When Embedded Systems meet Life Sciences: Microfluidic Biochips for Real-Time Healthcare

Abstract

Biochips are cyber-physical system with realistic potential to improve the healthcare process, e.g., by providing faster disease diagnosis and at-home direct treatment. We review the area of biochips on its way to becoming a strong research field. However, for the real breakthrough to happen, a stronger collaboration among disciplines is needed. Thus, we organized this tutorial at Embedded Systems Week, to inform researchers about the potential of biochips. We presented the vision, the research challenges and an overview of the design automation as well as of the fabrication work. We show how the conventional methods in embedded systems can be applied to biochips to optimize their execution and design. Besides that, we conducted a hands-on lab that allowed the participants to get in direct contact with biochips. The participants were highly inspired and we hope their interest will result in fruitful collaborations.

Full text

When Embedded Systems meet Life Sciences:
Microfluidic Biochips for Real-Time Healthcare
Mirela Alistar
Hasso-Plattner Institute
mirela.alistar@hpi.de
Jan Madsen
Technical University of Denmark
jama@dtu.dk
ABSTRACT
Biochips are cyber-physical system with realistic potential to
improve the healthcare process, e.g., by providing faster dis-
ease diagnosis and at-home direct treatment. We review the
area of biochips on its way to becoming a strong research field.
However, for the real breakthrough to happen, a stronger col-
laboration among disciplines is needed. Thus, we organized this
tutorial at Embedded Systems Week, to inform researchers
about the potential of biochips. We presented the vision, the re-
search challenges and an overview of the design automation as
well as of the fabrication work. We show how the conventional
methods in embedded systems can be applied to biochips to op-
timize their execution and design. Besides that, we conducted a
hands-on lab that allowed the participants to get in direct con-
tact with biochips. The participants were highly inspired and
we hope their interest will result in fruitful collaborations.
Keywords: biochips, real-time healthcare, design automa-
tion, routing, microfluidics, genetics
VISION: BIOCHIPS FOR REAL-TIME HEALTH
The conventional approach to healthcare (Figure 1) takes a
significant time from the installment of disease to the patient
being cured. In many cases, this delay may lead to the death
of the patient. It is sound to affirm that receiving healthcare
in time is critical.
That is the main reason behind the many phone applications
developed for healthcare monitoring. A future cyber-physi-
cal system based on biochips, will enable (i) a faster disease
diagnosis and (ii) at-home direct treatment.
Figure 1: A biochip-based system can shorten the delay from
the installment of the disease to the patient being cured.
Tsung-Yi Ho
National Tsing Hua University
tyho@cs.nthu.edu.tw
Robert Wille
Johannes Kepler University
robert.wille@jku.at
The main advantage of our system is accessibility: more peo-
ple will have access to healthcare. Another important ad-
vantage is a faster healthcare process, by reducing the delay
from the installment of the disease to the cure.
Figure 2: The main advantage of the system is accessibility:
more people will have access to healthcare.
HIGH-LEVEL SYSTEM
Figure 3 presents the envisioned system for on-patient real-
time detection of biomarkers. A phone controls the system
and collects the results. The data is stored in the cloud for
further analysis. The patients are later informed on the deci-
sion of the experts (‘e-doctors’) regarding their treatment.
Figure 3: Envisioned system for biomarkers detection.
Droplet-based biochips were introduced in the late 2000s as
a promising solution to a ‘lab-on-a-chip’ that can automate,
miniaturize and integrate complex biochemical applications
[4]. Figure 4 shows a prototype for the envisioned system.
Figure 4: A paper-based biochip controlled by the software
running on a phone. The communication is wireless and thus,
can be done at distance. A small prototype was demonstrated
for in-vitro diagnosis.

