Light Hazard Measurement on an Ophthalmic Instrument

David Melo

, Pedro Vieira

and Filipe Soares

Fraunhofer Portugal AICOS, Rua Alfredo Allen, 455/461, 4200-135, Porto, Portugal

Department of Physics, Faculdade de Ci

encias e Tecnologia, Universidade Nova de Lisboa,

Quinta da Torre, 2829-516, Caparica, Portugal

Keywords:

Light Hazard Measurement, Ophthalmic Instruments, Fundus Camera.

Abstract:

When the acquisition of images is performed on human body structures, the safety of the person on which

the images are being obtained acquires fundamental importance. This condition gathers even more relevance

when the goal of the image acquisition is related with health examination. To ensure that every physician, when

imaging human structures is undoubtedly secure that it will not harm or cause any non-tolerable discomfort

to the subject undergoing the exam, medical devices have to surpass a set of speciﬁc norms depending on the

nature of the structure being examined. More speciﬁcally related with instruments for eye related diseases

diagnostic and treatment, the ISO-15004-2 norm is the currently accepted. As some types of eye exams

rely on highly concentrated radiation, its power has to be limited in order to prevent irreversible damage.

In this paper, a Light Hazard Measurement for a speciﬁc prototype will be addressed, on which the power

calculation method for the emitted radiation will be described and the compliance of the limiting norms will

be veriﬁed. The main conclusion from the work here described is that the prototype can be classiﬁed as a

Group 1 Ophthalmic Device according to the ISO-15004-2 norm.

1 INTRODUCTION

In the work presented in this paper, a fundus cam-

era prototype named EyeFundusScope, currently un-

der investigation by Fraunhofer Portugal AICOS, is

submitted to a light hazard measurement. The goal

of this procedure is to understand if the prototype can

be used in screening and diagnostics situations, with-

out provoking any harm to the agents participating on

them, from patients to examiners.

The prototype uses light sources emitting at differ-

ent wavelengths, with one of them emitting on a nar-

row region of the near infrared the electromagnetic

spectrum and the other emitting broadband white

light. The light sources used were Light Emission

Diodes (LEDs) and the measurement of its power was

performed with a photodiode power sensor. The emis-

sion spectrum of each light source was obtained using

a spectrometer.

In this section the crucial principles and tools

needed for the measurements are introduced and in

the next ones, the procedure, the consequent results

and the conclusions that may be obtained from them

are described.

1.1 Eye Examination

The eyes along with the brain, the nervous system and

its information transfer channels ensure one of the ﬁve

ways human beings can acquire and understand in-

formation advent from its surrounding environment.

Taking this into consideration, a regular eye examina-

tion is of unaccountable importance.

According to the American Academy of Ophthal-

mology every adult should perform an ocular evalua-

tion, at least, every 5 years (Feder et al., 2016). The

frequency of the examinations for patients with, one

or more risk factors, such as diabetes or with an al-

ready diagnosed ophthalmological pathology should

naturally be increased (Feder et al., 2016).

During screening actions and regular medical ap-

pointments, several regions of the eye are observed.

Depending on the type of exam and the eye region

the physician wants to observe, radiation may have

to be directed towards the eye. For instance consid-

ering retinal examination, taking into account fundus

low reﬂectance, the retina can only be observed when

a considerable amount of light passes the pupil, en-

ters the ocular globe and reaches the fundus of the

eye (DeHoog and Schwiegerling, 2009; Hammer and

Schweitzer, 2002; Preece and Claridge, 2002). For

Melo, D., Vieira, P. and Soares, F.

Light Hazard Measurement on an Ophthalmic Instrument.

DOI: 10.5220/0007575800870098

In Proceedings of the 12th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2019), pages 87-98

ISBN: 978-989-758-353-7

the observation of more anterior structures the use of

light is not so imperative for illumination but can also

be used to understand several physiological mecha-

nisms as occurs in pupillometry, on which the pupil-

lary response to light of speciﬁc wavelengths is stud-

ied (Kelbsch et al., 2016; Rukmini et al., 2015).

1.2 Light Emission Diodes(LED)

Nowadays, Light Emission Diodes are available

in a wide-range of wavelengths, existing an al-

most continuum spectrum of wavelength availabil-

ity from near ultraviolet to near infrared existing

also monochromatic emitters at deeper infrared wave-

lengths (Capit

an-Vallvey and Palma, 2011).

They present several advantages when compared

with the traditional incandescent lamps namely, a

longer lifetime, higher brightness, lower power con-

sumption and the possibility of electrical modulation

at higher rates (Malinen et al., 1998; Patrick and

Fardo, 2000). Its usage in Healthcare Applications

is widespread not only as part of diagnosis instru-

ments, but also for disease treatment, more impor-

tantly within the scopes of dermatology, ophthalmol-

ogy and recently, neurology (Rahman et al., 2013;

Saltmarche et al., 2017; Eells et al., 2004). The tech-

nique is named photobiomodulation and can be de-

ﬁned as the irradiation of tissue with low-intensity

light in the infrared band, provoking alterations on the

cellular functions (Desmet et al., 2006).

