Development of an optical system for combined imaging and patterned stimulation of the rodentretina

Development of an optical system for combined imaging and patterned stimulation of the rodent

retina

Application to an optogenetic therapy for vision restoration

PAUL DELROT

Stockholm, Sweden 2013 Master of Science Thesis School of Engineering Physics

Royal Institute of Technology

TRITA-FYS 2013:69 ISSN 0280-316X ISRN KTH/FYS/–13:69–SE

i

ii Abstract

During this Master’s thesis at the Vision Institute (Paris, France), a non- contact optical system for in vivo combined imaging and patterned stim- ulation of the rodent eye was developed. The objective was to assess the efficiency of the optogenetic therapy, currently tested on rodents, whose aim is to heal blind or low-visual-acuity people.

To the best of our knowledge, previous systems did not reach the photo- metric requirement needed for the stimulation of optogenetic channels and were developed for bigger mammals than rodents. The mounted system al- lowed us to obtain non-contact mouse and rat eye fundi images with a wide field in both near-infrared and visible illumination. The first tests of the patterned stimulation arm allowed projection of simple patterns on the rat eye and showed that the photometric requirements are almost met for the activation of optogenetic channels. Future developments should include sta- ble alignment of the animal’s eye with the optical axis of the system, which would result in an improved quality of eye fundus images and of the patterned stimulation.

Supervisor:

Émilie Macé, PhD Postdoctoral Researcher, The Vision Institute,

Serge Picaud’s Team "Retinal information processing : pharmacology and pathology".

Examiner:

Linda Lundström, PhD Associate Professor,

Royal Institute of Technology / KTH, AlbaNova Biomedical and X-Ray Physics.

Commissioner:

The Vision Institute,

Serge Picaud’s team "Retinal information processing : pharmacology and pathology"

Paris, France.

Contents

1 Introduction 1

1.1 The eye anatomy . . . . 1

1.2 Eye diseases . . . . 2

1.3 Innovative therapies . . . . 3

1.4 Measurement techniques . . . . 4

1.5 Outline . . . . 5

2 Methods and materials 6 2.1 Optical design . . . . 6

2.1.1 Previous works . . . . 6

2.1.2 Materials . . . . 8

2.1.3 Initial scheme and planning of the optical design . . . . 8

2.1.4 The illumination arm . . . . 9

2.1.5 The imaging arm . . . . 12

2.1.6 The stimulation arm . . . . 13

2.2 Synchronizing the acquisition and stimulation . . . . 14

2.3 Mounting and optimizing the optical system . . . . 14

2.3.1 Mounting and aligning the system . . . . 14

2.3.2 Experimental protocol . . . . 15

2.3.3 Optimizing the system . . . . 15

2.3.4 Final system . . . . 17

3 Results and discussion 18 3.1 Illuminating and imaging . . . . 18

3.2 Stimulating . . . . 19

3.3 Future developments . . . . 21

4 Conclusion 23

5 Acknowledgements 24

A Aligning an afocal system 25

iii

CONTENTS iv

B Additional calculations 27

B.1 Choosing the collimation lens . . . . 27

B.2 Additional pictures . . . . 28

Chapter 1

Introduction

The optogenetic therapy developed at the Vision Institute, Paris, France, aims at restor- ing the vision of blind or low-visual-acuity people; this innovative therapy is currently tested on rodents. The way the therapy spreads in the rodent eye and its photometric characteristics require to develop an optical system able to assess the functional state of specific zones of the rodent retina in vivo , and this was the aim of this Master’s thesis. Previous systems only allowed to achieve this patterned functional imaging on bigger mammals and they did not reach the photometric requirements needed for the optogenetic therapy.

The context of this thesis is briefly introduced in Chapter 1, while the methods and materials needed to achieve the performances of the optical system are described in Chapter 2. In Chapter 3, the main results are shown and discussed, which allows us to draw the conclusions summarized in Chapter 4. Further details on the methods to align the optical system and on the optical design are given in Appendix A and B respectively.

1.1 Comparison of the human eye and the rodent eye anatomy

The eye is the organ of sight [1]. In a normal eye, light reflected or scattered by objects, or even emitted by sources is focused onto the retina. The retinal photoreceptors convert the luminous signals into neuronal electrical signals, which are transmitted to the visual cortex of the brain and interpreted.

In spite of a few anatomical differences (see Fig.1.1) between the human eye and the rodent eye, their function is similar. The incoming light is first focused by the cornea, a spherical transparent layer that accounts for the two-third of the human eye’s optical power. The iris, the variable pupil of the eye, limits the light flux entering the eye thus preventing saturation of the photoreceptors. The light is then focused by the lens onto

1

CHAPTER 1. INTRODUCTION 2 the retina.

In the human eye, the lens shape can be changed by the ciliary muscles in order to accommodate, whereas the rodents are believed to be unable to accommodate due to an atrophy of their ciliary muscles [2, 3]. Furthermore, the much bigger size of the lens in the rodent eye as compared to that of the human eye results in an optical power larger than expected for the rodent eye’s length. Hence rodents have a poor visual acuity [4]

.

As mammals, rodents are yet good models to study the retinal disorders affecting humans.

The retinal photoreceptors, cones and rods, are present in rodent retina as well as in human retina and the neural circuitry to the visual cortex is similar.

Figure 1.1: (a) The human eye anatomy, adapted from [1], (b) The rodent eye anatomy, adapted from [5], eyes dimensions from [6]

1.2 Eye diseases

Blindness may arise from several different pathologies, among which is retinitis pigmen- tosa.

