Sensors and Actuators A 162 (2010) 297–303
Contents lists available at ScienceDirect
Sensors and Actuators A: Physical
journa l homepage: www.e lsev ier .co
Smart o pl
Monica V ca C
Pietro Va Sar
a CRIM Lab, Scu
b NEURICAM s.
a r t i c l
Article history:
Received 30 Se
Received in re
Accepted 23 M
Available onlin
Keywords:
Biomedical
Imaging
Sensor
CMOS
ed fo
el ar
and
al pin
8m
igh se
rized
form
ff-th
ensor
1. Introduction
The continuous quest for painless diagnostic procedures in
the gastro-intestinal tract has resulted in greater interest in
endolumina
scopic caps
performs a
lowable cap
Jacobson’s
ture from t
endoscopy
during the
developed
approval in
mercialized
especially d
PillCamTM S
provides im
and biopsy
capsule [2,6
The main g
through im
For this rea
and innova
∗ Correspon
E-mail add
Given Imaging is a major worldwide industrial player in the
field of capsular endoscopy and commercializes solutions for dif-
ferent gastro-intestinal tracts: the PillCmTMSB, PillCmTM ESO and
PillCmTMCOLON. All these pills implement a complementary metal
0924-4247/$ –
doi:10.1016/j.l techniques, such as capsular endoscopy [1]. An endo-
ule is a swallowable self-contained microsystem which
sensing or actuating function in the body [2]. The swal-
sule concept first appeared in 1957 in Mackay and
paper on RF transmission of pressure and tempera-
he human body [3]. Although the concept of capsule
emerged as an alternative to traditional endoscopy
80s and 90s, the first capsule endoscope model was
by Given Imaging in 2000 [4] and received medical
Western countries in 2001. The first capsule was com-
by Given Imaging with the name of PillCamTMSB,
esigned for small bowel investigation. Essentially, the
B is a swallowablewirelessminiaturized camerawhich
ages. Despite research on actuation [5], drug delivery
techniques that may be implemented in an endoscopic
,7], the imaging unit is still the core part of the system.
oal of endoscopy is to inspect the inside of the body
aging techniques for diagnostic and surgical purposes.
son, image quality is a primary issue in both traditional
tive endoscopic devices.
ding author. Tel.: +39 050883483; fax: +39 050883496.
ress: m.vatteroni@sssup.it (M. Vatteroni).
oxide semiconductor (CMOS) imager acting as sensor, but they
present different frame rate characteristics. Alternative endoscopic
pills to those commercialized by Given Imaging are the EndoCap-
sule createdbyOlympus [8], theMiroCamTM by IMC [9], andOMOM
by Jinshan Science & Technology Company [10]. Although CMOS is
themost common technology, the EndoCapsule integrates a Charge
Couple Device (CCD). Resolution of these systems ranges between
256 × 256 and 1000 × 1000 pixels, while the frame rate ranges
between 2 and 7 frames per second (fps). The basic trade-off for a
vision system for capsular endoscopy is to be found between high
image quality and other features such as size, power consumption,
simple control interfaces, image resolution and frame rate. Off-
the-shelf chips can only partially fulfil these requirements. Some
sensors provide good image resolution and quality in a small size,
but lack adequate sensitivity and acceptable power consumption
[11]. Other imagers feature lowpower requirements and small size,
but still present poor output in terms of noise and image quality
[12]. Novel sensors have appeared on the market over the last few
months featuring low power consumption, good image quality and
small size [13].However, their analogueoutputmakes themunsuit-
able for capsular endoscopy because a companion chip is needed
for converting the analogue output into digital word. This results in
an increase in space and power consumption. For this reason, the
highly specific anddemanding requirements of capsular endoscopy
see front matter © 2010 Elsevier B.V. All rights reserved.
