Biosensors and Bioelectronics 38 (2012) 321–330

Contents lists available at SciVerse ScienceDirect

Biosensors and Bioelectronics

0956-56

http://d

n Corr

Institut

Tel.: þ8

E-m

yaemug

journal homepage: www.elsevier.com/locate/bios

Novel implantable imaging system for enabling simultaneous multiplanarand multipoint analysis for fluorescence potentiometry in the visual cortex

Takuma Kobayashi a,c,n, Mayumi Motoyama a,c, Hiroyuki Masuda a, Yasumi Ohta a,Makito Haruta a, Toshihiko Noda a,c, Kiyotaka Sasagawa a,c, Takashi Tokuda a,c,Hideki Tamura b,c, Yasuyuki Ishikawa b,c, Sadao Shiosaka b,c, Jun Ohta a,c

a Graduate School of Materials Science, Nara Institute of Science and Technology, Ikoma, Nara 630 0192, Japanb Graduate School of Biological Sciences, Nara Institute of Science and Technology, Ikoma, Nara 630 0192, Japanc CREST, Japan Science and Technology Agency, Kawaguchi, Saitama 332 0012, Japan

a r t i c l e i n f o

Article history:

Received 19 March 2012

Received in revised form

30 May 2012

Accepted 10 June 2012Available online 29 June 2012

Keywords:

CMOS image sensor

Implantable microsensor

Brain imaging

Fluorescence potentiometry

Freely moving animal

Visual cortex

63/$ - see front matter & 2012 Elsevier B.V. A

x.doi.org/10.1016/j.bios.2012.06.035

esponding author at: Division of Molecul

e of Traditional Asian Medicine, Kinki Univer

1 72 366 6661; fax: þ81 72 366 6661.

ail addresses: ta-kobay@med.kindai.ac.jp,

ura@gmail.com (T. Kobayashi).

a b s t r a c t

Techniques for fast, noninvasive measurement of neuronal excitability within a broad area will be of

major importance for analyzing and understanding neuronal networks and animal behavior in

neuroscience field. In this research, a novel implantable imaging system for fluorescence potentiometry

was developed using a complementary metal-oxide semiconductor (CMOS) technology, and its

application to the analysis of cultured brain slices and the brain of a living mouse is described. A

CMOS image sensor, small enough to be implanted into the brain, with light-emitting diodes and an

absorbing filter was developed to enable real-time fluorescence imaging. The sensor, in conjunction

with a voltage-sensitive dye, was certainly able to visualize the potential statuses of neurons and obtain

physiological responses in both right and left visual cortex simultaneously by using multiple sensors for

the first time. This accomplished multiplanar and multipoint measurement provides multidimensional

information from different aspects. The light microsensors do not disturb the animal behavior. This

implies that the imaging system can combine functional fluorescence imaging in the brain with

behavioral experiments in a freely moving animal.

& 2012 Elsevier B.V. All rights reserved.

1. Introduction

A technique for real-time analysis of physiological neural activitywithin a freely moving animal would make a significant contribu-tion to our understanding of neural networks and modification ofbehavior. In particular, a broad range and multipoint measurement,with enough spatiotemporal resolution for analyzing the differentialpropagation of neural activity between brain areas, is highly desired.In previous studies, positron emission tomography (PET), electro-encephalography (EEG), and a small microscope of the head instal-lation type have been developed for freely moving animals (Sossiand Ruth, 2005; Schulz et al., 2011; Ishida et al., 1993; Fan et al.,2011; Helmchen et al., 2001; Park et al., 2011). These compactinstruments for functional brain measurements would contribute inrevealing neural activities, which would provide insights into animalbehavior in neuroscience. However, a new measurement technology

ll rights reserved.

ar Brain Science, Research

sity, Osaka 589 8511, Japan.

is necessary for simultaneous clarification of chemical activity,dynamic anatomical character, and physiological processes withhigh spatiotemporal resolution in deep and wide area of the brain inconjunction with animal behavior.

Optical imaging is a useful and powerful tool because it enableswide range analysis with high spatiotemporal resolution withoutdestroying the tissue. In particular, voltage-sensitive dye imaging,which describes a variation in a membrane potential as a change influorescence intensity, is able to analyze the activity of each neurondirectly in real-time (Tasaki et al., 1968; Cohen et al., 1968, 1978;Davila et al., 1973; Salzberg et al., 1973; Grinvald et al., 1986;Grinvald and Hildesheim, 2004). A tiny implantable sensor forvoltage-sensitive dye imaging fulfills these requirements.

CMOS (complementary metal-oxide semiconductor)-based multi-modal microdevices can analyze various biological activities (Ng et al.,2006, 2008; Tamura et al., 2008; Ohta et al., 2009, 2011; Kobayashiet al., 2010a, Kobayashi et al., 2011; Sasagawa et al., 2011, 2012).A previous study demonstrated a novel on-chip imaging system thatenabled potentiometric fluorescence imaging with primary culturedneurons (Kobayashi et al., 2010a). CMOS image sensor with multi-electrode array for neural tissue was able to detect a local extra-cellular potential of the brain slice as a compact instruments

T. Kobayashi et al. / Biosensors and Bioelectronics 38 (2012) 321–330322

(Nakajima et al., 2012), but the broad contact imaging with cellularresolution for the visualization of nerve activity at the tissue or thebrain has not been achieved yet. The research reported here presentsan innovative imaging system for fluorescence potentiometry usingtiny implantable sensors. This proposed imaging system enables real-time fluorescence imaging in the brain. In addition, by using multiplesensors, multiplanar and multipoint measurements can be obtainedsimultaneously, providing multidimensional information from differ-ent aspects even in a complex structure.