Retinitis pigmentosa (RP) is an inherited eye disease affecting two million people in the

world [7]. RP leads to the progressive loss, generally in a few decades, of the sight. The

rods (the photoreceptors used for low-intensity vision) first degenerate, resulting in night

blindness, which can be difficult to become aware of. The degeneration of the peripheral

cones (the daylight-vision photoreceptors) follows this step and causes a tunnel vision in

daylight conditions, the degeneration may then progressively extend to the central part of

the retina and finally leads to incurable blindness. The best way to diagnose this disease

is the recording of an electroretinogram (ERG), a measurement of the electrical activity of

the photoreceptors. Since each cell layer of the retina produces a given component on an

CHAPTER 1. INTRODUCTION 3 ERG, the disappearance or the decrease in the amplitude of a component can be the sign of mid stages of RP. The only possible therapies (sunlight protection, vitaminotherapy) may slow down the progression of the disease [7] but currently the degenerative process cannot be stopped.

Other diseases, such as Age-related Macular Degeneration (AMD) may cause blindness or a decrease of the visual acuity. AMD generally affects people aged of more than 50 years old and conducts to a progressive loss of the central vision while the macula, the part of the retina with the maximal density of cones, degenerates [8]. Since this retinal atrophy is related to the lifelong accumulation of cellular defects, the aging trends in the Western countries are expected to give rise to an increase in the number of AMD- affected persons. However, the current therapies can only slow down and sometimes stop the progress of AMD and no therapies can restore vision.

1.3 Innovative therapies for vision restoration

The Vision Institute (Paris, France) is specialized in the study of eye diseases. More specifically, S. Picaud’s team, in which this Master’s thesis was conducted, is investigat- ing new therapies to restore vision of blind or partially-sighted people. Given the trends described above, the team focuses on retinitis pigmentosa and age-related macular de- generation. The two most promising axis of research are based on the work described in [9] on artificial optoelectronic retinal prosthesis and based on optogenetic therapy [10].

Artificial retinal prosthesis (see Fig.1.2a) are electronic chips using an array of electrodes to stimulate the remaining neural circuitry of the retina. These devices are supposed to function coupled with a pair of projection goggles worn by the implanted patient. The goggles are furnished with a camera that acquires images, and processed images are then projected onto the sub-retinally or epi-retinally implanted chips. The current number of electrodes per chip and the limited transmission rate of the camera result in a poor resolution and a narrow field of view for the implanted patient but preliminary results are very promising [11].

Optogenetics consist in the transfection of a photosensitive protein to a cell. The cell

then expresses the protein, which can be an ionic channel for instance. In the case of

optogenetic therapy for vision restoration, when excited with a light of convenient power

and spectrum, the ionic channel hyperpolarizes or depolarizes the cell of the retina it has

been transfected to. This excitation creates an action potential, an electric signal, which

can be transmitted up to the visual cortex if the downstream neural circuitry is still func-

tional. Current research focuses on the use of the ionic channels channelrhodopsins (see

Fig.1.2b the transfected protein has an attached GFP to localize transfected channels),

whose excitation peak is in the blue part of the visible spectrum, and halorhodopsins,

whose excitation peak is in the green or yellow part of the spectrum. Blind mice suf-

CHAPTER 1. INTRODUCTION 4 fering from retinitis pigmentosa could see anew when transfected with halorhodopsin in their light-insensitive cones [10]. The major drawbacks of this method are the high illu- minance threshold needed to activate the ionic channels (10

photons.cm

.s

that is 3.4mW.cm

) and the high transfection efficiency required to restore vision.

As a consequence of the high power threshold and spectral dependency of the optoge- netic therapy and of the retinal implants (these latters react to pulsed infrared light), a stimulation system able to send high power light with convenient wavelengths on the retina to activate the different therapies is needed.

1.4 Current means of assessing the efficiency of therapies

The efficiency of the newly elaborated therapies must obviously be assessed. Regarding the tests of retinal therapies on mammals, the typical measurement techniques are elec- troretinography (ERG), optical coherence tomography (OCT), cortical imaging and eye fundus.

ERG, whose principles are explained in section 1.2 is a measurement of the electrical activity of the retinal layers. To measure the retinal potentials, an electrode placed on the cornea integrates the field potentials induced by retinal activity. ERG is therefore a contact method, which requires a great stability of the system, and ERG cannot measure the activity of precise regions of the retina.

OCT is an in-depth imaging technique using the coherence properties of light. It can probe different layers of the retina. Using OCT one can produce 3D images of the retinal

(a) An epi-retinal implant on a rodent (b) Optogenetic therapy on a rodent

Figure 1.2: Different therapies for vision restoration. Image courtesy of Émilie Macé,

The Vision Institute

CHAPTER 1. INTRODUCTION 5 layers and then detect morphological abnormalities linked to the early stages of retinal disorders. In spite of intensive research in that field [12], full-field functional OCT is not yet a standard imaging procedure and is mostly used for anatomical measurements.

Cortical imaging is a functional imaging technique based on the absorption of light by the hemoglobin, which allows to quantify the blood volume change and therefore to detect the hemodynamic response concurrent to neuronal activity, notably in the visual cortex.

An eye fundus is an imaging technique based on the absorption and reflection of light by the different layers of the retina. The light reflected by the retina is then imaged onto a detector. This technique allows to observe the progress of retinitis pigmentosa, since, as its name indicates, the parts of the retina affected by mid-stage of RP appear yellow/white due to depigmentation and dark for late-stage RP.

Using an infrared light (IR) for the illumination, the physiological state of the retina [13] can be assessed with the eye fundus technique. This functional imaging is based on the differential absorption of IR by the retinal layers during stimulation (see Fig. 2.1).

Hence, if a precise zone of the retina is stimulated with visible light and if the eye fundi are recorded with infrared illumination, functional imaging of specific zones of the retina is possible.

1.5 Outline

We have now stressed the utility and the need of an optical device that could assess the functional state of specific regions of the rodent retina, especially when developing new therapies for eye diseases, such as the optogenetic therapy. It must be emphasized that the patterned feature of the optical system is crucial when it comes to heal diseases progressively affecting precise areas of the retina through targeting precise areas of the retina with viral vectors.

