An Implantable Microfluidic Device for Self Monitoring of Intraocular Pressure

Transcription

1 An Implantable Microfluidic Device for Self Monitoring of Intraocular Pressure 1,2 Ismail E. Araci, 1 Baolong Su, 1-3 Stephen R. Quake *, 4-6 Yossi Mandel * 1 Department of Bioengineering, Stanford University, Stanford, California, USA. 2 Department of Applied Physics, Stanford University, Stanford, California, USA. 3 Howard Hughes Medical Institute, Stanford University, Stanford, California, USA. 4 Department of Ophthalmology, Stanford University, Stanford, California, USA. 5 Hansen Experimental Physics Laboratory, Stanford University, Stanford, California, USA. 6 The Mina & Everard Goodman Faculty of Life Sciences, Bar Ilan University, Ramat-Gan, Israel. *Correspondence to: Stephen Quake: quake@stanford.edu and Yossi Mandel yossi.mandel@biu.ac.il Supplementary Note Sensitivity Calculation When the capillary forces are ignored, the system is in equilibrium in the case of identical liquid and gas pressure values. A step function increase in the liquid pressure disturbs the equilibrium and results in a steady flow towards gas reservoir. The flow continues until the gas pressure becomes equal to liquid pressure. According to the well known ideal gas law; Equation 1 P gas,init V gas,init =P gas,final V gas,final When the initial gas pressure is assumed as 1 atmospheric pressure (760 mmhg), the sensitivity, defined as displacement of the interface position in response to 1 mm Hg pressure change, can be derived from this as; 1

2 Equation 2 where V res, A ch and V ch are reservoir volume, channel cross section and channel volume respectively. When reservoir volume is much greater than the channel volume, sensitivity is simply dependent on the ratio between reservoir volume and channel cross section. In the more realistic case, the capillary effects and system compliance has to be taken into account. For a qualitative understanding of the role of these factors, we have used the equivalent circuit of the sensor inside a pressure chamber as shown in Supplementary Fig. S1. The liquid pressure, P is the only variable in this circuit model and C is the capacitance due to device compliance 36, ΔP fw,bw cap are the capillary pressure drop in forward and backward directions 37, R is the fluidic resistance dependent on the channel geometry and P air is the steadystate gas pressure. It can be seen in the equivalent circuit model (Supplementary Fig. S1) that, sensitivity (corresponds to fluidic flow and it is analogous to the integral of the current passing through the resistance R over time) will be reduced due to capillary pressure drop, ΔP bw,fw cap and capacitive effects of device compliance, C). We attribute the discrepancy between measured and theoretical sensitivity values to these factors. Besides the reduced sensitivity, the capillary effects cause a nonlinear behavior when they have positive values (hydrophobic surface) because in this case, pressure change required to move the interface has to be greater than ΔP bw,fw cap and thus indicated by a diode in the equivalent circuit 38. In order to eliminate the nonlinear behavior, the channel surfaces has to be rendered hydrophilic or equivalently channels has to be filled with a high lubricity liquid, in which cases ΔP cap bw,fw will become negative and degrading effects of the capillary pressure drop can be ignored. The optical effects of IOP sensor We have used Trioptics Optispheric IOL, to test the fabricated IOLs. The image formed on the focal plane of the IOL was focused to a CCD camera with AMT50 objective (50 mm focal length), as reported 13. Image quality 2