2. Material and methods

2.1. Development of an implantable fluorescence imaging system

As shown in Fig. 1A, a novel imaging system was constructedaccording to a previous study (Kobayashi et al., 2010a) but with freshtechnology and manifold modifications. The image sensor chip wasdesigned (width, 1.0 mm; length, 3.5 mm) and was fabricated using astandard CMOS process (0.35 mm 2-poly-4-metal CMOS; AustriaMicrosystems) (Tagawa et al., 2010). The sensor had a light-sensingpixel array, and each pixel consisted of a three-transistor CMOS activepixel sensor (APS) (Nixon et al., 1996). The total number of pixelswere 120�268, and each pixel size was 7.5 mm�7.5 mm. A red filter(high-pass liquid photoresist filter; 4600 nm, FUJIFILM arch. Inc.)was superimposed over the pixel array and around the sensor chip toprevent the penetration of the excitation light from the LEDs. The470–490 nm blue LEDs were used to efficiently excite a voltage-sensitive dye, RH795 (Molecular Probes, Inc.). The chip and blue orgreen LEDs (both blue (475 nm) and green (535 nm) bare LED chip,Epistar corp.) were assembled onto a polyimide substrate, which wasdesigned to install LEDs around the sensor chip. The sensor wascovered by a resin for waterproofing. An analog output data stream

Fig. 1. Novel implantable fluorescence imaging system for a freely moving mouse. Th

image taken by the sensor was transferred to the transit board, to the control boards, an

and were excited by the LEDs, as shown in (B). Bright field and fluorescence images we

halogen light source (E) and the fluorescence image (F) were taken by the sensor. The

from the sensor was pre-amplified and converted to 12 bit digitaldata. The digitized data was processed by laboratory-made softwareand stored to a computer, and the image was reconstructed on amonitor. To produce the dual imaging system shown in Fig. 6, smalltransit boards that the mouse could carry on its back, twin compactcontrol boards with an analog digital converter, and a coregulationsoftware, were all developed and integrated. Image data werecaptured at a frame rate of o60 fps.

2.2. Animal

B57BL/6J mice (SLC Co., Hamamatsu, Japan) were used in thepresent study. All animal procedures conformed to the animalcare and experimentation guidelines of Nara Institute of Scienceand Technology.

2.3. Whole-brain slice culture

A urethane-anesthetized mouse brain was rapidly isolated andcarefully maintained in an ice-cold Hanks’ solution (Invitrogen, Inc.).The dura was removed and the isolated brain was embedded in a 4%low melting agarose gel (Nacalai Tesque, Inc.). The brain block wassectioned transversely using a razor and the slice including the visualcortex was prepared on the membrane filter as shown in Fig. 3B.

3. Results

3.1. Evaluation of fluorescence imaging by the newly developed

imaging system

A novel real-time imaging system with multiple implantablemicrosensors for fluorescence potentiometry was developed

e fabricated imaging system is shown in (A). The information of the fluorescence

d to the personal computer (PC). The fluorescent beads were placed on the sensor

re taken using a stereomicroscope (C, D). The bright field image using an external

se images were then merged (G).

T. Kobayashi et al. / Biosensors and Bioelectronics 38 (2012) 321–330 323

(Fig. 1; see Section 2 for detail). To verify whether the sensor wasable to take a fluorescence image with cellular resolution andwaterproof, images of 15 mm fluorescent beads (F21012; Excita-tion/Emission wavelength¼559/580 nm, Molecular Probes Inc.)in PBS were taken by using either a stereomicroscope or thesensor (Fig. 1B–G). Comparing Fig. 1D with F, it can be seen thatthe sensor successfully discriminated each bead. Additionally, itwas possible to distinguish the individual fluorescent beads fromnon-fluorescent dust (Fig. 1G). These data indicate that the sensoris suitable for fluorescence imaging and gives results almost asgood as a microscope.

Next, it was verified that fluorescence signals inside the brainwere able to be detected by the sensor. Fluorescence images offluorescent beads that were implanted into the cerebral cortexusing a tungsten needle were taken by the sensor applied to thecortex (Fig. 2A–C). However, from these images, it is difficult tojudge the actual depth of the beads implanted. Depth profilinganalysis of beads using a living brain slice was then performed(see Sup. Inf. 1 for detail). In this analysis, two methods ofexcitation light irradiation were used. One was excitation byLEDs, which were installed around the entire circumference of thesensor chip (epi-illumination). The other was by an LED bar(transillumination) (Fig. 2D). The actual experimental and excita-tion methods are shown in Fig. 2E. When beads were excited byepi-illumination, their fluorescence decreased with distance fromthe sensor (Fig. 2F, upper column). When beads were excited by

Fig. 2. Depth profiling analysis of beads in a living brain tissue. 15 mm fluorescent be

washed away with PBS. A bright field (A) and fluorescence image (B) was taken. Blue ast

sensor on the cortex. (D) shows the LED bar, (E) shows a schematic diagram of the expe