This Master’s thesis is encompassed in a wider project whose goal is to probe the physi- ology of the whole circuitry of sight: from the upstream functional imaging of the retina to the downstream functional imaging of the visual cortex. The initial idea is to develop an optical system combining both patterned stimulation and conventional imaging, and then to couple it with an already operational cortical imaging setup in order to per- form accurate cortical retinotopies. The second step, once the imaging system will be stable, would be to perform IR functional imaging of the retina combined to patterned stimulation, and the whole optical system would be once again coupled to the cortical imaging. One could therefore assess the functional state of sight from one end of the visual circuitry to the other end.

In the following chapter, the detailed methods for the implementation of the optical

system are described, the results are discussed afterwards in Chapter 3.

Chapter 2

Methods and materials

This chapter describes the main steps of the methods allowing the development and optimization of an optical system performing non-contact patterned stimulation and imaging on rats and mice eyes.

2.1 Optical design

2.1.1 Previous works

The realization of the optical setup rests upon previous scientific publications. Zhu et al. [14] demonstrated the utility of a Digital Micro-mirror Device (DMD) for the stimulation of specific areas of the retina. This device is made up of a matrix of micro- mirrors, each of them being switchable between an ON and OFF angle, where an ON angle means that the incident light on the mirror is reflected towards the stimulation path of the optical system, an OFF angle sends the incident light into a beam dump.

High-definition patterns can therefore be projected onto the retina with a DMD. The utility of this device compared to a LCD projector, as was used in [13] for stimulation, resides in its photometric features. A LCD projector displays a limited intensity whereas the DMD can reflect lights of very high intensities.

As mentioned previously, functional imaging of the retina has already been achieved on big mammals. Schallek et al. [13] indeed achieved functional imaging of the cat eye. To do so, visible light was directly sent onto the retina with a LCD (see Fig.2.1a). The retina was meanwhile illuminated with NIR light, to which photoreceptors are insensitive, the IR retinal reflection is afterwards imaged. In a healthy retina, the photoreceptors activ- ity (due to visible light stimulation) induces among others a blood flow (hemodynamic response) and a deoxygenation of the hemoglobin. These two reactions are believed to create a change in the reflectance of IR light. Hence, by substracting the IR reflectance

6

CHAPTER 2. METHODS AND MATERIALS 7 image of a non-stimulated retina to the stimulated retina, one can visualize the functional response of the retina (see Fig.2.1b).

(a) Stimulation (b) Differential retinal reflectance

Figure 2.1: Retinal functional imaging of the cat eye, adapted from [13]

Owing to the large coefficient of reflection of the cornea and the lens with respect to that of the retina [15], most eye fundus systems are designed as shown in Fig.2.2. The illumination is made with a source ring (SR), conjugated with a hole mirror (HM) which is itself conjugated with the plane of the pupil of the animal. Such an annular illumination of the retina leaves a so-called corneal window in the centre of the eye to image the retina through HM and to avoid corneal reflections. Furthermore, the hole-mirror acts as a diaphragm and only the rays coming from the retina can reach the imaging path, and additional aperture stop (AS) blocks the illumination light reflected by the cornea and the retina.

Figure 2.2: Non-contact eye fundus setup, HM: hole mirror, AS: aperture stop, FS: field

stop, SR: source ring, OL: ophtalmic lens, adapted from [16]

CHAPTER 2. METHODS AND MATERIALS 8 2.1.2 Materials

For the purpose of the optical system, a wide field of view is required, to stimulate and im- age a retinal area which is as large as possible. The system should also be able to produce high illuminance levels on the retina, of the order of magnitude of 10

photons.cm

.s

to activate the optogenetic channels.

Thus, a DMD was preferred to a LCD to stimulate specific areas of the retina. A source ring conjugated with the cornea was also chosen to illuminate the retina, as done in most eye fundus systems. Yet, instead of designing a system based on a hole mirror, we decided to innovate and develop a system using the polarization of light. Such a polarimetric system is more compact than a hole-mirror system, which is very convenient as the system has to fit in an existing in vivo imaging setup. With a polarimetric system (see Fig.2.3), there is no need to conjugate the source ring with the hole mirror, which is here a polarizing beamsplitter (see Fig.2.4.b), and the source rings can be changed to fit the rat or the mouse eyes.

2.1.3 Initial scheme and planning of the optical design

Given these choices of materials, made before the beginning of the project, I had to design an optical frame. Imaging a retina requires an afocal system. If the eye is relaxed, which is the case when the animals are anesthetized, a parallel beam sent on the eye is focused on a point on the retina and vice versa the light coming from a point on the retina is a parallel beam at the output of the eye optics (see Fig.2.3). Hence, as the aim of the system is to stimulate and image points of the retina, it was decided to design a system mainly based on afocal systems. Furthermore, afocal systems are easy to align. Given all these preliminary choices, the initial scheme from which the optical system should be inspired is shown in Fig.2.3 .

The back and forth travel in the quarter wave plate gives the wave a half-wave phase shift, which allows the wave to go to the other arm of the polarizing beamsplitter (see Fig.2.4.b). The aperture stop in the imaging arm (which goes from the eye to the CCD) is used to block the reflections of the illumination light on the cornea and to select the light coming from the retina through the corneal window (see Fig.2.10c-2.10f).

The optical system was designed under the 2009 release of Zemax Radiant . I decided

initially to create a schematic model of the rat and mouse eye under Zemax with the

dimensions of [16, 17] (see Fig.2.4.a). It was also planned to follow this step with the

design of the illumination path as there is a constraint: the source ring should be conju-

gated with the plane of the cornea, the imaging and stimulation path would be simulated

afterwards.

CHAPTER 2. METHODS AND MATERIALS 9

Figure 2.3: Initial scheme of the system, the linear polarization state of the light is depicted at different positions

(a) Schematic of the mouse eye (b) A polarizing beamsplitter, from[18]

Figure 2.4: (a)Model of the rodent eye with different fields under Zemax , and (b)

principle of use of a polarizing beamsplitter 2.1.4 The illumination arm

The simulations were first run under Sequential Mode, and therefore without using po-

larization computations. Prior to the simulations, it was agreed to choose the optics from

CHAPTER 2. METHODS AND MATERIALS 10 a single optics provider in order to avoid complications when ordering the optics. The optics provider (Optosigma ) was selected because of its offer of high quality broadband

coatings.