sna.2010.03.034ptical CMOS sensor for endoluminal ap
atteronia,∗, Daniele Covib, Carmela Cavallotti a, Lu
ldastri a, Arianna Menciassi a, Paolo Darioa, Alvise
ola Superiore Sant’Anna, via Piaggio, 34, 56025Pontedera, Pisa 56100, Italy
r.l., Trento 38100, Italy
e i n f o
ptember 2009
vised form 22 March 2010
arch 2010
e 4 April 2010
a b s t r a c t
A custom CMOS image sensor design
ricated chip includes a 320 × 240 pix
series of DACs for internal references
the chip is guaranteed through 7 sign
totypes were produced using UMC 0.1
and colour-RGB versions. Due to its h
was electrically and optically characte
tion results show state-of-the-art per
which is less than half compared to o
monochrome imager), which makes s
single chip endoluminal applications.m/locate /sna
ications
lementelb,
torib
r low power endoluminal applications is presented. The fab-
ray, a complete read-out channel, a 10-bit ADC converter, a
digital blocks for chip control. The complete functionality of
s, used for the I2C-like input and LVDS output interfaces. Pro-
-CIS (CMOS Image Sensor) technology for both monochrome
nsitivity, a pinned photodiode was implemented. The imager
and preliminary ex-vivo tests were performed. Characteriza-
ance in terms of power consumption (<40mW for the core),
e-shelf sensors, and light sensitivity (0.1 lux@555nm for the
performance comparable to CCD technology performance for
© 2010 Elsevier B.V. All rights reserved.
298 M. Vatteroni et al. / Sensors and Actuators A 162 (2010) 297–303
have motiv
technology
because it i
technology
was design
power cons
few pins. Th
high sensiti
trol interfa
320 × 240p
image quali
try payload
low power
grated in th
for full auto
like input i
(LVDS) outp
0.18m-CIFig. 1. Layout of the Vector2 sensor (a). Block diagram o
ated the development of a novel image sensor. CMOS
was chosen for single chip endoluminal applications
s simpler to control and consumes less power than CCD
[14]. The imager presented in this paper, calledVector2,
ed to improve sensitivity and at the same time reduce
umption, by using a simple control interface requiring a
e main features targeted during the design phase were
vity [15,16], low power consumption and a simple con-
ce through a reduced number of pins. A resolution of
ixelswas considered to be a good compromise between
ty, chip dimensions and frame rate, in terms of teleme-
. All the internal blocks were designed to guarantee
consumption and easy chip control. The circuitry inte-
e chip comprises the analogue anddigital blocks needed
matic control of the sensor core through a 2-pin I2C-
nterface and a 4-pin low-voltage differential signaling
ut interface. Prototypes were fabricated using the UMC
S technology and characterized in both monochrome
and colour v
device, the
formance o
with the re
tests.
2. Imager
Vector2
monochrom
UMC 0.18
application
of pinned p
photodiode
as an active
is pinned by
tional junctf the Vector2 chip (b).
ersions. In order to exhaustively describe thepresented
imager architecture is presented in Section 2,while per-
f the device is presented in Sections 3 and 4 together
sults of the electro-optical characterization and ex-vivo
architecture and operation of the camera
is a monolithic 320 × 240 active-pixel colour-RGB or
e camera-on-a-chip sensor. The chip is fabricated with
m CMOS technology which is optimized for optical
s. This process was chosen because it allows the use
hotodiode technology, implemented in the UMC ultra-
pixel [17]. A pinned photodiode has the same structure
-pixel with the addition of an extra photodiode, which
depositing a p-plus doped thin silicon layer. This addi-
ion is connected to the read-out circuit by means of an
M. Vatteroni et al. / Sensors and Actuators A 162 (2010) 297–303 299
Table 1
Vector2 imager main characteristics.
