this study, continued research should test upper limits of pressure inside the channel, as this could lead to even better mixing results. Additional testing to determine optimal channel distances for mixing would also be beneficial. Due to the small required footprint of a RAPID chip, finding the minimum distance for sufficient mixing would allow for easier integration into multipurpose chips. Experimentation would involve testing linearly along the channel until finding appropriate mixing levels for the corresponding intended use. Sufficient lengths are expected to decrease as lower mixing is needed and increase as a higher overlap coefficient is required.

CONCLUSION

The objective of this study was to determine optimal microfluidic mixing designs for small-scale radiotracer synthesis and integrate the most optimal channel design into RAPID. To accomplish these, PDMS microfluidic chips were fabricated and subsequently tested using fluorescent dyes to assess mixing levels. The best mixing results were achieved using a 300 µm wide “Y” junction channel with a SW ratio of 10:1 at 400µL/min per syringe. The progress towards making novel radiotracers more accessible is still greatly impactful and brings humanity ever closer to fighting debilitating and deadly diseases like cancer, Alzheimer’s, and more.

ACKNOWLEDGMENTS.

This work was supported in part by the National Institute of Biomedical Imaging and Bioengineering of the National Institutes of Health under award number R01EB034665 and a Vanderbilt University Seeding Success grant. Internship opportunity provided by Research Experience for High School Students and Hillsboro High School’s Interdisciplinary Science and Research class.

REFERENCES

1. R. L. Siegel, T. B. Kratzer, A. N. Giaquinto, H. Sung, A. Jemal, Cancer statistics, 2025. CA: A Cancer Journal for Clinicians. 75, 1 (2025).
2. D. Crosby, et al. Early Detection of Cancer. Science. 375, 1 (2022).
3. W. A. Weber, M. Schwaiger, N. Avril, Quantitative assessment of tumor metabolism using FDG-PET imaging. Nuclear Medicine and Biology. 27, 683–687 (2000).
4. A. Prem Kumar, N. Singh, D. Nair, A. Justin, Neuronal PET tracers for Alzheimer’s disease. Biochemical and Biophysical Research Communications. 587, 58–62 (2021).
5. L. Li, et al. Comparison of 18F-FDG and 68Ga-DOTA-IBA in detecting bone metastases: a lesion-basis study. Scientific Reports. 15, 7 (2025).
6. Radiopharmaceutical – JF Part B – Noridian. JF Part B (2025), (available at https://med.noridianmedicare.com/web/jfb/fees-news/fee-schedules/radiopharmaceutical-fees).
7. H. Redman, “Radionuclide Safety Data Sheet Radionuclide: F-18 Forms: Soluble Physical Characteristics” (2020), (available at https://ehs.stanford.edu/wp-content/uploads/F-18-RSDS.pdf).
8. X. Zhang, et al. A simple microfluidic platform for rapid and efficient production of the radiotracer [¹⁸F]fallypride. Lab on a chip. 18, 1369–1377 (2018).
9. X. Zhang, et al. High-Yielding Radiosynthesis of [68Ga]Ga-PSMA-11 Using a Low-Cost Microfluidic Device. Molecular Imaging and Biology. 22, 1370–1379 (2020).
10. M. Mc Veigh, L. M. Bellan, Microfluidic synthesis of radiotracers: recent developments and commercialization prospects. Lab on a Chip. 24, 1226–1243 (2024).
11. C.-Y. Lee, C.-L. Chang, Y.-N. Wang, L.-M. Fu, Microfluidic Mixing: A Review. International Journal of Molecular Sciences. 12, 3263–3287 (2011).

---

---

Posted by buchanle on Friday, May 15, 2026 in May 2026.

Tags: microfluidic, Radiotracer, RAPID