The need for high quality broadband coatings arose from the superimposition of the stimulation and illumination arms; the illumination being in the NIR region [13], around 780nm and the stimulation being in the visible: around 570nm for the halorhodospins and around 480nm for channelrhodopsins. The polarizing beamsplitters and wave-plates were chosen with coatings fitting the wavelengths used. No coatings were available to al- low perfect transmission of both illumination/imaging and stimulation wavelengths in the polarizing beamsplitter used in the three arms (see in the top Fig.2.3). Owing to the ex- pected small flux reflected by the retina, the transmission of the NIR imaging wavelength was favored, though this choice induced a 40%-loss for the stimulation wavelength.

Based on [16], we planned to have a 15-mm front focal for the system. In the same way, the lateral magnification of the source ring was chosen to be g

' 3.5 so that the source ring can be easily built by the optomechanics workshop.

Since the retinal illumination is made with a ring, to achieve a wide illumination field on the retina rays must have a large incidence angle (see Fig.2.5b). We chose consequently to use the aspheric lens with the largest numerical aperture available as the ophtalmic lens N.A. = 0.69 f

= 23.5mm. As for the second lens of the first afocal (AF1 in Fig.2.5a), an achromatic doublet was chosen to limit the chromatic shift in the system since we want to focus or collect different wavelengths on the retina. Furthermore, achromatic doublets are usually optimized by the optics manufacturer to be as aplanetic (no spherical aberration and no coma) as possible.

Given all the enumerated constraints, several afocal systems were optimized using Zemax default merit function and compared using the spot diagram and geometrical image analysis tools. A couple of lenses was finally chosen to form the first afocal system AF1, the second lens of AF1 has a focal length f

= 81.0mm which fulfills the magnification requirements (see Eq.2.1 for the magnification of an afocal system).

The remaining lenses comprising the second afocal system (AF2) in the illumination arm were chosen to minimize the aberrations and keep the lateral magnification identical.

One should know that to minimize aberrations one has to favor the symmetry in an optical system, be it for the orientation of the lenses or the choice of the lens. As a consequence, the two lenses of the second afocal system were identical to the second lens of AF1 and orientated to form a symmetrical system.

Considering that in Sequential Mode the polarimetric behavior of the components cannot be simulated, it was decided to replace the polarizing beamsplitter by its equivalent glass thickness and a fold mirror (see Fig.2.5a). This simple simulation therefore takes into account the focal shift due to the crossing of the polarizing beamsplitter.

The dilated pupil size of mouse [17] and rats [4] are respectively in the range 1.8 − 2.0mm

and 3.5 − 5.0mm. We want to maximize the flux and set a corneal window, the effective

CHAPTER 2. METHODS AND MATERIALS 11 output area of the cornea, as large as possible to collect as much flux as possible. Thus, for the rat eye the source ring image in the plane of the cornea is set to 2 mm of internal diameter (the corneal window) and 1 mm of internal diameter for the mouse. The external diameter can theoretically be chosen as large as possible to let as much light as possible enter the eye through the pupil. The real size of the source rings is merely deduced from their image size in the plane of the cornea. The results of the simulation of an annular illumination of the rat eye in the plane of the cornea Fig.2.5c and in the plane of the retina Fig.2.5d are shown below.

(a) Simulation of the illumination arm (b) Annular illumination of the rodent eye

(c) Image of the source ring on the eye pupil (d) Illumination of the retina

Figure 2.5: Results of the simulation of the illumination arm under Zemax , 50,000 rays

were used for (c) and (d)

CHAPTER 2. METHODS AND MATERIALS 12 2.1.5 The imaging arm

The main constraint on the imaging arm is to be able to image a wide field. The end detector of the imaging arm was already available when I began to design the optical system. This detector is an industrial fast CCD camera, whose detector is a 10 mm- square matrix. For reasons of symmetry and high collection efficiency, it was decided to use an aspherical lens similar to the ophtalmic lens to focus the image onto the CCD instead of using a commercial objective for the camera (see Fig.2.6).

The lateral magnification g

of an afocal system working with a lens L1 of focal f

and a lens L2 of focal f

is given by the following equation:

g

= − f

f

(2.1)

Figure 2.6: Results of the simulation of the imaging arm under Zemax

The design of the second afocal in the imaging arm (AF3 in Fig.2.6) is based on the results

of the simulation of the illumination arm. By assuming that the iris is the pupil of the

system, the simulation showed that the field of illumination of the retina is 37.7

for the

rat and 30

for the mouse. The object field of view in the imaging arm must of course be

smaller than the illumination field as a point of the retina that is not illuminated cannot

be imaged. Hence, the object field of view was chosen to be 35

for the rat and 29

for

the mouse, which respectively corresponds to object field (size of objects on the retina)

of 3.7mm and 1.4mm.

CHAPTER 2. METHODS AND MATERIALS 13 The object field of the rat being the larger to image on a 10-mm large CCD matrix, it is the limiting case upon which we rest. The total magnification of the system is the product of the magnification of AF1 and AF3. Knowing the magnification of AF1, g

= −3.45 and desiring a total magnification enabling 3.7mm to be imaged onto 10mm that is to say g

= 2.70, we derive that the magnification of AF3 should be around:

g

= g

g

(2.2)

g

= −0.78 (2.3)

Finally, for reasons of symmetry, a lens identical to the second lens of AF1 was chosen to be the first lens of AF3. The second lens of AF3, an achromatic doublet, was chosen with a focal length of f

= 60mm, which results in magnifications g

= −0.72 and g

= 2.48.