Main characteristics Dimension Value
Resolution QVGA
Active area 320 × 240
Optical format Inch 1/9
Pixel pich m2 4.4 × 4.4
Fill factor % 25
Shutter type Rolling
Die size mm2 2.5 × 3.0
extra transfer gatewhichensures separationof the sensing junction
from the read-out node. The pinned structure of the photosensitive
element shields it from the Si − SiO2 interface, which is a source of
noise and leakage, reducing the dark current and enhancing sensi-
tivity. Furthermore, the additional junction, due to the extra layer,
increases intrinsic charge storage capacitance [18]. This technology
was mainly selected because of the possibility to reach high levels
of sensitivity [19], which is one of the main specifications to be
achieved. The sensor architecture is outlined in Fig. 1. The sensor
operates in
troller. The
carries the
advantage o
pins, as out
tocol, the m
was consid
order to re
lines. An I2C
control and
because it i
application
is limited to
which need
are also ava
possibility t
integrates a
nology [21]
a rolling sh
sensor.
The mai
Pixel pitch i
out transis
quarter vid
inch. As evi
pixel array
blocks, fina
Theprimarygoal of thedesignwas toobtainhighquality images.
Therefore, to achieve this result, it is important to minimize system
noise. Sources of noise in a CMOS imager are both optical and elec-
trical. Optical noise can be reduced with an optimized layout of the
pixel and with a protective shield covering the remaining circuitry.
The shield can be made of metal layers or achieved by post pro-
cessing coating. The latter technique was used for Vector2 since
available in the standard chipset processing masks by UMC. Elec-
trical noise contribution can be classified as spatial noise, called
Fixed Pattern Noise (FPN) due to process mismatch, and temporal
noise, called Pixel temporal Noise (PN). FPN is less critical and can
be reduced by one or more signal filtering steps. This operation is
carried out in Vector2, at column level, by a column data sampling
(CDS) block which performs a first signal subtraction to reduce the
pixel FPN.A
level, by a d
FPN introdu
PN is minim
dedicated t
and fully di
The read
digit
nver
omis
requ
nd a
te fo
to a
uired
tal to
chite
pixel
ce co
read
f a r
ncy n
e res
SOut
Out =
ffer g
e thr
possi
s a re
d to
CDS
of th
ock t
Out, i
used
Out =
S blo
es a
to thconjunction with a host microcomputer or microcon-
y are connected through a serial LVDS output, which
video data to the processor. The serial output has the
f high speed transmission through a small number of
lined in the specifications above. As regards data pro-
obile industry processor interface (MIPI) standard [20]
ered. A simplified version of this protocol was used in
duce the number of external connections and control
-like interface was implemented for the low rate input
setting of the chip. The I2C-like interface was selected
s well established, simple to use and suitable for this
. Therefore, thenumberof pinsused innormal operation
7 for simple chip control, as required by applications
a small number of connections. A number of test pins
ilable to guarantee a high level of flexibility and the
o shift control complexity to external logic. The imager
pixel array based on the UMC ultra-photodiode tech-
. However, it was decided that pixel driving should have
utter read-out in order to maximize sensitivity of the
n characteristics of the imager are reported in Table 1.
s 4.4m and fill factor is 25%, due to control and read-
tors integrated in the pixel. The optical format with
eo graphics array (QVGA) resolution results in 1/9 of
dent from Fig. 1(a), total die size is dominated by the
. Due to the other required conditioning and control
l chip dimension is 2.5 × 3.0mm2.
into a
This co
compr
earity
1mV a
adequa
Due
the req
of digi
The ar
single
interfa
The
tion o
freque
and th
VSResCD
VSigCDS
The bu
and th
to the
VbCDS i
tion an
The
means
DDS bl
VResCDS
and is
VResCDS
TheDD
provid
tional
Table 2Fig. 2. Vector2 test development board.
Vector2 image
Main charac
Master clock
Data rate
Pixel rate
Data format
Power consu
Operating tesecond subtraction is thenperformed, this timeat array
ata double sampling (DDS) block, which subtracts the
ced by the mismatch between the different CDS blocks.
ized by designing a read-out channel with techniques
o obtaining low-noise, i.e. special layout adjustments
fferential blocks were used wherever possible.