(F). In the upper/lower images, the beads were excited by LEDs surrounding the sensor

(G) The result of (F) is quantified. The longitudinal axis shows the fluorescence intens

fluorescence intensity is presented as a value of 256 stages. The fluorescence intensity a

(I) are quantifications of the results of ‘‘Surround’’ and ‘‘Bar,’’ respectively. The longitudin

left from the presumed center of the spot of light. Each dot indicates the average fluore

those dots. All scale bars¼200 mm. (For interpretation of the references to color in thi

the transillumination, their fluorescence also decreased withdistance from the sensor, but these were slightly increased nearthe excitation source (Fig. 2F, lower column). In all experiments,the current applied to the LEDs was constant. Fluorescenceintensity at the center of the beads is shown in Fig. 2G, anddistributions of the fluorescence intensity of 32 pixels from thecenter of beads are shown in Fig. 2H (epi-illumination) and I(transillumination). The results shown in Fig. 2G indicate thedetection sensitivity of the sensor in relation to the depth of themeasured object. The relative brightness analysis shown inFig. 2H and I revealed that the shape of the beads increased withdistance from the sensor, and it was more significant for tran-sillumination excitation than for epi-illumination excitation.These results show the resolving power of the sensor in the brain.As each excitation method has advantages and disadvantages,both excitation methods were used for subsequent analyses.

3.2. Fluorescence spectral analysis of voltage-sensitive dyes

The fluorescence spectrum of HEK293 cells stained withvoltage-sensitive dyes, RH795 (Grinvald et al., 1988, 1994;Carlson and Coulter, 2008) and Di-3-ANEPPDHQ (Obaid et al.,2004; Fisher et al., 2008),was first assessed using a spectrumanalyzer equipped with a photodiode array (see Sup. Inf. 2 fordetail). Each fluorescence spectrum of the cells stained withRH795 or Di-3 is shown in Fig. 3A (a) and (b). When the

ads were injected into the mouse cortex, and surplus un-implanted beads were

erisks in (B) and (C) refer to the same position. (C) Fluorescence image taken by the

riment. The fluorescence imaging results taken with the sensor of (E), are shown in

(‘‘Surround’’) or the LED bar (‘‘Bar’’), respectively. Arrows indicate bead positions.

ity, and the horizontal axis shows the distance from the beads to the sensor. The

ccording to the distance from the center of the beads in (F) is distributed. (H) and

al axis shows the relative brightness of 32 pixels (1 pixel¼7.5 mm) to the right and

scence intensity of several beads, and each approximation line is depicted based in

s figure legend, the reader is referred to the web version of this article.)

Fig. 3. Fluorescence spectrum of a voltage-sensitive dye was analyzed using cultured cells, investigation of a dyeing expansion process, and fluorescence potentiometry on

a brain slice. (A) HEK293 cells stained with RH795 were excited by a green laser (a) or blue/green LED (c) and (d). The cells stained with Di-3 were excited by a green laser

(b). In each graph, a blue/red line is the spectrum of the cells before/after the depolarization, respectively. A black and magenta line indicates an approximate curve. (B) A

whole-brain slice of an adult mouse (10 weeks old) was placed on a dish, as shown in (B). RH795 was injected into a spot in agarose gel that was dug so as to not penetrate,

and cultured in an incubator for 15 min (a), 3 h (b), 12 h (c) and 24 h (d). The upper images are bright field and the lower images are the corresponding fluorescence

images, taken using a fluorescence stereomicroscope. Scale bar¼1 mm. (C) An acute brain slice stained with RH795 was analyzed using a sensor. As shown in (C) a bright

field image using an external halogen light source (a) and a fluorescence image (b) were taken by the sensor. Pseudocolored images represent the change in fluorescence

intensity for 3 min after the addition of KCl solution on the slice. (c) þDMEM, (d) þ100 mM KCl, and (e) þ1 M KCl, scale bar¼200 mm. (f) An image of the density of neural

cell body drawn based on binning of a Nissl-stained primary visual cortex image. Each number represents a layer of the structure. (For interpretation of the references to

color in this figure legend, the reader is referred to the web version of this article.)

T. Kobayashi et al. / Biosensors and Bioelectronics 38 (2012) 321–330324

RH795- or Di-3-stained cells were depolarized with a highconcentration of Kþ , the fluorescence intensity was reduced orincreased to around �16.1% or þ15.7% (dF/F) (n¼6, 1), respec-tively, at a maximum. Similarly, the fluorescence spectrum ofRH795-stained cells was analyzed by activation with a blue LED

(490 nm), and its fluorescence intensity was reduced to around�13.3% (dF/F) (n¼5) (Fig. 3A (c)). The rate of the change whengreen LED (535 nm) was used was �7.62% (dF/F) (n¼2), it waslower than blue LED (Fig. 3A (d)). It was confirmed that the LEDsexcited the RH795 dye enough for changes in the fluorescence