2.1.6 The stimulation arm

As previously, the polarizing beamsplitter was simulated by a glass thickness and a fold mirror in the simulation arm (see Fig.2.7). The thickness of the other polarization elements, quarter-wave and half-wave plates, were neglected and would just change the distances between the afocal lenses during the alignment of the setup.

Figure 2.7: Simulation of the folded stimulation arm

The aim of the stimulation arm is to conjugate a point of the DMD with a point of the

retina. To gain room for the dichroic mirror and the half-wave plate (see Fig.2.3), it

was decided to form another afocal system of unity magnification (see AF4 in Fig.2.7)

CHAPTER 2. METHODS AND MATERIALS 14 by using the remaining lens of AF2 (see Fig.2.5a) with an identical lens. Three lenses of different focal lengths were chosen to collimate the light reflected by the DMD. These three different focal lengths allow to cover three different fields on the rat or mouse retinas, further information about the optical design of this part can be found in Appendix B.

2.2 Synchronizing the acquisition and stimulation

The acquisition of images to record the hemodynamic response concurrent to a stimulus required an Arduino chip in order to repeat a hardware trigger signal sent by the DMD (see Fig.2.8). The electronics of the DMD are indeed designed to send a single hardware trigger after the beginning of a stimulus. In order to record several images to follow the dynamic of the response, I had to create an Arduino program that repeats the hardware trigger of the DMD for each frame. It was also required to modify the C++ acquisition and display software provided with the camera.

Figure 2.8: Acquiring a series of images after a stimulus

2.3 Mounting and optimizing the optical system

2.3.1 Mounting and aligning the system

The optical system was aligned using the materials and methods described in Appendix

A. Great care was given to the reference axis defined by the alignment laser. The

ophtalmic lens was placed at the border of the breadboard so that the animal is outside

of the breadboard during imaging. This configuration allows to change the orientation

and position of the animal without risking to move involuntarily optical components.

CHAPTER 2. METHODS AND MATERIALS 15 The imaging arm was first aligned. The illumination arm was afterwards aligned using the imaging arm optical axis, and so was finally the stimulation arm.

2.3.2 Experimental protocol

The experiments were performed in vivo and the animals (rats and mice) were conse- quently anesthetized with 100mg/kg ketamine and 10mg/kg xylazine injected intraperi- toneally. One to three drops of atropine 1% in collyrium solution was deposited on the cornea of the animals so as to dilate their pupils. The dilatation of the pupils enables to send and collect as much light as possible into the eye. In order to prevent eye dryness and the whitening of the cornea, which decreases the transmission of light to the retina, a gel with A vitamin was frequently provided onto the cornea of the animals.

The animals were hold on a commercial holder furnished with the Micron IV Retinal Imaging Microscope of Phoenix Research Labs . To image and stimulate a specific

retinal zone of an eye, one should first of all turn on the illumination light. The rodent should afterwards be placed so that the image of the source ring is in the plane of the cornea and co-centered with eye. The first step is to modify the orientation of the animal in order to observe the optic disk (see Fig.2.9b). Following the observation of the optic disk, one can use it as a landmark and orient the animal to observe a given retinal zone.

2.3.3 Optimizing the system

For the image acquisition, the gain of the camera was set to the minimum in order to reduce the noise, and the images were exported either in RAW-8 bit or in Monochrome- 8 bit to allow a convenient image processing. The exposure time was set manually to exploit the full dynamic range of the camera without saturating the CCD pixels.

Once the alignment step was completed, preliminary tests were run on the imaging arm in order to check if it could properly image a rodent retina. The rodent was prepared as described in section 2.3.2. As showed in Fig.2.9a, no relevant fundus images were obtained due to numerous parasitic reflections of the illumination light on the polarizing elements. The reflectance of the retina is indeed very low, between 0.1 and 10% for human eyes depending on the wavelength and the retinal area [19], and small reflections arising from the optics may veil the useful image.

The system was re-aligned several times with no improvements. Hence, it was agreed

to check if the imaging system was operational by using fluorescent imaging, which is

not sensitive to back reflections of the illumination light since the imaged wavelength is

different from the illumination wavelength and can be discriminated. A suitable epifluo-

rescence filter cube was borrowed from a fluorescent microscopy setup and conveniently

introduced in the imaging arm. By using a powerful light source and injecting a rat with

CHAPTER 2. METHODS AND MATERIALS 16 diluted fluorescein, an angiography image of the rat retina was obtained (see Fig.2.9b).

This image permitted to assess that the imaging arm was functional and that further improvements of the system were needed to acquire clear images.

(a) Parasitic reflections in the imaging arm

(b) Processed image of a rat retinal angiography

Figure 2.9: Checking if the imaging arm is operational

Further investigations led to the conclusion that the main sources of back reflections were the polarizing beamsplitter shared by the three arms (see Fig.2.3 in the top) and the quarter-wave plate close to the ophtalmic lens (see Fig.2.3 in the top left). The illumination light is indeed reflected by these two elements towards the imaging arm and the detector.

As for the polarizing beamsplitter, it was shown that the reflections originate from the transmitted p-polarized component of the illumination light (see Fig.2.10a-2.10d). By polarizing the illumination light prior to its entrance in the cube, the amount of stray light was considerably reduced (see Fig.2.10b-2.10e).

The reflections arising from the quarter-wave plate were handled by tilting this plate

(which does not change much the output polarization) so that the aperture stop (which

location can be seen in Fig.2.3 in the top-right) blocks them (see Fig.2.10f). One will

also understand the utility of the aperture stop by seeing the corneal reflections and the

corneal window images in the plane of the aperture stop. Such a strong reflection of

the source ring was obtained by exchanging the animal cornea with the more reflective

alignment target (see Fig.2.10c).