-out is completed by converting the analogue signal
al signal by means of a pipeline ADC architecture [22].
ter architecture was chosen since it represents a good
e between speed, power consumption and output lin-
ired by the application. Considering a noise level below
full signal range of 1V, 10-bit resolutionwas considered
r our purposes.
strict constraint on the number of external connections,
voltage referencesweregenerated internallybymeans
analogue converters (DAC) integrated onto the chip.
cture is completed with row and column decoders for
selection and custom digital blocks for sensor core and
ntrol.
-out of the analogue pixel outputs startswith the selec-
ow, by means of a row decoder. Pixel FPN and low
oise are filtered, line by line, carrying the signal, VSigPix,
et value, VResPix, to the CDS amplifiers [23]. CDS output,
is proportional to these signals as:
[VSG + (VResPix − VSigPix) × GSFpix + VbCDS] × GSFCDS (1)
ain of the pixel and CDS, GSFpix and GSFCDS respectively,
eshold VSG are technology- and layout- dependent due
ble mismatch between different transistors. The signal
ference set by the user to avoid direct ground connec-
allow a controlled shift of the output signal.
outputs are then sequentially selected one at a time by
e column decoder, and further filtered in series by the
o remove the column FPN [24]. A CDS reference value,
s obtained by setting the CDS in the reset configuration
for this filtering.
[VSG + VbCDS] × GSFCDS (2)
ck is a fully-differential switched capacitor blockwhich
differential output, VOutDDS. These signals are propor-
e signal and reset outputs of the CDS read-out by a gain
r electrical characteristics.
teristics Dimension Value
MHz 25
MHz 100
MHz 10
Bit 10-serial
mption mW < 40 (@30 fps, 27 ◦C)
mperature C −40/ + 80
300 M. Vatteroni et al. / Sensors and Actuators A 162 (2010) 297–303
Table 3
Vector2 optical performance.
Parameter Unit Monochrome Colour-RGB
Sensitivity lux 0.11@555nm,27 ◦C, 30ms 0.32@555nm, 27 ◦C, 30ms
W/m2 1.70 × 10−4,@27 ◦C, 30ms 4.65 × 10−4@27 ◦C, 30ms
Responsivity V/lux*se ◦C 0.53@555nm, 27 ◦C 0.12@555nm, 27 ◦C
V/W/m2*s 360@27 ◦C 81@27 ◦C
Dynamic range dB 50 60
SNR dB 46 (max) 53 (max)
Pixel Temporal Noise(PN) % 0.70 0.25
Fixed Pattern Noise (FPN) % 0.86 1.67
Fig. 3. ite sen
factor, GDDS
VOutDDS = V
The GDDS ga
theuserand
switched ca
with a prec
Finally, t
by the on-c
ADCOUT =
ltage
volt
especVector2 power responsivity as function of irradiation power for the black and wh
, but also to a common mode signal,VCM.
CM ± GDDS × (VREF − (VResCDSOut − VSigCDSOut)) (3)
The vo
erence
0.3V rin is incorporated in the DDS block, programmable by
setvia I2C. Thegain is setbychanging the ratiobetween
pacitances. Available gain values are between 0 and 8
ision of 256 steps.
heDDSoutput is converted intoadigitalword,ADCOUT,
hip 10-bit pipeline ADC and then serialized.
(V+OutDDS − V−OutDDS)
(VREFP − VREFN)
× 29 (4)
All the d
the imager.
3. Imager
A test b
Vector2 chi
a dedicated
Fig. 4. Vector2 SNR as function of irradiation power for the black and white sensor andsor and the colour-RGB sensor in the three different colours.
range accepted at the ADC input is defined by two ref-
ages, VREFP and VREFN, with typical values of 1.5V and
tively.
escribed functions are related to the analogue core of
characterization
oard was designed and developed to characterize the
p (Fig. 2). The set-up is composed of a main board and
‘eye PCB’ designed for specific chip control. This set-
the colour-RGB sensor in the three different colours.
M. Vatteroni et al. / Sensors and Actuators A 162 (2010) 297–303 301
r and the colour-RGB sensor in the three different colours.
up was use
summarize
1.8V and 3.