LED principle of operation is based on electrolu-

minescence, the production of light by ﬂow of elec-

trons (Schubert, 2018). More speciﬁcally for the pro-

duction of white light, a phosphor is usually added to

a blue light LED chip. The phosphor is excited by the

blue light coming from the chip and absorbs part of

it. Due to the excitation, it emits light of other visi-

ble wavelengths, usually in the green and red regions

of the electromagnetic spectrum that together with the

blue light partially transmitted, generates white light

(Chen, 2005; Schubert, 2018).

1.3 Photodiode Power Sensors

Photodiode Power Sensors are constituted by a semi-

conductor that produces a ﬂow of electrons (current)

dependent on the amount and wavelength of photons

(light intensity) that reach its working surface (Senior,

1986). From the current produced the power can be

obtained.

The power output of the photodiode power sensor

reveals a high dependence on the wavelength of the

radiation being measured. So, they can only be used,

as a direct power meter, when the light source being

measured has a narrow emission band. With this in

consideration, the photodiode operator can select the

expected wavelength of the radiation (hereinafter de-

scribed as λ

). Although, as the suppliers normally

provide information on the Sensor Spectral Response

(Figure 1) methods for the measurement of broader

emission bands situations can be conceived.

Figure 1: Spectral Response of the PD300-3W with the re-

placeable ﬁlter in. R(λ) is the relative sensitivity of the pho-

todiode for the wavelength λ.

The power measured by the photodiode is usually

performed on the highest peak of the spectrum and

is deﬁned by the following equation, according to the

White Paper by the Doctor Eﬁ Rothem ”Measuring

LED Power and Irradiance with Calibrated Photodi-

odes” available at Ophir website (Rothem).

Meas

R[λ

]

R(λ) × I(λ) dλ (1)

In which P

Meas

is the power read from the photo-

diode output display, R(λ) is the relative sensitivity of

the photodiode for radiation of wavelength λ, I(λ) is

the spectral intensity of the radiation being measured

and λ

is the wavelength selected on the photodiode.

Photododiodes are considered valuable and ac-

curate tools when the light source to be studied

presents an extremely narrow wavelength such as

lasers (Ready, 1971).

To validate Equation 1 it is possible to test if for

an ideal single-wavelength light source, accordingly

to the manufacturers the optimal situation for use of

photodiodes, the power measured by the photodiode

is equal to the real power.

The Equation 1 for the ideal light source presented

in Figure 2 (example of the spectrum of a laser emit-

ting at 550 nm ) becomes:

Meas

R[555]

R(λ) × I(λ) dλ (2)

And the real power for any type of light source is

given by the Equation 3.

Real

I(λ) dλ (3)

BIODEVICES 2019 - 12th International Conference on Biomedical Electronics and Devices

Figure 2: Spectral distribution of an ideal single wavelength

light source.

By the spectrum is easily deductible that I(λ) = 0

for every wavelength except when λ = 555nm. So

as shown in Equation 4 it is possible to prove that

Meas

= P

Real

. For validation of the equality, the value

I(λ) is not replaced by the intensity value from

the spectrum (Figure 2).

Meas

R[555]

R[555] × I(λ) dλ

Meas



R[555]



R[555]

I(λ) dλ

Meas

I(λ) dλ = P

Real

(4)

This way, it is shown that the Equation 1 for the

power measured by the photodiode, is congruent with

the fact that photodiodes are accurate direct power

meters for narrow band sources.

2 POWER MEASUREMENT

In this section, light hazard measurements will be pre-

sented for a smartphone-based handheld fundus cam-

era prototype. This prototype uses two LEDs as light

sources, one with a narrow peak on the near infrared

(NIR) region of the electromagnetic spectrum and the

other emitting broadband white light. The method for

examination of the retina with this prototype is char-

acterized by ﬁrstly illuminating the fundus with NIR

light for guidance. When the examiner is satisﬁed

with the image being observed, a button is pressed

and white light is turned on, only for a few millisec-

onds (typically about 50 to 100 ms is sufﬁcient). The

coloured image obtained is stored and becomes avail-

able for consultation on the smartphone. It was es-

tablished like this, to prevent miosis, an effect that

consists on the narrowing of the pupil by action of

the Autonomous Nervous System, when visible light

reaches the retina (Wang et al., 2016; Tran et al.,

2012). As miosis is not veriﬁed when infrared light

reaches the retina (Shen and Mukai, 2017), to limit

the use of white light only to the minimum to obtain

a color image of the fundus, this method was chosen

to allow a comfortable examination for the patients

and to avoid the need of mydriatic agents (drugs that

dilate the pupil) (Yuvacı et al., 2015).