CHAPTER 2. METHODS AND MATERIALS 17

(a) Schematic illumination with unpolarized light

(b) Schematic illumination with polarized light

(c) Image of the source ring in the plane of the cornea

(d) Real system with unpo- larized light illumination

(e) Real system with polar- ized illumination

(f) Parasitic reflection on the quarter-wave plate

Figure 2.10: Getting rid of the parasitic reflections

2.3.4 Final system

New tests were run with a more powerful illumination light source in order to obtain clear images without using fluorescence imaging. These tests were first run with green light as the contrast is stronger than in the NIR region due to the absorption of hemoglobin.

The results of these tests are introduced and discussed in the next chapter.

Chapter 3

Results and discussion

In this chapter, the results obtained with the optical system are described and discussed.

Future developments for improving the system are also introduced.

3.1 Illuminating and imaging

Mouse and rat eye fundi were acquired with the mounted optical system by using white (see Fig.3.1a-3.1b), green (525nm) and NIR light (780nm). The images do not exhibit a good contrast, maybe because of the monochrome 8-bit encoding format of the camera.

The best contrast was achieved with green light illumination (see Fig.3.1c). Green light is indeed best absorbed by hemoglobin, enhancing the contrast between the vessels and the surrounding tissues.

Moreover, the experiments showed that it is extremely difficult to align the animals eye, especially the tiny mouse eye, with the optical axis of the system. In contrast to contact eye fundus systems, the animal breathing is also a source of image instability. A homemade holder was therefore designed to gain stability and allow an easier alignment of the mouse eye.

In spite of the poor contrast, the resolution is good enough, on-axis and off-axis, for the initial purpose of the setup. The image quality and the wide field of view (31

for the rat, 30

for the mouse) allows indeed to use the optic disk as a landmark and target specific areas of the retina.

It can be seen in Fig.3.1a-3.1c that back reflections on the ophtalmic lens were processed and deleted. The addition of an adapted mask (see Fig.3.1c) before the ophtalmic lens allowed to suppress this back reflection, unfortunately to the detriment of the center of the field. Back reflections were expected [20] as there are numerous optics in common between the illumination arm and the imaging arm. The system used in [16] (see Fig.2.2),

18

CHAPTER 3. RESULTS AND DISCUSSION 19

(a) Mouse eye fundus (b) Rat eye fundus (white) (c) Rat eye fundus (green) Figure 3.1: Eye fundi obtained with the optical system and white light illumination (a) and (b) and green light (525 nm) illumination (c)

where there is just one optics between the hole mirror and the eye, is a better approach to eliminate back reflections on the imaging optics.

Furthermore, a lot of time was lost during the optimization step due to tests with NIR illumination light instead of testing the setup with visible light, especially green light, with which back reflections are easier to detect. Retinal images are also somewhat more apparent and with a better contrast when imaged with green light, which helped me to find the good focus for the camera.

Optimizations are yet a necessary step when it comes to the development of any system and the performances achieved with the illumination and imaging arms are satisfactory with regards to the criteria set at the beginning of the thesis.

3.2 Stimulating

The stimulation arm of the optical system is operational since patterns are collimated and of adequate dimensions at the output of the ophtalmic lens, as can be seen in Fig.3.2a- 3.2c. Projection of simple patterns, such as the one depicted in Fig.3.2b, were realized on the retina (see Fig.3.2d). It should be noticed that the black spot in Fig.3.2d is part of the projected pattern and not due to the mask shown in Fig.3.1b since that one was removed, nor to the image substraction of back reflections since the back reflections erased in Fig.3.1a-3.1c are correlated to the use of the illumination arm, which was not used for this acquistion.

However, combined patterned stimulation and imaging were quite unstable because of

great difficulties to keep aligned the rodent eye with the optical axis of the setup. Hence,

combined projections and imaging of more complex patterns (chequered or grid patterns)

on the retina were not successful.

CHAPTER 3. RESULTS AND DISCUSSION 20 Photometric measurements were made with a wattmeter to check if the high power stimulation requirement was fulfilled. It appears that only 4% of the flux reflected by the DMD is incident on the eye (see Fig.3.3). This poor stimulation efficiency should be compared with the theoretical ideal value of 15%, which was computed with the manufacturer’s data.

(a) Stimulation pattern "ALP" dis- played on the DMD and collimated to- wards the eye

(b) Pattern displayed by the ALP, black: no light reflected, white: reflec- tion toward the eye

(c) Collimated image of the pattern of Fig3.2a at the output of the ophtalmic lens

(d) Fundus of a rat eye stimulated with green light (525 nm) with the pattern of Fig.3.2b

Figure 3.2: The stimulation arm, (d) the image was filled with black in the left-bottom to allow contrast enhancement

The stimulation efficiency could be doubled by changing the way the DMD is illuminated.

Indeed, instead of using a compact polarimetric system with a polarizing beamsplitter and

a quarter-wave plate as shown in Fig.2.3, one could directly use the diagonal switching

CHAPTER 3. RESULTS AND DISCUSSION 21 feature of the DMD and set a direct oblique illumination of the DMD. This method was not implemented due to a lack of time.

In spite of the poor stimulation efficiency, a 5 ∗ 10

ph.cm

.s

stimulation intensity was achieved in the plane of the eye with an off-the-shelf 525 nm illumination light. This intensity corresponds to the threshold for the activation of the optogenetic channels as measured in [10]. By using a more powerful source and implementing the previously described enhancement, a stimulation intensity of 10

ph.cm

.s

could be achieved, such an intensity enables to properly activate most optogenetic channels [10].

Figure 3.3: Photometric measurements on the stimulation arm

3.3 Future developments

In spite of very promising results for a prototype, many enhancements are needed to improve the repeatability and stability of the system in order to perform the stimulation of optogenetic channels on the retina.

As a consequence, a self-designed holder for the animals was ordered from a 3D printer manufacture and should simplify the image acquisition and the alignment of animals.

In the same way, a system indicating the correct distance to which the eye should be positioned with regards to the ophtalmic lens would also make less complex the alignment of the animals. The animals were indeed positioned by superimposing the image of the source ring on the cornea, which is feasible with visible light illumination but more difficult to realize with NIR illumination light. Using a camera with a larger bit-depth could also improve the contrast of the fundus images.