25MHz and
cases. Comp
fully verifie
characteriz
and linearit
ation was m
with a fram
cal sensitiv
optical char
sensor.
The mai
was measur
for the colo
gration tim
555nm wav
are reporte
Figs. 3 and4
and the colo
SNR performFig. 5. Vector2 PN as function of irradiation power for the black and white sensoFig. 6. Vector2 FPN as function of irradiation power for the black and white sensor and
d for both electrical and optical characterization. As
d in Table 1 the chip is supplied with voltages equal to
3V. Typical values for the master and output clocks are
100MHz respectively, with a duty cycle of 50% in both
lete functionality of the Vector2 sensor was success-
d using the test board. Most of the internal blocks were
ed independently,mainly to check output voltage swing
y of response. Power consumption during normal oper-
easured as a crucial parameter, being less than 40mW
e rate of 30 fps. Significant parameters, such as opti-
ity, noise and dynamic range, were extracted through
acterization of both the monochrome and colour-RGB
n test results are shown in Table 2. Optical sensitivity
ed as 0.11 lux for the monochrome sensor and 0.32 lux
ur-RGB sensor. Sensitivity was measured with an inte-
e of 30ms at ambient temperature (27 ◦C) and with a
elength. Dynamic range and signal to noise ratio (SNR)
d in Table 3 as average and maximum values, and in
as functionof the lightpower forboth themonochrome
ur-RGB sensors. It is quite clear thatDynamic range and
ance are better for the colour-RGB sensor given the Fig. 7. Exthe colour-RGB sensor in the three different colours.
ample of ex-vivo image acquired with the Vector2 optical sensor.
302 M. Vatteroni et al. / Sensors and Actuators A 162 (2010) 297–303
lower responsivity. Noise numbers are also reported as a percent-
age of the average value over the full signal range and as a function
of illuminationpower (Figs. 5 and6). PN [25], consistingof temporal
noise, is higher for the monochrome sensor (0.70%), but still lower
than the lev
FPN, which
version (1.6
4. Ex-vivo
When w
consider th
ture impor
from many
important t
quality and
can be adju
Ex-vivo
the Vector2
receive prel
tion. Tests w
attached to
An example
is obtained
As expected
quality is g
can also be
processing.
Addition
era setting
5. Conclus
A CMOS
minal appli
a trade-off
power cons
developed t
cal characte
terms of po
The monoc
and 27 ◦C),
(@555nm a
devices for
tion is less t
for wireless
E x-vivo an
showing go
formances (
further exte
acterization
Samples of
miniaturize
Acknowled
The wor
Commissio
2006-03397
Jerome Par
EPFL (Ecole
support and
References
[1] R. Eliakim
terol. 24 (
[2] C. McCaffrey, O. Chevalerias, C. Mathuna, K. Twomey, Swallowable-capsule
technology, IEEE M PVC 7 (1) (2008) 23–29.
[3] R.Mackay, B. Jacobson, Endoradiosonde, Nature 179 (4572) (1957) 1239–1240.
[4] G. Iddan, G. Meron, A. Glukhovsky, P. Swain, Wireless capsule endoscopy,
Nature (2000) 405–417.
aldas
chanis
nts, IE
Moglia
ate a
–361
akam
entero
heorg
minat
ang, J
t clini
hnolo
ehara
ed Te
p://ww
p://ww
p://ww
. Cave
tout,
domiz
osc. 6
adid-
os act
97) 12
. Liu, B
ge se
nquit
8m
OS SE
heuw
Solid
21–27
uidas
ollste
ger te
–929
p://ww
p:/ww
amam
ms/s p
owler
ge Se
. Men
emb
tems,
ian, B
ive pix
heuw
Appl
phie
Vatte
l engi
rom t
iCam,
me r
ks for
he is r
stem
ferenc
age
oscop
Covi g
2001
2005
na (It
red to
for im
ce 20
optica
t syst
Cava
Camp
tudenel perceived by the human eye, which is over 5% [26].
represents spatial noise, is double in the colour-RGB
7%) compared to the monochromatic version (0.86%).