Fig. 4. Fluorescence potentiometry and a vascular image based on a bloodstream fluctuation using the sensor in mouse visual cortex. The visual cortex was stained with RH795

after a craniotomy in an anesthetized mouse, the sensor was placed on the brain and the LED bar inserted into the brain as shown in (A). (B) is a bright field image taken by the

sensor using an external halogen light source. Blue asterisks in (A) and (B) refer to the same position. Cortex was excited by green LED. (C) is a fluorescence image taken by the

sensor. Pseudocolored images represent the changes in fluorescence intensity for 3 min after a microinjection experiment. The cortex was excited using both the LED bar and the

omni-LED of the sensor. In (D), PBS was injected, and in (E), 500 mM KCl was injected. Black broken circles indicate the injection sites. Cortex was excited by blue LED. (F) is a

fluorescence image taken by the sensor. Vascular images were compared using different excitation methods; using only the LED bar (G), only omni-LED of the sensor (H), and

both of LED bar and omni-LED of the sensor (I). Pseudocolored images that represent the changes in the fluorescence intensity reveal a vascular pattern. There are mainly thick

blood vessels at (a) and (b) in (F). (J) shows the increase (black) and decrease (white) of the bloodstream at a certain moment. Pseudocolored images represent the changes in

fluorescence intensity for 3 min after a microinjection experiment. The cortex was excited using both the LED bar and the omni-LED of the sensor. In (K), nothing was injected, in

(L), PBS was injected, in (M), 500 mM KCl was injected, in (N and O), PBS or 500 mM KCl was injected into the unstained cortex, and in (P and Q), PBS or 500 mM KCl was injected

in the cortex stained with PKH26. White and black broken circles indicate the injection sites. In the fluorescence image taken by the sensor (R), there is a thick blood vessel in the

upper center of the image. (S) The bloodstream of that vessel is represented as the changes of pseudocolor in time-lapse images (white arrows). Scale bar¼200 mm. The images in

(C–E) or (F–M) indicate the same cortical position. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

T. Kobayashi et al. / Biosensors and Bioelectronics 38 (2012) 321–330 325

T. Kobayashi et al. / Biosensors and Bioelectronics 38 (2012) 321–330326

intensity to be detected, without being greatly inferior to thelaser, and also that the sensor could detect small changes influorescence when the cells were excited by LEDs.

3.3. Displaying a temporal diffuse staining pattern by the RH795 dye

on the cortex, and the fluorescence potentiometry of the cultured

cortex slice using the sensor

To optimize the dyeing conditions for the RH795 dye for thecortex, it was beneficial to observe a cortical slice periodicallyafter the staining procedure. In order to carry this out, whole-brain slice culture was performed. The transverse section of thebrain containing the visual cortex was cultured, and part of thesurface of the cortex was stained with RH795 as shown inFig. 3B. The 0.1 mg/ml, 1 ml RH795 dye was carefully injectedinto a gel pocket adjacent to the surface of the cortex. Becausethe gel pocket did not penetrate to the bottom, the dye dropstayed inside it for a while and then gradually sank into thecortex. RH795 emitted very little fluorescence in itself. Sequen-tial fluorescence images revealed that the region that wasstained with RH795 continued to spread gradually from thecenter of the injection point, as time advanced. Locally, allcortical layers were stained in around 15 min. Fluorescencepotentiometry was then executed by placing the sensor on theacute brain slice stained with RH795, as shown in Fig. 3C. Almostall the cortical layers were imaged by the sensor. When the KClsolution was added to the cortex, the fluorescence intensitydecrements were detected by the sensor. These data indicatethat the cortical cells were depolarized by the potassium ions,and the responses occurred in a dose-dependent manner. Com-pared to the results shown in Fig. 3C (f), several robustresponses, shown in Fig. 3C (e), appear to be caused locally atsimilar positions to the cortical neurons.

3.4. Evaluation and optimization of the excitation method through

dynamic fluorescence imaging of the vascular morphology and

fluorescence potentiometry in the visual cortex

To verify whether the developed sensor was able to measureneural cell activity in the brain of a living animal, fluorescenceimaging of the mouse visual cortex was performed using thesensor, as shown in Fig. 4. The visual cortex was stained withRH795, as shown in Fig. 3B. In the first instance, fluorescenceimaging was performed using the two previously mentionedexcitation methods, either in combination or separately (Fig. 2E).The optimum excitation conditions were investigated by visua-lizing the blood vessel. The visualization of the feeble change ofthe bloodstream associated with the nerve activity is importantfor the functional brain imaging. Continuous measurement ofchanges in fluorescence intensity from the cortex showed theappearance of several streaky lines. As the eyes of the mousewere sealed, it can be concluded that such changes in fluores-cence intensity were mainly derived from a bloodstream frac-tion, as they were too slow for neural activity, and the shapeswere unique. Though it is not easy to clearly see the blood vessel

Fig. 5. Physiological optic analysis with an electrophysiologic test and fluorescent pote

asterisks in (D and E) refer to the same position. A tungsten needle was inserted into t

recorded electrophysiologically (B) when the right eye was stimulated with a light (red

intermittent light (red broken line in (C)). Fluorescence potentiometry using the senso

placed in position as shown in (A), and the changes in fluorescence intensity were me

stimulated by intermittent light in a similar way to (C). Each pseudocolored image rep

and E) refer to the same position. Scale bar¼200 mm. (F) Twenty four trials of the tota

(C) were averaged every 10 ms and added together, and these are indicated as a relative

stimulation or non-stimulation in a similar manner to (D and E) were averaged and are