In addition to the previously described enhancements, preliminary studies were made in

the end of this Master’s thesis to design a compact system using the hole-mirror technique

CHAPTER 3. RESULTS AND DISCUSSION 22

[16], which is theoretically less prompt to back reflections of the illumination light.

Chapter 4

Conclusion

In this Master’s thesis, I could, from the animal handling to the optical design and optimizations, develop an operational prototype of an optical system for in vivo non- contact combined imaging and patterned stimulation of mice and rats eyes.

The requirement of a wide imaging field was validated. Optimizations focused on the removal of back reflections of the illumination light on the imaging optics at the expense of the field center. Simple patterns were successfully projected on a rat retina and future developments should allow to reach a stimulation intensity high enough to activate the optogenetic channels.

Patterned stimulation of the rodent retina coupled to cortical imaging would permit to assess the functional state of specific areas of optogenetically treated animals. One could, for instance, project grid patterns with different spatial frequencies and observe the cortical response of the animal in order to assess the visual acuity of animals.

An enhanced system, using the hole-mirror configuration, could allow to reach the con- trast necessary to perform differential functional imaging of the retina. Coupled func- tional imaging of the retina and the visual cortex would create a frame enabling one to assess the functional state of the whole visual neural circuitry.

23

Chapter 5

Acknowledgements

My special thanks go to my supervisor, Emilie Macé, whose enthusiasm and patience incited me not give up when the project became clearly tricky. I also would like to thank her for letting me so much autonomy, from the handling of animals to the optical conception and development, which really helped me to reason and act as a scientist on my own instead of relying on the work of others. I definitely think that it would have been difficult to find another degree project matching so much my interests.

My best wishes of success to the whole team, especially Olivier, Serge and Stéphane, whose work and experiments are fascinating and on the limit of being pure science- fiction.

I would like to thank as well my examiner, Linda Lundström, for helping me to catch the ins and outs of KTH’s specifications for a degree project. What is more, I would not have so much enjoyed my stay and studies in Sweden without the help and patience of Kjell Carlsson as an international coordinator as well as a teacher. Tack så mycket Kjell!

Finally, I would like to thank my family and friends for their kind support, and especially my girlfriend, Marie, who has always been there to incite me to go forward in life.

24

Appendix A

Aligning an afocal system

The afocal system is the basis of our optical system, it allows easy image formation and therefore easy optical conception. To align an afocal system one should first gather an alignment laser, an iris diaphragm, an economy plane mirror, an alignment target and a ruler. One should patiently and carefully follow the steps below in order to achieve a correct alignment.

1. The laser, which is reference of the setup, should be carefully aligned at the good height. For that purpose a ruler may be used but a self-designed target on a paper at the correct height is also convenient. One can use off-the-shelf collimated lasers but the best solution is to collimate the fiber end of an alignment laser with a lens (with the auto-collimation method) so that the collimated beam covers the full aperture of the lenses of the setup.

2. Place and center the iris diaphragm close enough to the laser to let room for the afocal system (see Fig.A.1.a) but far enough so that the diaphragm does not limit the aperture of the beam when opened at maximum. Image the beam at the output of the diaphragm with a sheet of paper. To check if the diaphragm is well-aligned, open at maximum the aperture diaphragm, close slowly the aperture and observe if the circular beam is uniformly being vignetted. If a part of the beam is disappearing faster than its counter-opposite, re-align the diaphragm. Repeat this step until the closure of the diaphragm produce a perfect circular vignetting.

3. Close the iris to let a tiny beam pass through the aperture. Place and center afterwards the alignment target far enough from the alignment laser for the afocal system to fit in the interval (see Fig.A.1.a). Open and close slowly the aperture to check if the alignment target is well-centered on the beam.

4. Close the iris to let a tiny beam pass through the aperture. Place the first lens, L1 in Fig.A.1.b, so that the output beam is still well-centered on the alignment target.

5. With the same iris setting, place a plane mirror on the back of the mount of L1

25

APPENDIX A. ALIGNING AN AFOCAL SYSTEM 26 (see Fig.A.1.c). If the lens is tilted with respect to the optical axis, the reflection of the beam should be apparent on the diaphragm. Hold the mirror and tilt the lens until the reflected beam is superimposed or co-centered (if the two beams are not of the same size) with the incident beam on the diaphragm.

6. Repeat steps 4 and 5 until the output beam is well-centered on the target and the reflection is superimposed or co-centered with the aperture. The method should converge quickly.

7. Place the second lens of the afocal system approximately at the right distance of L1 (the right distance is the sum of the focal length of the two lenses).

8. Hold a plane mirror at the back of the mount of L2. Move and tilt L2 so that the reflected beam is perfectly superimposed with the beam coming from the aperture.

If L2 is not exactly at the right distance of L1, the reflected beam will not be of the same size as the beam passing through the aperture (see Fig.A.1.d).

9. Repeat steps 3 for L2 and 8 until the output beam is aligned both with the align- ment target and iris diaphragm (see Fig.A.1.e).

Figure A.1: Aligning an afocal system

Appendix B

Additional calculations

B.1 Choosing the collimation lens

The illumination and imaging arms being completed, a collimation lens to image the DMD pattern on the rodent retina had to be chosen. As it can be seen in Fig.B.1, the pattern displayed by the DMD is collimated, sent into a first afocal system of unity magnification (AF4 in Fig.2.7) and afterwards in the ophtalmic afocal system (AF1 in Fig.2.7), the output collimated beam is finally imaged by the eye optics onto the retina. The two consecutive afocal systems are equivalent to a single afocal system of lateral magnification equal to the product of the lateral magnification of each afocal system.

In the case of the rat eye, a square pattern of half-side 10mm (y in Fig.B.1) should be imaged on an retinal area of half-side 1.75mm, which corresponds to a stimulation half-angle θ

= 20

.