tests
orking with an endoscopic camera, it is necessary to
e non-standard settings required by the camera to cap-
tant biological information in the image, which differ
other types of imaging systems. For this reason it is
o perform tests on biological tissues, to evaluate image
to set the sensor accordingly. In particular, colour gain
sted in order to receive the best response.
images were acquired with the colour-RGB version of
imager and with non-optimized optics, in order to
iminary indications on image quality for target applica-
ere conducted on freshly excised porcine colon tissue
the test bench, which also included the test board.
of an acquired image is shown in Fig. 7. The image
by applying demosicing and background subtraction.
by the results obtained on noise performance, image
ood in terms of uniformity. Original colour rendition
considered good and can be further improved by image
al ex-vivo tests will be carried out to optimize the cam-
and to better understand imager performance.
ions and future work
image sensor specifically designed for wireless endolu-
cations was presented. As required by the application,
was identified for crucial parameters such as chip size,
umption and image quality, and a custom design was
o meet the required specifications. Electrical and opti-
rizations demonstrated that targeted requirements in
wer consumption and high sensitivity have been met.
hrome sensor has a sensitivity of 0.11 lux (@555nm
while colour-RGB imager sensitivity is equal to 0.32 lux
nd 27 ◦C). These characteristics are comparable to CCD
single chip endoluminal applications. Power consump-
han40mWinboth cases. Thismakes the sensor suitable
endoluminal applications such as capsular endoscopy.
d in-vivo tissue images were preliminarily acquired,
od image uniformity, also guaranteed by low-noise per-
PN < 0.53% and FPN < 1.67%). TheVector2 chipwill be
nsively tested toachievecompleteandsystematic char-
, focusing on the acquisition of endoluminal images.
the Vector2 imager will be integrated in a complete
d wireless vision system for capsular endoscopy.
gments
k described in this paper was funded by the European
n within the VECTOR FP6 European project (EU/IST-
0). We would like to thank Stephane Chamot and
ent from the Microvision & Microdiagnostics Group at
Polytechnique Federale de Lausanne) for their valuable
work during the optical characterization phase.
, Video capsule endoscopy of the small bowel, Curr. Opin. Gastroen-
2) (2008) 159–166.
[5] P. V
me
me
[6] A.
upd
353
[7] T. N
tro
[8] C. G
exa
[9] S. B
Firs
tec
[10] A. U
Alli
[11] htt
[12] htt
[13] htt
[14] D.R
Gos
ran
End
[15] O. Y
cm
(19
[16] X.C
ima
ogu
[17] 0.1
CM
[18] A. T
in:
pp.
[19] R. G
R. H
ima
927
[20] htt
[21] htt
[22] S. H
6.3
[23] B. F
Ima
[24] S.K
S. M
sys
[25] H. T
act
[26] A. T
and
Biogra
Monica
electrica
Physics f
for Neur
she beca
she wor
where s
vision sy
eral con
CMOS im
and end
Daniele
(Italy) in
traps. In
of Bolog
transfer
oratory
Area sin
electro-
suremen
Carmela
from the
a Ph.D. s
Pisa.tri, R. Webster, C. Quaglia, M. Quirini, A. Menciassi, P. Dario, A new
m for mesoscale legged locomotion in compliant tubular environ-
EE J RO 25 (4) (2009) 1–11.
, A. Menciassi, P. Dario, A. Cuschieri, Capsule endoscopy: progress
nd challenges ahead, Nat. Rev. Gastroenterol. Hepatol. 6 (2009)
.
ura, A. Terano, Capsule endoscopy: past, present, and future, J. Gas-
l. 43 (2) (2008) 93–99.
he, R. Iacob, I. Bancila, Olympus capsule endoscopy for small bowel
ion, J. Gastrointest. Liver Dis. 16 (2007) 309–313.
.Y. Park, S. Jeong, Y.H. Kim, H.B. Shim, T.S. Kim, D.H. Lee, S.Y. Song,
cal trial of the miro capsule endoscope by using a novel transmission
gy: electric-field propagation, Gastrointest. Endosc.