The level of the emitted light intensity is con-

trolled by Pulse Width Modulation (PWM) (Dyble

et al., 2005). On the smartphone application the op-

erator can select this level on a range between 0 and

100 % of the maximum value allowed.

The emission spectrum of each LED was obtained

using a AvaSpec-ULS2048-USB2-VA-50 spectrome-

ter supplied by Avantes. The instrument used to ob-

tain the power values was a photodiode power sensor

PD300-3W. Photodiodes principle of operation was

already described in Section 1.3.

In order to collect all of the emitted light, in both,

spectrum and power measurements, the aperture of

the measurement device (spectrometer or photodiode)

had to be centered and placed at the focal distance of

the optical system covered by the prototype.

Its usable range goes from the near ultraviolet

wavelengths to the near infrared region (360-1100

nm) so it was able to recognize the full spectrum of

both LEDs.

The White LED is a broadband light source with

two peaks, being the ﬁrst narrow and in the blue re-

gion and the other with a wider band and a peak on

the yellow\green regions of the spectrum. The NIR

LED only has one peak.

The several steps one must consider when per-

forming this type of measurement in such conditions

for each LED, may be resumed by the following:

1. Obtain the spectra with the spectrometer;

2. Normalize the spectrum according to the maxi-

mum counts;

3. Calculate the wavelength of the highest peak by

post-processment of the spectrum (in this example

python libraries such as NumPy and SciPy were

used);

4. In the photodiode select the wavelength obtained

in the step 3;

5. Select several levels for the intensity of light

through the smartphone application - in this setup

Light Hazard Measurement on an Ophthalmic Instrument

the selected values were 30, 50, 75 and 100 % of

the current allowed by the hardware;

6. As photodiodes are very wavelength sensitive, it

is needed to multiply the Spectral Response pre-

sented in Figure 1 by the Spectrum obtained on

step 2;

7. By the direct proportionality relationship between

the area beneath the spectral curve and the power,

ﬁnd the real power emitted at each peak wave-

length.

The Step 6 is justiﬁed by the fact that the Power

measured by the Photodiode is given by the Equation

1 in Section 1.3, at the same time that the power emit-

ted by a light source is derived with the Equation 3.

To bring more clarity to the method described be-

fore, in the following subsections, not only the ﬁnal

radiant power values obtained will be shown, but also

the set of calculations required for each LED.

2.1 White LED

The white LED is a OSLON

 Black LUW H9GP

supplied by OSRAM. Chromatically it has a temper-

ature of 6500 K, which is considered a cool white

with a strong prevalence of lower wavelengths (Violet

and Blue regions of the spectrum). As referred previ-

ously four different levels for the intensity of the emit-

ted light were measured. The changes on the spec-

trum along the intensity levels were not representa-

tive, only a slight increase on the signal-to-noise ratio

with the increase of LED intensity. The normalized

spectrometer output for the White LED is presented

in Figure 3.

The power measurements obtained on the photo-

diode are presented in Table 1.

Figure 3: Spectrum of White LED at maximum intensity

(100%) obtained with the AvaSoft 2048.

Besides the calculations of these values it was re-

quired the multiplication of each spectrum obtained

by the spectral response. The curves for the LED

spectral intensity, spectral response, and result of its

multiplication are presented in Figure 4.

Figure 4: Top - In red is presented I(λ), the normalized spec-

tral intensity of the light source and in blue is presented

R(λ) the relative sensitivity of the photodiode. Bottom - Is

presented in black the result of the multiplication of the two

curves presented in the Top image - R(λ)×I(λ), hereinafter

referred as Weighted Spectrum.

To use the Equation 1, the sensitivity of the pho-

todiode at the speciﬁc wavelength is needed. For the

visible LED, λ

= 439 nm, so R[λ

] = R[439].

The variables needed for the Equation 1:

• R[439] - 0.62

• P

Meas

- 0.60 mW

Replacing the variables for these values, the Equa-

tion 1 becomes:

0.60 =

0.62

R(λ) × I(λ) dλ

0.372 =

R(λ) × I(λ) dλ

(5)

With the last equality of the Equation 5, and con-

sidering that the expression

R(λ) × I(λ) dλ is given

by the area beneath the Weighted Spectrum curve,

R(λ) × I(λ) (Figure 4 - Bottom), a direct proportion-

ality relationship between power and area can be es-

tablished. The variable Spectrum Area is the area be-

neath the curve presented in Figure 3, corresponding

to the emission spectrum of the LED and named I(λ)

in Equation 3.

With the variables presented next, the real power

Real

) emitted by the White LED can be calculated.