TECHNOLOGY
Biochips are electronic devices that can perform the tasks
traditionally performed by a human wet lab technician with
a pipette, such as in-vitro diagnosis.
Figure 5: The OpenDrop biochip has four reservoirs and 8x8
electrodes holding a single droplet. The biochip can move the
droplet by applying a voltage to a neighboring electrode.
As shown in Figure 5, biochips consist of electrodes, with
each electrode capable of holding a droplet. To move a drop-
let, biochips apply electrical voltage on a neighbor electrode.
The voltage attracts the droplet and the droplet moves. In
such way, biochips can run biochemical processes (‘bio-pro-
tocols’), i.e., sequences of operations on droplets, including
dispensing, splitting and mixing (Figure 6).
Figure 6: To execute biochemical processes or bio-protocols,
biochips (a) dispense, (b) split, (c) merge and (d) mix droplets.
CHEAP FABRICATION
Significant research efforts have been directed towards fab-
ricating a cheap and reliable biochip. The most successful
fabrication techniques so far are based on chromium elec-
trodes on a glass substrate (Figure 7), gold electrodes on a
printed circuit board, as in Figure 5, and silver electrodes
printed on photographic paper.
Figure 7: Dropbot is built using vapor deposition of chrome on
a glass substrate. This fabrication method is the most expen-
sive, but the glass substrate can be easily sterilized and thus re-
used multiple times [1].
The recent development of paper-based biochips (Figure 8)
is a significant step towards our vision. These biochips are
printed on gloss photographic paper using a home inkjet
printer that was previously loaded with conductive ink. Pa-
per-based biochips are very cheap, and thus disposable. Con-
sequently, contamination can be avoided in a simple manner
(i.e., by disposing the biochips after each use), without the
need of sterilization.
Figure 8: Paper-based biochips have ink-jet printed electrodes.
This biochip uses a cheap ink based on carbon-nanotubes for
conductivity [3]. These biochips are disposable, making them
highly suitable for point-of-care diagnosis.
However, there are several limitations to paper-based bio-
chips: (i) short shelf life and (ii) limited number of elec-
trodes, because, unlike PCB, multiple layering is not possi-
ble on paper.
SYNTHESIS SOFTWARE
To synthesize a bio-protocol means translating it into droplet
movements. The synthesis of bio-protocols is a NP-complete
problem, very similar to the conventional embedded systems
problem of scheduling a task graph on a given architecture
with a defined set of resources. Traditionally the synthesis is
divided in a series of four steps: (1) binding of the droplets
(tasks) to the electrode modules (resources), (2) scheduling
of the droplets, (3) placement of electrode modules on the
biochip grid and (4) routing the droplets from one module to
another.
Figure 9: The traditional synthesis problem consists of binding
and scheduling the tasks (here, droplets) on the given resources
(here, electrodes) that need to be placed on the architecture
(here, biochip). Droplets also act as data (messages) that need
to be routed from one resource to another.
We review a novel approach that synthesizes in one-step, us-
ing Boolean satisfiability [2]. Thus we formulate the synthe-
sis problem as a sequence of decision problems. The pro-
posed SAT encoding proved to provide better results than
the traditional approach.

Figure 10: One-step synthesis using Boolean satisfiability [2].
SOLVING THE ROUTING CONFLICTS
Determining the routes of the droplets is NP-complete. Rout-
ing of droplets has to be done such that (i) there is no unde-
sired droplet merging and (ii) the bio-protocol completes in
time. We call (i) the fluidic constraint and (ii) the timing con-
straint. Figure 11 depicts the fluidic constraint.
Figure 11: When two droplets are too close they merge instantly
together by default. A minimum spacing of one electrode is re-
quired for both stationary and moving droplets.
The routing problem can be summarized as follows:
Given as input: a netlist of n droplets D = {d1, d2,…, dn},
the locations of m, the blockages B = {b1, b2,…, bm}, and
the timing constraint Tmax, we have as objective to route all
droplets from their sources to their targets, satisfying the
fluidic and timing constraints. Figure 12 further shows the
example of dynamic control interference between droplets
and control lines on paper-based biochips.
Figure 12: (a) Routed control lines (b) Droplets and are sched-
uled to move rightward at the same time (c) The control line
of d1 adversely affects the movement of droplet d2 due to the
voltage interference (d) The control line of is rerouted to re-
solve the conflict [5].
The problem is solved by modeling the biochip as a network-
flow graph with specific minimum cost flow formulation.
After we determine an escape route, we re-route the droplets
using A* search. Researchers have also proposed a co-de-
sign routing that adjusts the cost function depending on the
current routing scenario.
HANDS-ON LAB
The participants (including 2 female researchers) were in-
volved in a series of practical tasks meant to consolidate the
knowledge presented so far. All tasks were executed in
teams of two, facilitating collaboration and networking.
First, the participants were challenged to search around for
the combination of fluid and substrate that would give the
highest contact angle. We tested diluted soap, coffee, milk
and water on the following substrates: phone cover, confer-
ence badge, disposable waxed cup and tea bag envelope. The
participants learnt the importance of the contact angle (and
its properties) for biochips.
Figure 13: Coffee, soap, milk and water droplets on the back of
a conference badge.
The second task for the participants was a series dilution of
flavored droplets to achieve the ideal concentration for a
contact angle of 120 degrees.
Figure 14: Participants perform a series dilution on several flu-
ids.
Then, the participants were exposed to a live demonstration
of a biochip moving and mixing droplets. As a third task,
each team had to develop their own bio-protocol for a sleep
enhancer, a Pennsylvania aroma, cloth detergent flavor and
insect repellent. For this task, the teams were given 8 fluids
of different aromas that had to be combined to define the bio-