(For interpretation of the references to color in this figure legend, the reader is referre

in the image shown in Fig. 4F (a) and (b), the change offluorescence intensity associated with the blood flow enabledthe vessel to be visualized as a pseudo colored image (Fig. 4G–J).Such a fluorescence change was probably caused by the absor-bance of hemoglobin. Many deeper blood vessels, including thatshown in Fig. 4F (a), were only detected by using the LED bar,although the shape became slightly blurred (Fig. 4G). Mean-while, the clear shape of the blood vessels could only be detectedby using LEDs on the sensor; however, the deep vessels could notbe seen (Fig. 4H). When the cortex was excited by both the LEDbar and the LEDs of the sensor, the relatively deep blood vesselscould be detected with a clear visible shape. Additionally, boththe entry and efflux of the blood flow can be simultaneouslydistinguished and represented as a white and black coloredvessel, as seen in Fig. 4J. Furthermore, the data analysis focusedon the bloodstream enabled visualization of fluorescence fluc-tuation depending on the blood flow (Fig. 4R). Using thesedetermined optimum excitation conditions, fluorescence poten-tiometry with a chemical stimulation was performed. Thechange in fluorescence intensity is shown as pseudocoloredimages (Fig. 4K–M) three times later after injecting nothing(control), PBS, and KCl into the cortex, respectively. The micro-injection was performed using a glass microcapillary. The PBS orKCl solutions were injected into the cortex at a rate of 1 ml/10 s,at the side of the sensor (Fig. 4A). Though a subtle reaction wasdetected at the injection point of PBS, a more intense reactionwas detected at the injection point of KCl. Such an intensereaction was not observed on the cortex, which was not stainedwith RH795 (Fig. 4N and O) or PKH26 (sham) (Fig. 4P and Q).These data indicate that the potential changes of the corticalneurons that were stimulated and depolarized could be success-fully visualized by the sensor. Because the neural activity wasstrongly measured using blue LED than green LED (Fig. 4E andM), blue LED was used in the later experiments.

3.5. Physiological analysis in the visual cortex using optical

potentiometry and electrophysiology

Before verifying whether the physiological response could bevisualized by the sensor, the local field potentials in the visualcortex, caused by light stimulations in the eye, were measuredusing electrophysiology. A fine tungsten needle was inserted3.5 mm posterior and 2.3 mm lateral to the bregma and 2.0 mmdeep into the skull surface in the left hemisphere. The left eye wasthen occluded and the right eye was dilated with mydriatics andstimulated by light. The local field potentials associated with asingle light stimulation are shown in Fig. 5B. Intermittent lightstimulations were applied during 1 min intervals by blinking for aduration of 2 s, and the responses were recorded (Fig. 5C). Themeasured potential values, with or without light stimulation,were averaged, and the relative values are shown in Fig. 5F. It ispossible to distinguish between the activities due to the lightstimulation from those that are independent activities. Fluores-cence potentiometry was performed continuously. The center ofthe sensor was placed at a location of 3.0–3.5 mm posterior and

ntiometry. (A) shows the experimental scheme. The blue asterisk in (A) and white

he left primary visual cortex of the mouse, and the local field potential (LFP) was

line in (B)). The LFPs are shown in (C) when the right eye was stimulated with an

r was performed on the mouse visual cortex stained with RH795. The sensor was

asured under the dura (D) or above the dura (E). The right eye of the mouse was

resents the changes in the fluorescence intensity for 38 ms. Blue asterisks in (A, D

l potential changes during the stimulation or non-stimulation in a similar way to

negative value on the graph. Six trials of the total fluorescence changes during the

indicated as a relative value on the graph. Error bars represent standard deviation.

d to the web version of this article.)

T. Kobayashi et al. / Biosensors and Bioelectronics 38 (2012) 321–330 327

Fig. 6. Fluorescent potentiometry in the visual cortex of both hemicephalon using an ambicerebral fluorescence imaging system for a freely moving mouse. Using a newly

developed dual imaging system, voltage-sensitive dye imaging was performed. An application image of two sensors for both visual cortices is indicated (A). The mouse

wears a jacket to hold the connector of the sensor (B). A schematic image is shown in (C). All images were taken by the sensor (D–G). The images that were taken of the left

brain are shown in the left side, and the images that were taken of the right brain are shown in the right side. Bright field images were taken by the sensor using an

external halogen light source (D). Fluorescence images of the fluorescent beads implanted into the cortex were taken by the sensor (E). In the freely moving mouse, as

shown in (B), fluorescence images of cortices stained with RH795 are shown in (F) as a pseudocolor. When the right brain was stimulated by microinjection of KCl solution

after the cortices were stained with RH795, as shown in (C), the changes in fluorescence intensity for 20 s after the stimulation are shown in (G) as a pseudocolor. Blue

asterisks in (C) and (D) refer to the same position. Black broken circles indicate the injection sites. Scale bar¼200 mm. (For interpretation of the references to color in this

figure legend, the reader is referred to the web version of this article.)