• Weighted Spectrum Area (WS

area

) - 49.46

• Spectrum Area (S

area

)- 62.63

• Power with photodiode sensitivity inﬂuence

Sens

) - 0.372 mW

BIODEVICES 2019 - 12th International Conference on Biomedical Electronics and Devices

Table 1: White LED Measured Power at 439 nm, P

Meas

, and Real Power, P

Real

, for each peak and for both peaks.

Smartphone Level

Wavelength

Meas

- 439 nm P

Real

- 439 nm P

Real

-555 nm P

Real

- Total

30 0,18 mW 0,044 mW 0.096 mW 0.140 mW

50 0,31 mW 0,077 mW 0.165 mW 0.242 mW

75 0,45 mW 0,111 mW 0.240 mW 0.351 mW

100 0,60 mW 0,150 mW 0.321 mW 0.471 mW

Considering the relationship of proportionality pre-

viously established, the following Equation can be

used:

Real

Sens

× S

area

W S

area

Real

= 0.471 mW

(6)

To compare the values obtained with the ISO

Norms, it is needed to know the percentage of the to-

tal power corresponding to each peak. In order to have

an accurate separation of the two peaks a Gaussian Fit

function from SciPy library was used. Two Gaussian

Fits were considered, one for each peak, with the pa-

rameters:

• Wavelength of the Center;

• Amplitude of the Gaussian;

• Width of the Gaussian;

The graphical results of the ﬁtting are presented in

the Figure 5.

Figure 5: Top - In black is presented I(λ), the spectrum of

the light source and in red is presented the Gaussian Fit

function. Bottom - Peaks are individually presented, the

ﬁrst in green and the second in blue.

From the area of each peak it is possible to calcu-

late each peak proportion on the total power, as pre-

sented by the following equations.

• A

Peak1

(%) - 31.8%

• A

Peak2

(%) - 68.2%

Peak1

= A

Peak1

× P

Real

Peak1

= 0.318 × 0.471

Peak1

= 0.150 mW

Peak2

= A

Peak2

× P

Real

Peak2

= 0.682 × 0.471

Peak2

= 0.321 mW

(7)

In Table 1 it is possible to verify the ﬁnal values

for the power of the White LED, for both peak wave-

lengths.

2.2 NIR LED

The Near Infrared LED is an Infrared Emitter

MTE1081C, supplied by Marktech Optoelectronics.

The normalized spectrum of the light emitted by the

LED is presented in Figure 6. Likewise what occurred

with the White LED, the changes on the spectrum

along the four intensity levels were not representative.

The method is slightly different to the one previously

presented for the White LED due to the existence of

only one peak.

Figure 6: Spectrum of the NIR LED at maximum intensity,

(100%) obtained with the AvaSoft 2048.

The power values measured with the Photodiode

are presented in Table 2.

The result of the multiplication of the NIR Spec-

trum by the Spectral Response of the Photodiode is

presented in Figure 7.

Light Hazard Measurement on an Ophthalmic Instrument

Figure 7: Top - In red is presented I(λ), the spectrum of

the light source and in blue is presented R(λ) the relative

sensitivity of the photodiode. Bottom - Is presented in black

the result of the multiplication of the two curves presented

in the top image - R(λ) × I(λ) - Weighted Spectrum.

Similarly to what occurred with the multiplication

of the white spectrum by the spectral response (Fig-

ure 4), the effect of the wavelength sensitivity of the

photodiode is noticeable. The sensitivity of the pho-

todiode for NIR wavelengths is much higher than it

is for lower wavelengths, fact that can be conﬁrmed

by comparison of the y-scale of the weighted spectral

response of both LEDs (Figures 4 and 7 - Bottom).

Calculations will be shown for the maximum in-

tensity level.

• R[820] - 2.57

• P

Meas

- 0.57 mW

Replacing the above values in Equation 1:

0.57 =

2.57

R(λ) × I(λ) dλ

1.465 =

R(λ) × I(λ) dλ

(8)

The variables needed for the real power (P

Real

)

calculation are:

• Weighted Spectrum Area (WS

area

) - 106.0

• Spectrum Area (S

area

)- 41.40

• Power with photodiode sensitivity inﬂuence

Sens

) - 1.465 mW

Real

Sens

× S

area

W S

area

Real

= 0.57 mW

(9)

The major difference between the used method for

each LED is related with the number of peaks. The

NIR LED only has one peak and because of that the

Gaussian Fit is not required, and P

Real

can be consid-

ered directly proportional to all the area beneath the

curve I(λ). With this approximation for the compar-

ison with the ISO norms, all the radiation emitted by

the NIR LED is centered at 820 nm.

The results for every level of intensity are pre-

sented in Table 2.