3 of U.S. Air Force (USAF) targets projected through the lens was not affected by the presence of the IOP sensor (Supplementary Fig. S3). Quantitative assessment of the sensor effect on image quality was further determined by calculating the modular transfer function (MTF) of an image produced by an intraocular lens with and without the sensor. We found that even in the worst case scenario, where the sensing channel is positioned on the optical axis and equator, the MTF was not significantly affected by the sensing channel. The results were obtained for both tangential and sagittal focusing (Supplementary Fig. S4). Similar data were obtained of other lens configurations where the sensing channel is in the mid-periphery (data not shown). These results clearly show that IOP sensor does not cause image distortions. The sensing features in our device can theoretically contribute to the glare through scattering of light; for example, in the worst case scenario, the channel on the lens can cause total internal reflection (TIR, Supplementary Fig. S5a). We evaluated the potential glare image formation by the microfluidic channel by performing ray tracing analysis using Zemax optical design program. We found that light source arriving at 15 degrees angle off the optical axis produced negligible glare images on the retina. A large optical detector covering the entire posterior retina was used to perform ray tracing of 5,000 projected rays arising from a light source positioned at infinity at an angle of 15 degrees. We modeled the worst case scenario where the microfluidic channel was designed with 100µm 100µm dimension, filled with air and positioned either on the optical axis (similar to Supplementary Fig. S3b) or 1.5 mm away from the optical axis (similar to Supplementary Fig. S3c). The ray tracing analysis clearly showed that the glare images were not of significant clinical importance, with size of up to 150µm and irradiance 3-4 orders of magnitude below the retinal image (Supplementary Fig. S5b). As expected from the small dimension of the channel, the glare area is significantly smaller than what was calculated to be induced by the edge of clinical available IOL design 41. Glare image can be further reduced by round channel architecture 16. 3

4 Sensor Life time Besides high sensitivity, one other requirement of a microfluidic IOP sensor is the hermetic sealing. However, ideal hermetic sealing is challenging to realize, thus even for best sealant materials (i.e. glass or ceramics) air leakage has to be taken into account due to the long lifetime requirement of the device (years). The sealant material of choice for this study was parylene-c which is known to have four orders of magnitude lower permeability than PDMS, comparable to epoxy (~3 cc.mm.m 2.day.atm) 39 (See also here). For a 1 (0.3) µm thick parylene-c layer there was a drift of 2 (6) µm minute 1 at 7 mmhg pressure. This can be reduced by three orders of magnitude by using a thicker parylene-c or a sealing material with lower permeability such as glass 40. We propose that the drift can be corrected during readout process upon detection of the interface. Supplementary Fig. S6 shows the feasibility of this approach by demonstrating that three measurements taken in 24 hours intervals show little variation when drift corrected. For measurements done consecutively in the time scales of t = sensitivity/drift the drift compensation will not be required (Supplementary Fig. S6). 4

5 Supplementary Fig. S1 Equivalent electrical circuit of the sensor inside a pressure chamber. P is the liquid pressure, C is the capacitance due to device compliance, ΔP cap fw,bw are the capillary pressure drop in forward and backward directions, R is the fluidic resistance dependent on the channel geometry and P air is the steady-state gas pressure. 5

6 Supplementary Fig. S2 Enhancement of interface visibility by retroillumination. Gas/fluid interface visibility could be enhanced by retroillumination of the eye through a side lighting at the level of the limbus or slightly more posteriorly. A reddish homogenous background enhanced channel and gas/fluid interface visibility both through a surgical microscope (a,b) and by an iphone camera (c,d). Retroillumination is routinely used during ophthalmic examination for enhancement of various ocular pathologies 6

7 Supplementary Fig. S3. USAF target images for 4 different lenses a) Lens without features b) Lens with µm straight channel (and scales) right on the equator c) Lens with µm, 2 mm diameter circular channel (and scales) d) Lens with µm, 4 mm diameter circular channel (and scales). Inlets show the corresponding lens configurations 7

8 Supplementary Fig. S4 MTF of two lenses, one without any features (hollow black lines) and one with a straight channel and scales positioned on the optical axis (filled black lines), is measured for both tangential (dashed lines ) and sagittal (solid lines)) focusing. 8

9 Supplementary Fig. S5 Schematics of an IOP sensor, demonstrating the negligible effect of glare produced by the embedded sensor channels. a). When the incidence angle is larger than the critical angle, TIR from the channel surface can cause glare; the close-up look of the region in green rectangle is shown on the right. The Zemax simulation result showing the glare caused by 3 mm diameter circular channel with µm 2 rectangular cross section, light incidence angle is 15 degrees b). Ray tracing analysis demonstrates minimal effect of the channel on retinal image. 9

10 Supplementary Fig. S6 The long term stability measurement results. The curves are recorded in 24 hours intervals and corrected for 2 µm min 1 drift. Each set of data shows the displacement of interface position with respect to chamber pressure for a µm 2 cross section channel and a reservoir volume of µm 3. 10