, K. Hoshina, Capsule endoscope norika system, Minim. Invas. Ther.
chnol. 12 (5) (2003) 227–234.
w.aptina.com/products/image sensors/mt9v013d00stcp/#overview.
w.ovt.com/products/part detail.php?id=78.
w.ovt.com/products/part detail.php?id=55.
, D.E. Fleischer, J.A. Leighton, D.O. Faigel, R.I. Heigh, V.K. Sharma, C.J.
E. Rajan, K. Mergener, A. Foley, M. Lee, K. Bhattacharya, A multicenter
ed comparison of the endocapsule and the pillcam sb, Gastrointest.
8 (3) (2008) 487–494.
Pecht, B. Mansoorian, E.R. Fossum, B.P. Pain, Optimization of noise in
ive pixel sensors for detection of ultra low light levels, Proc. Spie 3019
5.
. Fowler, H. Do, S.Mims, D. Laxson, B. Frymire, Highperformance cmos
nsor for low light imaging, in: International Image Sensor Workshop,
, Maine, USA, 2007, pp. 327–330.
cmos image sensor umc ultra photodiode process cell rule, G-1C-
NSOR 18-UMC/ ULTRA PD-CELL Ver. B.D P-B, UMC foundry, 2007.
issen, Cmos image sensors: State-of-the-art and future perspectives,
State Circuits Conference, 2007. ESSCIRC 2007. 33rd European, 2007,
.
h, T.-H. Lee, P. Lee, D. Sackett, C. Drowley, M. Swenson, L. Arbaugh,
in, F. Shapiro, S. Domer, A 0.6m cmos pinned photodiode colour
chnology, in: ElectronDevicesMeeting, 1997, Tech.Digest., Int. (1997)
, IEDM’97.
w.mipi.org/.
w.europractice-ic.com/technologies UMC.php?tech id=018um.
i, L.F. L, O. Yadid-Pecht, Cmos image sensor employing 3.3 v 12 bit
ipelined adc, Sens. Actuators A 135 (2007) 119–125.
, X. Liu, Charge transfer noise in image sensors, in: 2007 International
nsor Workshop, ogunquit, Maine USA, 2007, pp. 51–54.
dis, S.E. Kemeny, R.C. Gee, B. Pain, C.O. Staller, Q. Kim, E.R. Fossum,
er, Cmos active pixel image sensors for highly integrated imaging
IEEE J. Solid-State Circuits 32 (1997) 187–197.
. Fowler, A.E. Gamal, Analysis of temporal noise in cmos photodiode
el sensors, IEEE J. Solid-State Circuits 1 (2001) 36.
issen, Digital Imaging: Image Capturing, Image Sensors, Technologies
ications, CEI-Europe, 2004.
s
roniwasborn in LaSpezia, Italy, in1975. She receivedanM.S. degree in
neering from theUniversity of Pisa (Italy) in 2001 and a Ph.D. degree in
heUniversity of Trento (Italy), in 2008. From2002 to 2008, sheworked
Trento (Italy), as Pixel Engineer and analogue designer, and in 2005
esponsible for the development of CMOS image sensors. Presently,
the Scuola Superiore Sant’Anna in Pisa (Italy) as postdoctoral fellow,
esponsible for the research and development of image sensors and
s for biomedical applications. She is the author and co-author of sev-
e and journal publications and of three patents. Her interests include
sensors, low-noise analogue electronics, high dynamic range pixels
ic vision systems.
raduated in physics (summa cum laude) from the university of Trento
where he worked on the active control of magnetic fields for atomic
he received an MBA from the Alma Graduate School – University
aly). After joining Neuricam in 2000, he took part in the design and
production of CMOS optical sensors and set up the electro-optical lab-
aging sensors’ characterization. He has been head of the VLSI Design
02. He currently works as project manager in the field of advanced
l systems design. His research interests focus on optical distancemea-
ems and miniaturized camera modules for endoscopy applications.
llotti received a degree in biomedical engineering (with honours)
us Bio-MedicoUniversity in Rome inDecember 2007. She is currently
t in biorobotics at the CRIM Lab of the Scuola Superiore Sant’Anna in
M. Vatteroni et al. / Sensors and Actuators A 162 (2010) 297–303 303
Luca Clementel received a B.S. degree in communication engineering from the Uni-
versity of Trento in 2001 developing a digital neural network implemented in FPGA.