Table 2: NIR LED Measured Power at 820 nm - P

Meas

and

NIR LED Real Power at 820 nm - P

Real

. SL - Smartphone

Level; WL- Wavelength.

Power

Meas

Real

30 0,17 mW 0,168 mW

50 0,30 mW 0,299 mW

75 0,44 mW 0,439 mW

100 0,57 mW 0,570 mW

As can be perceived by comparison of the P

Meas

and P

Real

in Table 2, the values are equal or very iden-

tical.

This can be explained by the fact that on the region

where the NIR LED emits radiation (≈ 758-850 nm -

Figure 6), the Photodiode Spectral Response (Figure

1) is considerably ﬂat. Consequently, for the integral

R(λ) × I(λ) dλ, on the spectrum region of interest,

R(λ) will be nearly constant and similar to R[820].

3 ISO COEFFICIENTS

CALCULATION

In order to ensure the safety of the prototype, the val-

ues calculated for the power in the previous section

must be converted to quantities that can be compared

with the limits displayed on the ISO Norms. So in

this section, the calculations required for this conver-

sion will be presented along with the limits applica-

ble to the wavalength of the radiation in question, ac-

cording to the ISO Standards ISO 15004-2 and ISO

10940 (ISO 15004-2, 2007; ISO 10940, 2009). The

ﬁrst is for light hazard protection of Ophthalmic In-

struments and the second deals with the features of

Fundus Cameras (including light hazard protection).

As the limits provided on the ISO Norms are pre-

sented in irradiance, the areas of the irradiated struc-

tures have to be calculated.

The Field of View obtainable with this prototype

is 45

◦

so, one way to calculate the retinal area illu-

minated is by using the formulas on the Equations 10

and 11 (ISO 10940, 2009):

BIODEVICES 2019 - 12th International Conference on Biomedical Electronics and Devices

ω = 4πsin

(

)

(10)

A = (1.7cm)

× ω

(11)

Where ω is the illumination solid angle in stera-

dians, α is the full cone angle in degrees and A is the

retinal area illuminated in square centimeters. Con-

sidering a Field of View of 45

◦

the solid angle is 0.478

sr and consequently the retinal area is 1.381 cm

Knowing the area it is possible to calculate the ir-

radiance with the formula:

E =

dΦ

(12)

Where E is the irradiance given in mW/cm

, Φ is

the radiant power given in mW, referred as P

Real

the previous subsection, and A is the area illuminated

in square centimeters. The radiant power was consid-

ered to be constant all over the area, so the equation

becomes:

E =

(13)

Therefore, the values for the maximum level of

intensity for the peaks of both LEDs are:

• Retinal Irradiance White - 439 nm Peak =

0.109 mW/cm

• Retinal Irradiance White - 555 nm Peak =

0.232 mW/cm

• Retinal Irradiance NIR - 820 nm = 0.413

mW/cm

For the other eye structures the utmost scenario

was considered in which the area for calculation of the

irradiance was the area of the smallest circle of illu-

mination possible with the prototype. In other words,

was considered the area of the circle of illumination

at the focal distance from the prototype. It was empir-

ically calculated to be of about 0.071 cm

, resulting

from a circle with 3 mm diameter.

• Corneal and Anterior Segment Irradiance

White - 439 nm Peak = 2.113 mW/cm

• Corneal and Anterior Segment Irradiance

White - 555 nm Peak = 4.521 mW/cm

• Corneal and Anterior Segment Irradiance NIR

- 820 nm = 8.028 mW/cm

As for some hazards the irradiance is weighted

according to the wavelength of the radiation, the co-

efﬁcients for both thermal, R(λ), and photochemical

aphakic hazard, A(λ) provided by the ISO standards,

have to be considered for each peak wavelength ex-

amined (ISO 15004-2, 2007):

• White LED, 439 nm Peak: R(λ) - 1; A(λ) - 1;

• White LED, 555 nm Peak: R(λ) - 1; A(λ) -

0.0078;

• NIR LED, 820 nm Peak: R(λ) - 0.58; A(λ) - 0;

Considering the principle of operation of the pro-

totype described before, ISO norms compliance was

addressed for 4 different situations.

• Continuous NIR LED at maximum intensity;

• Continuous White LED at maximum intensity;

• Pulsed White LED at maximum intensity;

• Real acquisition simulation with the NIR LED

at maximum intensity, and pulsed White LED at

maximum intensity.

The distinction between continuous and pulsed

occurs because in the ISO norms it is considered that

when an instrument emits a pulse or a set of pulses

that last less than 20 seconds the limits that must be

calculated are different from continuous wave instru-

ments (ISO 15004-2, 2007). For each conﬁguration

different hazards must be calculated and are going to

be presented in the next subsections.