protocol. The participants also determined the routes and the
scheduling of their own bio-protocol.
Figure 15: Side view of the real-biochip moving a droplet.
DISCUSSIONS AND FUTURE WORK
The participants reported that the tutorial was informative,
and that the information load was not too heavy. Five partic-
ipants would have liked even more technical details and they
all reported that the hands-on lab was easy to understand
and perform: ‘although I have never ever done anything like
that before I still got a feeling I learnt it’. All participants
found the tutorial inspiring and when asked about collabo-
ration interest, one suggested that he could use his tech-
niques (in Matlab, Simulink) to model and verify fluids. One
participant found the team work very inspiring and it al-
lowed her to interact and learn more from the others. An-
other participant suggested to split the tutorial over two days,
with a break in between, giving them the chance to sleep
over the knowledge and get more ideas.
Overall, the participants were extremely enthusiastic
and interested in the topic.
As future work, we presented potential applications in syn-
thetic biology, a living pill (Figure 16) and cancer cures.
Figure 16: The living pill analyzes the biomarkers and pro-
duces the cure [6].
CONCLUSIONS
We believe that the area of biochips is becoming a strong
field that require strong interdisciplinary collaborations. A
lot of the classical methods in embedded systems can be ap-
plied to biochips to optimize their execution and design.
Moreover, existing embedded systems can be integrated
with biochips to form advanced cyber-physical systems.
Since the field is just emerging, any research contribution
has significant impact. Thus, we strongly encourage en-
gagement of embedded systems researchers in address-
ing the current problems concerning biochips.
Through running such tutorial, we aim at informing the re-
searchers about the potential of biochips. The participants
were interested and we hope their interest will result in re-
search collaboration as we aim at expanding the field.
REFERENCES
1. Ryan Fobel, Christian Fobel, and Aaron R. Wheeler.
2013. DropBot: An open-source digital microfluidic
control system with precise control of electrostatic driv-
ing force and instantaneous drop velocity measurement.
In Applied Physics Letters, 102, 19, 193513-193520.
2. Oliver Keszocze, Robert Wille, Tsung-Yi Ho, and Rolf
Drechsler. 2014. Exact one-pass synthesis of digital mi-
crofluidic biochips. Proceedings of the 51st Annual De-
sign Automation Conference. ACM, 1-6.
3. Hyojin Ko, Jumi Lee, Yongjun Kim, Byeongno Lee,
Chan‐Hee Jung, Jae‐Hak Choi, Oh‐Sun Kwon, and
Kwanwoo Shin. 2014. Active Digital Microfluidic Pa-
per Chips with Inkjet‐Printed Patterned Electrodes. Ad-
vanced Materials 26.15, 2335-2340.
4. Fei Su, Krishnendu Chakrabarty, and Richard B. Fair.
2006. Microfluidics-based biochips: technology issues,
implementation platforms, and design-automation chal-
lenges. In IEEE Transactions on Computer-Aided De-
sign of Integrated Circuits and Systems, 25, 2. 211-223.
5. Sying-Jyan Wang, Katherine Shu-Min Li, and Tsung-Yi
Ho. 2016. Congestion-and timing-driven droplet routing
for pin-constrained paper-based microfluidic biochips.
21st Asia and South Pacific Design Automation Confer-
ence (ASP-DAC). IEEE.
6. Lina Schukur, Barbara Geering, Ghislaine Charpin-El
Hamri, Martin Fussenegger. 2015. Implantable synthetic
cytokine converter cells with AND-gate logic treat ex-
perimental psoriasis. Synthetic Biology. 17, 118, 1-12.