T. Kobayashi et al. / Biosensors and Bioelectronics 38 (2012) 321–330328

2.0–2.3 mm lateral to the bregma and was turned 451 from thelateral axis (Fig. 5A). Light stimulation was applied in a similarmanner, and the results from the sensor being placed on thecortex, under or over the dura are shown in Fig. 5D and E,respectively. It is suggested that the pseudocolored imagesrepresent the burst depending on the light stimulations thatwas caused at the visual cortex. Though the signals from thesensor placed over the dura were slightly weaker and moreblurred than the signals from the sensor placed under the dura,it is possible to distinguish the response of the period of ‘‘on’’from ‘‘off’’. Also, judging from the quantitative analysis (Fig. 5F),which summarizes the fluorescence intensities of each of theimages in Fig. 5D and E, the response due to the light stimulationwas able to be detected by the sensor.

3.6. Ambicerebral fluorescence imaging system in a freely moving

animal

Optical imaging in the brain of free-moving animal, withbroad scope, and from many viewpoints pleiotropically, wasrequired for the analysis of the propagation over multiple brainareas. It is especially important to analyze the activities of bothhemicephalon at the same time. These requirements encour-aged us to develop a novel multiple imaging system, with twoimplantable sensors applied to both the visual cortices stainedwith RH795, as shown in Fig. 6A–C. The dual imaging systemthat can visualize both left and right visual areas at the sametime (Fig. 6D) can also preserve the fluorescence image withcellular resolution (Fig. 6E). In a freely moving mouse, the

fluorescence image could be taken by the sensor without anoise (Fig. 6F). However, the fluorescence values varied in thebrain without anesthesia. This suggests that the brain persis-tently increases baseline noise, because the physical movementof the mouse was not generally related to the level of noise.Thus, images of the fluorescence variations when the rightvisual cortex was stimulated with KCl injection, with a localaveraging of 2�2 pixels, similar to a binning process, areshown in Fig. 6G. The results indicate that the differentialresponse of the cortex could be successfully visualized by thisnovel dual imaging system.

4. Discussion

Recently, the activity of the brain is mapped and its function isgoing to be clarified by non-invasive wide-field imaging techni-ques; such as fMRI (Miyawaki et al., 2008), fOCT (Maheswari,et al., 2003; Rajagopalan and Tanifuji, 2007), and a photoacousticimaging (Liao et al., 2010; Zhang et al., 2011). These techniqueshave advantages of noninvasiveness and penetrability, while theyhave disadvantages of less portability and indirect measurementof nerve activity. On the other hand, the implantable imagingtechnique needs a less-invasive craniotomy, but has highly port-ability and can directly analyze neural activity. Although manystudies have been conducted on optical imaging of the animalbrain with voltage-sensitive dyes, the work described here hasmoved this area of research forward by, for the first time, using animplantable sensor. Of particular note, the plurality of the sensor

T. Kobayashi et al. / Biosensors and Bioelectronics 38 (2012) 321–330 329

permits the imaging of physiological activity in a freely movinganimal at multiple points simultaneously. In this respect, implan-table imaging system has a number of advantages over previousimaging systems.

This research examined two distinct excitation methods.One involved using an LED bar implanted within the brain, andthe other was the application of LEDs on the sensor. Theformer method was able to activate deeper parts of the tissue,but with a spatial resolution inferior to the latter (Fig. 2E–I).Indeed, both excitation methods provided vivid images in thehemodynamic analysis (Fig. 4G–I). It is possible that the signalshown in Fig. 4M was derived from almost the entire corticallayer, with a thickness of 800 mm, taking into account theresults of the studies with fluorescent beads shown in Fig. 2.However, it should be taken into account that the thickness ofthe cortical layer fluctuates depending on the cortical area andage of the mouse. It is clearly preferable for the brain toremain intact for accurate physiological analysis, if at allpossible. In Fig. 5, and 6, the signal could be detected eventhrough the dura. In this case, the signals were probablyderived from the area above cortical layer IV, again takinginto account the results of the fluorescent bead studies (Fig. 2)that show that the sensor can acquire signals from a depth ofapproximately 400–500 mm. In conclusion, it was shown thatthe most appropriate excitation method could be chosendepending on the specific imaging purpose.

The fluorescence of cells stained with RH795 dye representedthe potential status of the cell membrane, for example, a restingmembrane potential. Fig. 3C (b) and (e) show the same slice,which demonstrates that there are cells present in the areasgiving a weak signal in Fig. 3C (e). The signal pattern in Fig. 3C(e) corresponds approximately to the binned image of the Nisslstaining in Fig. 3C (f), which represents only neural cell bodies.These results indicate that Fig. 3C (f) mainly represents neuronalactivity, suggesting that voltage-sensitive dye imaging using thenovel system accurately revealed the neural activity of the braintissue. It was ascertained by trypan blue staining that the cells ofthe experimental brain slices were alive, even after culturing,staining with RH795, and administration of KCl (data not shown).Fig. 4N, and O show that no response can be recorded forunstained tissue, and so this rules out any contribution to theoptical imaging data shown in Fig. 5. In addition, the response of38 ms is too fast for intrinsic signals derived from the blood-stream or a flavoprotein, and the signal is too strong for adiscrimination between changes in oxy and deoxyhemoglobinconcentration of the blood (Kalatsky and Stryker, 2003; Goetzeet al., 2010; Bouchard et al., 2009; Tohmi et al., 2006; Hishidaet al., 2011). Accordingly, it is concluded that fluorescencepotentiometry of cortical neurons was convincingly demon-strated using the implantable imaging system.