3.1 Continuous NIR

For this conﬁguration the values that have to be cal-

culated are:

• Unweighted Corneal and Lenticular Infrared

Radiation Irradiance, E

IR-CL

;

IR-CL

2500

∑

770

× ∆λ (14)

With,

dΦ(λ)

dA × dλ

Φ(λ)

A × ∆λ

(15)

Where E

is the spectral irradiance given in

mW/(cm

.nm), Φ is the radiant power of the ra-

diation, given in mW, A is the area illuminated in

square centimeters and ∆λ is the wavelength inter-

val on which the irradiance was measured. This

approximation will be considered on every haz-

ard and is only possible because the radiant power

was considered to be constant all over the area and

all the emission wavelengths near the peak, were

considered to be the peak itself. Due to this ap-

proximation it is mathematically possible to ig-

nore the dependence on the value of ∆λ, as can be

observed next.

Light Hazard Measurement on an Ophthalmic Instrument

IR-CL

2500

∑

770

Φ(λ)

A ×



∆λ



∆λ

IR-CL

2500

∑

770

Φ(λ)

IR-CL

2500

∑

770

IR-CL

= 8.028 mW/cm

(16)

• Unweighted Anterior Segment Visible and In-

frared Radiation Irradiance, E

VIR-AS

;

VIR-AS

1200

∑

380

× ∆λ (17)

There is only one emitted peak in the NIR spec-

trum, with 820 nm as center wavelength. So, the

result for this hazard is equal to E

IR-CL

, therefore

VIR-AS

= 8.028 mW/cm

• Weighted Retinal Visible and Infrared Radia-

tion Thermal Irradiance, E

VIR-R

;

VIR-R

1400

∑

380

× R(λ) × ∆λ (18)

VIR-R

1400

∑

380

E × R(λ)

VIR-R

= 0.413 × 0.58

VIR-R

= 0.240 mW/cm

(19)

3.2 Continuous White

For this conﬁguration the values that have to be cal-

culated are:

• Weighted Retinal Irradiance for Photochemi-

cal Aphakic Light Hazard, E

A-R

;

A-R

700

∑

350

× A(λ) × ∆λ (20)

A-R

700

∑

350

E × A(λ)

A-R

= 0.109 × 1 +0.232 × 0.0078

A-R

= 0.109 + 0.00181

A-R

= 0.111 mW/cm

(21)

• Unweighted Anterior Segment Visible and In-

frared Radiation Irradiance, E

VIR-AS

; The

method for the calculation of this hazard was al-

ready demonstrated for the NIR LED in the previ-

ous subsection, so only the result will be shown.

VIR-AS

1200

∑

380

VIR-AS

= 2.113 + 4.521

VIR-AS

= 6.634mW/cm

(22)

• Weighted Retinal Visible and Infrared Radia-

tion Thermal Irradiance, E

VIR-R

; The method

for the calculation of this hazard was already

demonstrated for the NIR LED in the previous

subsection so, only the result will be shown.

VIR-R

1400

∑

380

E × R(λ)

VIR-R

= 0.109 × 1 +0.232 × 1

VIR-R

= 0.341mW/cm

(23)

3.3 Pulsed White

For the pulsed mode, information on the duration of

each pulse is necessary. In the current prototype, the

user can select the duration of ﬂash. Was veriﬁed em-

pirically that 0.095 seconds produced fairly good re-

sults when imaging an eye model so, it will be con-

sidered the time of the pulse for comparison with the

ISO Norms comparison. The energy of the ﬂash nor-

mally used is the maximum intensity possible for the

White LED, so irradiance values are the same as those

for the continuous white calculations (Section 3.2).

For the Pulsed Intruments Hazards the results to

be compared will take into consideration the spectral

radiant exposure, H

, which can be derived from the

Irradiance, E

, according to the following equations.

H =

∆t

E × dt = E × ∆t (24)

And so,

= E

× ∆t (25)

Where E

is spectral irradiance, H

is spectral ra-

diant power and ∆t is the duration of the exposure to

the radiation also named exposure time. The value of

is given in mJ/(cm

.nm).

For the same reason that for the continuous

sources, ∆λ is not relevant. For this conﬁguration the

values that have to be calculated are:

• Weighted Retinal Visible and Infrared Radia-

tion Radiant Exposure, H

VIR-R

;

VIR-R

1400

∑

380

× ∆t) ×R(λ) × ∆λ (26)

BIODEVICES 2019 - 12th International Conference on Biomedical Electronics and Devices

VIR-R

1400

∑

380



Φ(λ)

× ∆t



× R(λ)

VIR-R

1400

∑

380

E × ∆t × R(λ)

VIR-R

= 0.109 × 0.095 ×1 + 0.232 ×0.095 × 1

VIR-R

= 0.0324 mJ/cm

(27)

• Unweighted Anterior Segment Visible and In-

frared Radiation Radiant Exposure, H

VIR-AS

;

