Académique Documents
Professionnel Documents
Culture Documents
Digital Microfluidics-Based
Biochips
Microfluidic
Lab-on-Chip
Conventional Biochemical
Analyzer
2
Microfluidics
Continuous-flow biochips: Permanently etched
microchannels, micropumps and microvalves
Digital microfluidic biochips: Manipulation of
liquids as discrete droplets (digital microfluidics)
3
Integration of microfluidics: one of the system-
level design challenges (≤ 50nm/beyond 2009)
2003 International Technology Roadmap for
Semiconductors (ITRS)
Heterogeneous SOCs
-Mixed-signal
-Mixed-technology
4
Outline
Motivation
Background
Related prior work
Architectural-level synthesis for digital
microfluidic biochips
Sequencing graph model
Mathematical programming model
Heuristics for scheduling problem
Simulation experiments
Conclusions
5
Background
Novel microfluidic platform invented at Duke University
Droplet actuation is achieved through an effect called
electrowetting
Electrical modulation of the solid-liquid interfacial tension
6
Background (Cont.)
Actuation principle of digital microfluidics
Droplet Transport
A droplet can be transported by
removing a potential on the
current electrode, and applying a
potential to an adjacent electrode.
7
Background (Cont.)
Digital microfluidics-based biochips
TRANSPORT
TRANSPORT DISPENSING
DISPENSING MIXERS
MIXERS REACTORS
REACTORS DETECTION
DETECTION
8
Background (Cont.)
The in-vitro measurement of glucose in human
physiological fluids
Glucose + H 2 O + O 2 Glucose
Oxidase
→ Gluconic Acid + H 2 O 2
2H 2 O 2 + 4 - AAP + TOPS Peroxidase
→ Quinoneimi ne + 4H 2 O
9
Background (Cont.)
Digital microfluidics-based biochip used in glucose
measurement
Dispensing/Transportation:
Sample droplet (glucose) and reagent
droplet (glucose oxidase, peroxidase,
4-AAP and TOPS), are dispensed into
the microfluidic system from reservoirs.
Mixing:
Length of the control electrode Sample droplet and reagent droplet are
L = 1.5mm mixed in a mixer (i.e. 2x2 array mixer).
Height between two plates
H = 475nm
Thickness of insulator layer Optical Detection:
(parylene C) = 800nm Assay result (quinoneimine) is detected
Thickness of hydrophobic film
by a green LED and a photodiode.
(Teflon AF) = 60nm
10
Background (Cont.)
Detection of lactate, glutamate and pyruvate has also been
demonstrated.
Biochip used for multiplexed in-vitro diagnostics on human
physiological fluids
Scheduling of operations
Binding to functional
resources
Physical design
12
Related Prior Work
Synthesis of integrated circuits: well-studied
problem
MEMS simulation & synthesis tools
CAD tools for microfluidic biochips
Physical-level simulation: CFD-ACE+, FlumeCAD
System-level design: CoventorWare (targeted at
continuous flow system)
13
Architectural-Level Synthesis
Sequencing graph model
Multiplexed in-vitro diagnostics
14
Sequencing Graph Model (Cont.)
Node representing the input operation
Sj Rj
Ij Ij+m
15
Sequencing Graph Model (Cont.)
Node representing different types of mixing operations
S1 Ri S2 Ri Sm Ri
…...
M1 M2 Mm
16
Sequencing Graph Model (Cont.)
Node representing the detection operations
S i+R1 Si +R2 Si+Rn
D1 D2 …... Dn
17
Sequencing Graph Model (Cont.)
Assumption 4: In contrast to the above operations,
droplet movement on a digital microfluidic array is
very fast. We can ignore the droplet movement
time for scheduling assay operations.
18
Sequencing Graph Model (Cont.)
NOP Input
operations:
2mn Nodes
… …… …… …… …… …… …
S1 S1 Sm Sm R1 Rn R1 Rn
Mixing
M1 M1 Mm Mm operations:
………… ………… …………
2mn+1 3mn mn Nodes
……… ……… ………
Detection
D1 ………… Dn ………… D1 ………… Dn operations:
3mn+1 4mn
… … …
mn Nodes
multiplexed bioassays 30
this time period
D1
19
Mathematical Programming Model
Objective Constraints
First define a binary variable Dependency constraints
Stj ≥ Sti + d(vi) if there is a dependency
1 if operation vi starts at
X ij = time slot j. between vi and vj
0 otherwise
Resource constraints
Starting time of operation vi : Reservoirs/dispensing ports
T
St i = ∑ j × X ij Nr reservoirs/dispensing ports assigned to
j =1 each type of fluid (Nr = 1)
20
Mathematical Programming Model (Cont.)
Evaluated for a problem of the modest size:
Plasma and serum are sampled and assayed for glucose,
lactate and pyruvate measurements; i.e., m = 2, n = 3
Assume Nr = Nd = 1, and Na = 4
Time Reservoirs Reconfigurable Optical Detectors
step for S1 for S2 for R1 for R2 for R3 Mixers for D1 for D2 for D3
1
S1 S1 S1 S2 S2 S2 R1 R2 R3 R1 R2 R3 1 6 7 12
2
3
I1 I1 I1 I2 I2 I2 I3 I4 I5 I3 I4 I5 2 8 18
4
3 9 13
1 2 3 4 5 6 7 8 9 10 11 12 5
6
4 10 14
d(Ii)=1 7
5 11 15
8 24
M1 M1 M1 M2 M2 M2 d(M1)=5 16
9
13 14 15 16 17 18 d(M2)=3 17 19
10
d(D1)=5 11 20
D1 D2 D3 D1 D2 D3 d(D2)=4 12
19 20 21 22 23 24 d(D3)=6 13
14 21
22
15 23
NOP 16
17
22
Simulation Experiments
Lower bound (LB)
LB = m×max{d(D1),… d(Dn)}+min{d(M1),… d(Mm)}+d(Ii)+1
Upper bound (UB)
UB = m× max{d(D1),…d(Dn)}+k×max{d(M1),…d(Mm)}+max(m, n)×d(Ii)+ 1
Reservoirs
Mixer with Detector with
Reservoirs smallest d largest d Phase I
1
Input operations:
Max duration =
Min{d(M 1), Worst case: 2mn storage units needed
…, d(Mm)} max(m, n)
Phase III
Detection operations:
Max duration
= m×max{d(D1),
…, d(Dn)}
(a) (b) 23
Simulation Experiments (Cont.)
Five examples (four samples) S1: Plasma, S2: Serum, S3:
Urine, S4: Saliva, Assay1: Glucose assay, Assay2: Lactate assay,
Assay3: Pyruvate assay, Assay4: Glutamate assay
Example Description
Example 1 S1 and S2 are assayed for
(Nr=Nd=1,Na=3) m=2, n=2 Assay1 and Assay2.
1.2
Ratio of Heuristic/Lower Boun
0.8
GA/LB
M-LS/LB
0.6
0.4
0.2
0
1 2 3 4 5
Example
25
Conclusions
A system design methodology to apply classical
architectural-level synthesis techniques to digital
microfluidics-based biochips
An optimal strategy based on integer linear
programming for scheduling assay operations
under resource constraints
Two heuristic techniques that scale well for large
problem instances
M-LS: computationally more efficient
GA: yields lower completion times for bioassays
A clinical diagnostic procedure used to evaluate
the proposed methodology
26