In an experiment similar to the one represented in Fig. 4M,two areas of high signal intensity were often observed, not only atthe point where KCl was injected but also at an additional pointwhere KCl was not injected. The reaction that was caused by KClin the primary visual cortex (lower island in Fig. 4M) probablypropagated to the secondary visual cortical area (upper island inFig. 4M). Two similar islands were observed in Fig. 5D, and E. Ineach sequential image, the lower/upper islands are assumed to bethe response of the primary/secondary visual cortex, respectively,because the mouse was irradiated by one point light source(Antonini et al., 1999; Tagawa et al., 2005).

The functional fluorescence imaging system using implantablesensors presented in this paper, enables ambulatory intravitalimaging of an animal, which is extremely difficult using conven-tional technology. Simultaneous interior physiological imagingby multiple sensors produces an unprecedented, extensive,

and multipoint analysis of an entire brain. This device makes itpossible to analyze the physiology and brain function of a freelymoving animal in combination with genetic studies and variousbehavioral experiments. Such an integral analysis for the right andleft brain would also contribute to analyzing the cause of aninterhemispheric asymmetry and associated disease. Additionally,the small scale of the sensor presents the opportunity to investi-gate a local internal reaction of tissue cultured in vitro, such as anorgan cultured in a three-dimensional gel or an embryo in utero orin ovo (Kobayashi et al., 2009, 2010b). Furthermore, the compactand lightweight instrumentation developed in this study wouldcontribute to the field of aerospace medicine and bioastronauticsas it overcomes the several problems associated with space andtransportation weight.

5. Conclusions

In this paper, a novel implantable CMOS image sensor systemfor voltage-sensitive dye imaging was developed, and actually,the neural activity was successfully visualized by using it. Easyorgan culture for the whole-brain slice was newly established andits physiological activity was visualized by the contact imagingfor the first time. A novel dual implantable imaging system forfreely moving animal was developed, and actually, the neuralactivity of both visual cortex was successfully visualized in awakemouse in real-time.

Acknowledgments

This work was supported by the Core Research for EvolutionalScience and Technology (CREST) from Japan Science and Technol-ogy Agency (JST), and the Grant-in-Aid for Scientific Research(23246068) and the Young Scientists (22800044, 24700411) fromthe Japan Society for the Promotion of Science (JSPS).

Appendix A. Supporting information

Supplementary data associated with this article can be foundin the online version at http://dx.doi.org/10.1016/j.bios.2012.06.035.

References

Antonini, A., Fagiolini, M., Stryker, M.P., 1999. Journal of Neuroscience 19,4388–4406.

Bouchard, M.B., Chen, B.R., Burgess, S.A., Hillman, E.M., 2009. Optics Express 17,15670–15678.

Carlson, G.C., Coulter, D.A., 2008. Nature Protocols 3, 249–255.Cohen, L.B., Keynes, R.D., Hille, B., 1968. Nature 218, 438–441.Cohen, L.B., Salzberg, B.M., Grinvald, A., 1978. Annual Review of Neuroscience 1,

171–182.Davila, H.V., Salzberg, B.M., Cohen, L.B., Waggoner, A.S., 1973. Nat. New Biologist

241, 159–160.Fan, D., Rich, D., Holtzman, T., Ruther, P., Dalley, J.W., Lopez, A., Rossi, M.A., Barter,

J.W., Salas-Meza, D., Herwik, S., Holzhammer, T., Morizio, J., Yin, H.H., 2011.PLoS One 6, e22033.

Fisher, J.A., Barchi, J.R., Welle, C.G., Kim, G.H., Kosterin, P., Obaid, A.L., Yodh, A.G.,Contreras, D., Salzberg, B.M., 2008. Journal of Neurophysiology 99, 1545–1553.

Goetze, B., Schmidt, K.F., Lehmann, K., Altrock, W.D., Gundelfinger, E.D., Lowel, S.,2010. NeuroImage 49, 1622–1631.

Grinvald, A., Frostig, R.D., Lieke, E., Hildesheim, R., 1988. Physiological Reviews 68,1285–1366.

Grinvald, A., Hildesheim, R., 2004. Nature Reviews Neuroscience 5, 874–885.Grinvald, A., Lieke, E., Frostig, R.D., Gilbert, C.D., Wiesel, T.N., 1986. Nature 324,

361–364.Grinvald, A., Lieke, E.E., Frostig, R.D., Hildesheim, R., 1994. Journal of Neuroscience

14, 2545–2568.Helmchen, F., Fee, M.S., Tank, D.W., Denk, W., 2001. Neuron 31, 903–912.

T. Kobayashi et al. / Biosensors and Bioelectronics 38 (2012) 321–330330

Hishida, R., Watanabe, K., Kudoh, M., Shibuki, K., 2011. Journal of NeuroscienceMethods 201, 315–321.

Ishida, N., Kasamo, K., Nakamoto, Y., Suzuki, J., 1993. Brain Research 608, 52–57.Kalatsky, V.A., Stryker, M.P., 2003. Neuron 38, 529–545.Kobayashi, T., Tagawa, A., Noda, T., Sasagawa, K., Tokuda, T., Hatanaka, Y., Tamura,

H., Ishikawa, Y., Shiosaka, S., Ohta, J., 2010a. Japanese Journal of AppliedPhysics 49, 117001.