VIR-AS

1400

∑

380

× ∆λ (28)

VIR-AS

1400

∑

380



Φ(λ)

× ∆t



VIR-AS

1400

∑

380

E × ∆t

VIR-AS

= 2.113 × 0.095 +4.521 × 0.095

VIR-AS

= 0.630 mJ/cm

(29)

3.4 Real Acquisition Simulation

Lastly, it is necessary to test the prototype under real

acquisition conditions in order to predict if it will ful-

ﬁll all the safety requirements on real use cases. As

was previously referred, the prototype uses NIR light

when the operator is performing the retinal align-

ment so a good image of the retina without provok-

ing mydriasis can be obtained and only acquires the

image with white light when observes referable opti-

cal structures for the disease being diagnosed (optic

disk, vessels...). Consequently, a good simulation of

an intended use case can be to consider that the ex-

amination takes about 5 minutes, and 5 images of the

retina are acquired. According to the ISO Norms, in

case of consecutive use of different light sources, the

limit for radiation reaching each surface of the eye is

given by (ISO 15004-2, 2007):

(E,H,L)

Limit

(E,H, L)

Limit

+ ... +

(E,H, L)

Limit

≤ 1 (30)

Where E is the irradiance, H is the radiant expo-

sure, L is the radiance and i is the i

source. As the

irradiance and the radiance are analogous, measure-

ments on only one of them are required. In the work

here presented only irradiance was used.

For pulsed sources the results obtained must be

multiplied by the number of pulses considered so,

when ﬁve pulses are considered the spectral exposure

results become:

5 Pulses

= 5 ∗ H

Single Pulse

(31)

As a consequence :

• Result H

VIR-R

- 0.162 mJ/cm

• Result H

VIR-AS

- 3.15 mJ/cm

The ratio presented above must be veriﬁed for dif-

ferent groups of hazards accordingly to the structure

they refer to, so the following equations have to be

veriﬁed.

• Retinal Hazards:

Result E

VIR-R

Limit E

VIR-R

Result H

VIR-R

Limit H

VIR-R

0.240

Limit E

VIR-R

0.162

Limit H

VIR-R

(32)

• Anterior Segment Hazards:

Result E

VIR-AS

Limit E

VIR-AS

Result H

VIR-AS

Limit H

VIR-AS

8.028

Limit E

VIR-AS

3.15

Limit H

VIR-AS

(33)

• Corneal and Lenticular Hazards:

Corneal and Lenticular Hazards only consider ra-

diation emitted between 770 and 2500 nm, so the

pulsed white light does not present an hazard for

these structures.

As for all the other conﬁgurations, the results will be

presented in the following section (Section 4).

4 RESULTS AND DISCUSSION

In this Section the comparison of the results obtained

for the prototype with the norms currently considered

for Ophthalmic Devices will be demonstrated. Haz-

ards results and limits for the modes of operation of

the prototype are presented in Table 3. The compar-

ison between the results of the measurements with

the limits was performed for the maximum intensity

(100%), even though lower levels of light can also al-

low the acquisition of fairly good images.

Besides the certiﬁcation that all the results are be-

low Group 1 limits it is necessary to perform the cal-

culations for the real acquisition conditions in order

to predict if the prototype will fulﬁll all the safety re-

quirements on real use cases. As previously stated in

subsection 3.4, in case of consecutive use of differ-

ent light sources, the following equations have to be

veriﬁed (ISO 15004-2, 2007):

Light Hazard Measurement on an Ophthalmic Instrument

Table 3: Group 1 ISO Norms Compliance. The cells ﬁlled in green represent the Hazards for which the results calculated

were below the Group 1 Limit.

Mode Hazard Result Group 1 Limit Compliance

Continuous NIR

IR-CL

8.028 mW/cm

20 mW/cm

VIR-AS

8.028 mW/cm

4000 mW/cm

VIR-R

0.240 mW/cm

700 mW/cm

Continuous White

A-R

0.111 mW/cm

0.220 mW/cm

VIR-AS

6.634 mW/cm

4000 mW/cm

VIR-R

0.341 mW/cm

700 mW/cm

Single Pulse White

VIR-R

0.0324 mJ/cm

= 1.026 J/cm

1026 mJ/cm

VIR-AS

0.630 mJ/cm

25t

= 13.879 J/cm

13879 J/cm

• Retinal Hazards:

0.240

Limit E

VIR-R

0.162

Limit H

VIR-R

0.240

700

0.162

3432

= 0,000390 ≤ 1

(34)

• Anterior Segment Hazards:

8.028

Limit E

VIR-AS

3.15

Limit H

VIR-AS

8.028

4000

3.15

20754

= 0.00216 ≤ 1

(35)

• Corneal and Lenticular Hazards:

For Corneal and Lenticular Hazards there is no

Pulsed ratio because on the acquisition mode the

prototype only emits white light.