The magnifying power G of an optical system is defined by (see Fig.B.1 for the nota- tions):

G = θ

θ (B.1)

In the case of an afocal system, the magnifying power is equal to the inverse of the lateral magnification of the system. Hence, using Eq.2.1, the total magnifying power of the two afocal systems between the DMD and the eye is G = 3.45.

What is more, with our model the rodent eye is equivalent to a single lens of focal length f

= 4.85mm for the rat and f

= 1.68mm for the mouse. The image size y

on the eye is given by:

y

= f

tan(θ

) (B.2)

Similarly:

y = f

tan(θ) (B.3)

27

APPENDIX B. ADDITIONAL CALCULATIONS 28 Thus, we can derive from the equations B.1-B.2-B.3 that:

f

= y

tan













arctan y

f

!

G













(B.4)

Hence, according to Eq.B.4, to stimulate 40

of the rat or mouse retina, a collimator lens of focal length f

= 99mm would be required. In the same way, to stimulate 30

of the rat or mouse retina, a collimator lens of focal length f

= 130mm. We chose consequently, in our optics dealer catalog, lenses of focal lengths 81mm, 100mm and 150mm to stimulate different fields of the retina.

Figure B.1: Schematic of the reduced stimulation arm

B.2 Additional pictures

APPENDIX B. ADDITIONAL CALCULATIONS 29

(a) Spot diagram of the imaging arm with a mouse eye, computed under Zemax

(b) Spot diagram of the imaging arm with a rat eye, computed under Zemax

Figure B.2: Spot diagram of the imaging arm, simulated in the plane of the CCD

Figure B.3: Schematic of the final system (left) and real final system (right), FM:

Fold Mirror, AS: Aperture Stop, PB: Polarizing Beamsplitter, λ/4:Quarter-Wave Plate,

λ/2:Half-Wave Plae, POL: Linear Polarizer, DM: Dichroich Mirror, SR: Source Ring,

VIS: Visible stimulation light. Thick green line: stimulation arm, dashed red line: illu-

mination arm, thick red line: imaging arm

Bibliography

[1] Mark F. Bear, Barry W. Connors, and Michael A. Paradiso. Neuroscience, Exploring the brain. Lippincott Williams & Wilkins, third edition, 2007.

[2] D. Woolf. A comparative cytological study of the ciliary muscle. Anat. Rec., 44:145–

163, 1956.

[3] P. Artal, P. Herreros de Tejada, C. Munoz Tedo, and D.G. Green. Retinal image quality in the rodent eye. Visual Neuroscience, 15:597, 1998.

[4] S. Remtulla and P. E. Hallett. A schematic eye for the mouse and comparisons with the rat. Vision Research, 25(1):21, 1984.

[5] Joan Stein-Streilein and Hiroshi Keino. Regulatory T Cells and Clinical Application, chapter CD8+ T Regulatory Cells in Eye Derive Tolerance. Springer US, 2008.

[6] Schematic mouse, rat and human eyes. cited October 24, 2013.

http://prometheus.med.utah.edu/marclab/protocols.html.

[7] Christian Hamel. Retinitis pigmentosa. Orphanet Journal of Rare Diseases, 1:40, 2006.

[8] F. G. Holz et al. Age-Related Macular Degeneration. Springer-Verlag Berlin Heidel- berg, 2013.

[9] Daniel Palanker, Alexander Vankov, Phil Huie, and Stephen Baccus. Design of a high-resolution optoelectronic retinal prosthesis. Journal of Neural Engineering, 2:S105, 2004.

[10] Volker Busskamp, Jens Duebel, David Balya, et al. Genetic reactivation of cone photoreceptors restores visual responses in retinitis pigmentosa. Science, 329:413, 2010.

[11] James D. Weiland, Wentai Liu, and Mark S. Humayun. Retinal prosthesis. Annual Review of Biomedical Engineering, 7:361–401, 2005.

[12] Bizheva, R. Pflug, B. Hermann, B. Povazay, H. Sattmann, P. Qiu, E. Anger, H. Re- itsamer, S. Popov, J.R. Taylor, A. Unterhuber, Ahnelt P., and W. Drexler. Opto-

30

BIBLIOGRAPHY 31 physiology: depth-resolved probing of retinal physiology with functional ultrahigh- resolution optical coherence tomography. PNAS, 103(13):5066–5071, 2006.

[13] Jesse Schallek, Hongbin Li, Randy Kardon, Young Kwon, Michael Abramoff, Pe- ter Soliz, and Daniel Ts’o. Stimulus-evoked intrinsic optical signals in the retina:

Spatial and temporal characteristics. Investigative Ophthalmology Visual Science, 50(10):4865, 2009.

[14] Peixin Zhu, Otto Fajardo, Jennifer Shum, Yan-Ping Zhang Schärer, and Rainer W.

Friedrich. High-resolution optical control of spatiotemporal neuronal activity pat- terns in zebrafish using a digital micromirror device. Nature Protocols, 7(7):1410, 2012.

[15] A. Hugues. A schematic eye for the rat. Vision Research, 19:569, 1979.

[16] Dietmar Link, Clemens Strohmaier, Bernd U. Seifert, Thomas Riemer, Herbert A.

Reitsamer, Jens Haueisen, and Vilser Walthard. Novel non-contact retina camera for the rat and its application to dynamic retinal vessel analysis. Biomedical Optics Express, 2(11):3094, 2011.

[17] Christine Schmucker and Frank Schaeffel. A paraxial schematic eye model for the growing c57bl/6 mouse. Vision Research, 44:1857, 2004.

[18] Meadowlark optics catalog, 2012.

[19] François C. Delori and Kent P. Pflibsen. Spectral reflectance of the human ocular fundus. Applied Optics, 28(6):1061–1077, 1989.

[20] Edward DeHoog and James Schwiegerling. Fundus camera systems: a comparative

analysis. Applied Optics, 48(2):221–288, 2009.