Kobayashi, T., Tamura, H., Hatanaka, H., Motoyama, M., Noda, T., Sasagawa, K.,Tokuda, T., Ishikawa, Y., Shiosaka, S., Ohta, J., 2011. Proceedings of IEEE BioCAS,

110–113.Kobayashi, T., Yasuda, K., Araki, M., 2009. Development, Growth and Differentia-

tion 51, 723–733.Kobayashi, T., Yasuda, K., Araki, M., 2010b. Development, Growth and Differentia-

tion 52, 351–363.Liao, L.D., Li, M.L., Lai, H.Y., Shih, Y.Y., Lo, Y.C., Tsang, S., Chao, P.C., Lin, C.T., Jaw, F.S.,

Chen, Y.Y., 2010. NeuroImage 52, 562–570.Maheswari, R.U., Takaoka, H., Kadono, H., Homma, R., Tanifuji, M., 2003. Journal of

Neuroscience Methods 124, 83–92.Miyawaki, Y., Uchida, H., Yamashita, O., Sato, M.A., Morito, Y., Tanabe, H.C., Sadato,

N., Kamitani, Y., 2008. Neuron 60, 915–929.Nakajima, A., Kimura, H., Sawadsaringkarn, Y., Maezawa, Y., Kobayashi, T., Noda, T.,

Sasagawa, K., Tokuda, T., Ishikawa, Y., Shiosaka, S., Ohta, J., 2012. Optics

Express 20, 6097–6108.Ng, D.C., Tamura, H., Mizuno, T., Tokuda, T., Nunoshita, M., Ishikawa, Y., Shiosaka,

S., Ohta, J., 2008. Sensors and Actuators A 145–146, 176–186.Ng, D.C., Tamura, H., Tokuda, T., Yamamoto, A., Matsuo, M., Nunoshita, M.,

Ishikawa, Y., Shiosaka, S., Ohta, J., 2006. Journal of Neuroscience Methods156, 23–30.

Nixon, R.H., Kemeny, S.E., Pain, B., Staller, C.O., Fossum, E.R., 1996. IEEE Journal ofSolid-State Circuits 31, 2046–2050.

Obaid, A.L., Loew, L.M., Wuskell, J.P., Salzberg, B.M., 2004. Journal of NeuroscienceMethods 134, 179–190.

Ohta, J., Kobayashi, T., Noda, T., Sasagawa, K., Tokuda, T., 2011. IEICE Transactionson Communications E94.B, 2454–2460.

Ohta, J., Tokuda, T., Sasagawa, K., Noda, T., 2009. Sensors (Basel) 9, 9073–9093.Park, J.H., Platisa, J., Verhagen, J.V., Gautam, S.H., Osman, A., Kim, D., Pieribone, V.A.,

Culurciello, E., 2011. Journal of Neuroscience Methods 201, 290–295.Rajagopalan, U.M., Tanifuji, M., 2007. Optics Letters 32, 2614–2616.Salzberg, B.M., Davila, H.V., Cohen, L.B., 1973. Nature 246, 508–509.Sasagawa, K., Ando, K., Kobayashi, T., Noda, T., Tokuda, T., Kim, S., Iino, R., Noji, H.,

Ohta, J., 2012. Japanese Journal of Applied Physics 51, 02BL01.Sasagawa, K., Matsuda, T., Davis, P., Zhang, B., Li, K., Kobayashi, T., Noda, T., Tokuda,

T., Ohta, J., 2011. In: Conference Proceedings of the IEEE Engineering inMedicine and Biological Society, pp. 2917–2920.

Schulz, D., Southekal, S., Junnarkar, S.S., Pratte, J.F., Purschke, M.L., Stoll, S.P.,Ravindranath, B., Maramraju, S.H., Krishnamoorthy, S., Henn, F.A., O’Connor, P.,Woody, C.L., Schlyer, D.J., Vaska, P., 2011. Nature Methods 8, 347–352.

Sossi, V., Ruth, T.J., 2005. Journal of Neural Transmission 112, 319–330.Tagawa, A., Minami, H., Mitani, M., Noda, T., Sasagawa, K., Tokuda, T., Tamura, H.,

Hatanaka, H., Ishikawa, Y., Shiosaka, S., Ohta, J., 2010. Japanese Journal ofApplied Physics 49, 01AG02.

Tagawa, Y., Kanold, P.O., Majdan, M., Shatz, C.J., 2005. Nature Neuroscience 8,380–388.

Tamura, H., Ng, D.C., Tokuda, T., Naoki, H., Nakagawa, T., Mizuno, T., Hatanaka, Y.,Ishikawa, Y., Ohta, J., Shiosaka, S., 2008. Journal of Neuroscience Methods 173,114–120.

Tasaki, I., Watanabe, A., Sandlin, R., Carnay, L., 1968. Proceedings of the NationalAcademy of Sciences of the United States of America 61, 883–888.

Tohmi, M., Kitaura, H., Komagata, S., Kudoh, M., Shibuki, K., 2006. Journal ofNeuroscience 26, 11775–11785.

Zhang, Y., Hong, H., Cai, W., 2011. Cold Spring Harbor Protocols, 1015–1025.