He joined Neuricam Srl, Trento, in 2001, where he designed digital architectures in
programmable logic devices for vision systems such as glue logic for demonstra-
tion baseboards of optical sensors and complex image processing algorithms. He is
currently an HDL developer and a project manager in the field of intelligent vision
systems design.
Pietro Valdastri received a degree in electronic engineering (with honours) from
the University of Pisa in February 2002. In the same year he joined the CRIM Lab
of the Scuola Superiore Sant’Anna in Pisa as a PhD student. In 2006 he obtained
a Ph.D. in bioengineering from the Scuola Superiore Sant’Anna discussing a the-
sis titled “Multi-Axial Force Sensing in Minimally Invasive Robotic Surgery”. He is
now assistant professor at CRIM Lab, with main research interests in the field of
implantable robotic systems and active capsular endoscopy. He is currently work-
ing on several European projects for the development of minimally invasive and
wireless biomedical devices.
Arianna Menciassi received a degree in physics (with honours) from the Uni-
versity of Pisa in 1995. In the same year, she joined the CRIM Lab of the Scuola
Superiore Sant’Anna in Pisa as a Ph.D. student in bioengineering with a research
programme on the micromanipulation of mechanical and biological micro objects.
In 1999, she received a Ph.D. degree discussing a thesis titled “Microfabricated Grip-
pers for Micromanipulation of Biological and Mechanical Objects”. She is currently
professor of biomedical robotics at the Scuola Superiore Sant’Anna, Pisa. Her main
research interests are in the fields of biomedical micro- and nano-robotics, micro-
fabrication technologies, micromechatronics and microsystem technologies. She is
currently working on several European projects and international projects for the
development of micro and nano-robotic systems for medical applications.
Paolo Dario received a degree in mechanical engineering from the University of
Pisa in 1977. Currently, he is professor of biomedical robotics at the Scuola Supe-
riore Sant’Anna, Pisa. He also set up and teaches the Mechatronics course at the
School of Engineering, University of Pisa. He has been a visiting professor at the
Ecole Polytechnique Federale de Lausanne (EPFL), Lausanne, Switzerland, and at
Waseda University, Tokyo, Japan. He is the director of the CRIM Lab of Scuola Supe-
riore Sant’Anna, where he supervises a team of around 70 researchers and Ph.D.
students. His main research interests are in the fields of medical robotics, mecha-
tronics andmicroengineering, and specifically in sensors and actuators for the above
applications. He is the coordinator of many national and European projects, the edi-
tor of two books on robotics and the author of over 200 journal papers. He is a
member of the Board of the International Foundation of Robotics Research. He is
an associate editor of the IEEE Transactions on Robotics and Automation, a member
of the Steering Committee of the Journal of Microelectromechanical Systems and
a guest editor of the Special Issue on Medical Robotics of the IEEE Transactions on
Robotics and Automation. He serves as president of the IEEE Robotics and Automa-
tion Society and as the co-chairman of the Technical Committee onMedical Robotics
of the same society.
Alvise Sartori received an M.A. degree in Physics from the University of Oxford in
1978 and a Ph.D. in Geophysics from Imperial College, London, in 1983. He then
joined the central research laboratory of Olivetti, where he carried out research on
modelling of fluido-dynamic systems and design of digital CMOS integrated circuits.
In 1990 he joined IRST, a Research Institute in Trento, Italy, where he was incharge
of the VLSI Design Laboratory. Since 1998, he is President and CEO of NeuriCam SpA,
Trento, a company he co-founded in 1998, active in the fabless production of chips
and systems for computer vision.