1 of 1

Shiley eye project

Robert Haeuser, Wanfang Wu, Lihua Wei, Ji In Kim

Department of Mechanical and Aerospace Engineering at University of California San Diego

Sponsored by Dr. Ben Xu and Shiley Eye Institute

Overview

Future Improvements

Impact on Society and Safety

Acknowledgements

Dr. Ben Xu for constant input on the design and access to 3-D imager
Steve Roberts for expert help and advise on the RC to OpAmp circuit
Dr. Tustaniwskyj for constant advice and guidance during this venture
Pedro Franco Navarro for constant advice and guidance during this venture
Tung Nguyen for his help with coding and testing

References

https://www.gsa.gov/portal/content/101308
OSHA Safety and Health Regulations for Construction standard number 1926.56
SS-405, lighting for office and industry, SAIF corporation

Summary of Hardware Performance

Impact:

Glaucoma is one of the leading causes of blindness worldwide. A crucial structure in the development of glaucoma is the drainage channel (labeled open/closed below).

This structure is controlled by the same muscles that modulate pupil diameter. To the extent of the author’s knowledge, no study has yet been done on the drainage channel under dynamic pupil conditions. Thus, this device will improve knowledge of the biomechanics of the drainage channel which could lead to better glaucoma prevention methods.

Safety:

Device is deemed safe.
The largest operating current is 20mA which is far below the 100mA fatal threshold¹. Additionally, all wires are insulated, and the circuit is in a ABS plastic vessel.
For brightness, maximum LEDs intensity was measured as 800 lux with 90% of cases under 500 lux. For comparison, OSHA recommends first aid stations and general offices to be at 300 - 500 Lux². IESNA Lighting Handbook recommends 1000 lux for performance of visual tasks of low contrast or very small size, such as drawing work³. An overcast day is 1000 to 2000 lux. Thus, LEDs are within normal day-to-day intensities.

The Shiley eye institute asked for a smart light source that would induce the pupil to constrict to specific user-chosen diameter for interior cornea imaging. The software sends brightness commands after determining pupil size from analyzing the camera feed on the cornea scanner. This was achieved by wiring an Arduino to a current control circuit and 4 red LEDs, which is controlled by Matlab algorithms on the cornea scanning machine.

Components

Ardiuno

Arduino receives commands from Matlab and outputs PWM voltage signal which eventually determines LED brightness

Current control circuit

Low-pass filter converts PWM voltage signal (square wave) to a constant DC voltage
Resulting DC voltage goes through Op-Amp and transistor, which converts the voltage source to current source to better control LED brightness

Matlab code

Pupil detection code that detects pupil and measure its size from the eye imager video
Controller code that compares the current pupil size to the desired value and change brightness through closed-loop feedback

Screen Cover + Case

A tent-like cover that will surround the monitor of OCT machine computer to reduce ambient incident light into the patient’s eye. (Patient needs to be in a dark environment)
A case to put Arduino and circuit board.

Figure 3. Block diagram of the system

Figure 4. Light source circuit diagram

Accessory

MATLAB Code: Pupil Detection

Current Control Circuit

Figure 7. Screen cover

Screen Cover - The screen cover was designed to be removable and placed onto the monitor so the testing area can be as dark as possible to have the greatest change in pupil size.
Hardware Casing - The case is 3D-printed by ABS material. Its size is 140 x 70 x 85mm. The case will contain the Arduino board and most part of the circuit, including a small fan that cools down the circuit.

Recommendation	Justification
Change the size and angles of the screen cover	This screen cover was designed for general purpose use but it can be redesigned for more optimal use for specified space or user
RC to Op amp Circuit changed for different resistors and lower power source	By using a lower power source the circuit will not heat up as much and will be able to be left on without any worry
Optimization of Matlab code	The code was written to drive the pupil to one requested size but the code could be used to automatically driven to several sizes to be photographed by the imager.
3-point circle detection	Current pupil detection algorithm calls on matlab function imfindcircles() which in 15% of cases gives diameter error of up to 0.3mm. Hypothesized the artefacts from eyelashes confused the hough transforms used in imfindcircles. New algorithm would find 3 points on the pupil circumference and algebraically solve for the diameter.

The completed circuit consists of:

Low pass filter with voltage gain 0.385
Voltage controlled current source (VCCS)
A sensing resistor to control current flow
A TQ2-4.5V Relay that completely open the circuit when needed

Figure 8. CAD model of the case

Figure 5. Pupil detection process

Pupil detection Code

Captures image of pupil from video and turn it into grayscale
Applies binary filter to convert every pixel into either complete white or complete black
Pre-processing to get rid of reticle for more accurate results (runs filter 4 times)
Uses Hough circle transform algorithm to detect circle (pupil) from the filtered image and measure its diameter in pixel
Converts pixel scale to mm and outputs pupil diameter in mm
Processing time is approximately 0.1 ~ 0.2 sec per image

MATLAB Code: Controller

Figure 6. Schematic of closed loop system

PID controller with Graphical User Interface (GUI)

Opens up a window that asks user to input desired diameter (in mm)
Compute error by subtracting measured diameter (obtained from pupil detection code) from desired value
From this error, compute new output PWM voltage using PID algorithms to change brightness of LED

Glaucoma is a leading cause of blindness worldwide. Ophthalmologists use anterior segment optical coherence tomography (AS-OCT) imaging devices to study the internal anatomy of the eye. The iris is an intraocular structure that plays a crucial role in the pathogenesis of certain forms of glaucoma. While the configuration of the iris can be modulated by light, researchers lack a device that gives them precise control of the iris and pupil size. This has limited their ability to study iris biomechanics and its contribution to the development of glaucoma. With this light source, researchers hope to study how changes in the iris and drainage angle(labeled open/closed, see figure 2) contribute to anterior segment disease.

Figure 1. Front view of the eye

Background

Figure 2. Cross-section view of the eye

circle detection

Figure 9a. Pupil response over time (target size = 5.5mm)

Figure 9b. Pupil response over time (target size = 3.5mm)

Shown in figures 9a, b, the closed-loop system is able to drive the pupil diameter near the desired diameter in less than 3 seconds and hold the pupil at the desired diameter with oscillation amplitude less than 0.3mm for at least 2 seconds for the OCT machine to take images. In short, the closed-loop system meets the requirement for this project.