The values for the radiant exposure limits were

obtained by replacing the value of t on the equations

that set the limits (H

VIR-R

- 6t

; H

VIR-AS

- 25t

) by

the product of the exposure time for a single pulse and

the number of pulses.

5 CONCLUSION

Considering the results presented on the previous sec-

tion, it is possible to conclude that the EyeFundusS-

cope prototype surpassed the Group 1 Ophthalmic

Device tests, according to the ISO 15004-2 Norm.

Group 1 Instruments can be deﬁned as instruments for

which no potential light hazard exists (ISO 15004-2,

2007).

If the use case presented in Subsection 3.4 that

considers obtaining 5 images during 5 minutes, is not

applicable, and a different amount of ﬂashes are di-

rected to the human eye, the calculations required to

attest the hazard can be easily performed, by replicat-

ing the ones presented in this paper. Considering the

differences of orders of magnitude between the result

and the limit (presented in the Results subsection 4),

it is not likely that a different setup would led to a

change of classiﬁcation of the prototype.

By presenting a method for the measurement of

radiant power that uses devices considerably cost-

effective when compared to the regularly used equip-

ment (Spectroradiometers (Calandra et al., 2017;

Hong et al., 2017)), we believe that the work here pre-

sented may be beneﬁcial for all the researchers that

want to perform similar tests.

To the best of the authors knowledge, in the litera-

ture there is not much information on power measure-

ments for broadband LEDs with photodiodes. Con-

sidering this, the work presented in this paper can be

of considerable importance in the ﬁeld of Biomedi-

cal Metrology and provide valuable knowledge for the

designers of biological and biomedical instruments,

who regularly deal with radiation safety issues.

BIODEVICES 2019 - 12th International Conference on Biomedical Electronics and Devices

ACKNOWLEDGEMENTS

We would like to acknowledge the ﬁnancial support

obtained from North Portugal Regional Operational

Programme (NORTE 2020), Portugal 2020 and the

European Regional Development Fund (ERDF) from

European Union through the project Symbiotic tech-

nology for societal efﬁciency gains: Deus ex Machina

(DEM), NORTE-01-0145-FEDER-000026.

For supporting with the necessary material and

guidance we would also like to acknowledge Lab-

orat

orio de

Optica, Lasers e Sistemas, a contract-

research laboratory within the Physics Department of

FCUL (University of Lisbon).

It is also important to acknowledge Ricardo

Peixoto, Jo

ao Gonc¸alves, Sim

ao Felgueiras and Jo

Oliveira, the Fraunhofer AICOS team members re-

sponsible for assembling and developing the proto-

type.

REFERENCES

Calandra, D. M., Di Martino, S., Riccio, D., and Visconti,

A. (2017). Smartphone based pupillometry: An em-

pirical evaluation of accuracy and safety. In Battiato,

S., Gallo, G., Schettini, R., and Stanco, F., editors,

Image Analysis and Processing - ICIAP 2017, pages

433–443, Cham. Springer International Publishing.

Capit

an-Vallvey, L. F. and Palma, A. J. (2011). Recent de-

velopments in handheld and portable optosensing—a

review. Analytica Chimica Acta, 696(1):27 – 46.

Chen, H. (2005). Method for manufacturing a triple wave-

lengths white led.

DeHoog, E. and Schwiegerling, J. (2009). Fundus cam-

era systems: a comparative analysis. Appl Opt,

48(2):221–228. 19137032[pmid].

Desmet, K. D., Paz, D. A., Corry, J. J., Eells, J. T., Wong-

Riley, M. T., Henry, M. M., Buchmann, E. V., Con-

nelly, M. P., Dovi, J. V., Liang, H. L., Henshel,

D. S., Yeager, R. L., Millsap, D. S., Lim, J., Gould,

L. J., Das, R., Jett, M., Hodgson, B. D., Margo-

lis, D., and Whelan, H. T. (2006). Clinical and ex-

perimental applications of nir-led photobiomodula-

tion. Photomedicine and Laser Surgery, 24(2):121–

128. PMID: 16706690.

Dyble, M., Narendran, N., Bierman, A., Klein, T., Dy-

ble, M., Narendran, N., Bierman, A., and Klein, T.

(2005). Impact of Dimming White LEDs : Chromatic-

ity Shifts Due to Different Dimming Methods.

Eells, J. T., Wong-Riley, M. T., VerHoeve, J., Henry, M.,

Buchman, E. V., Kane, M. P., Gould, L. J., Das, R.,

Jett, M., Hodgson, B. D., Margolis, D., and Whelan,

H. T. (2004). Mitochondrial signal transduction in ac-

celerated wound and retinal healing by near-infrared

light therapy. Mitochondrion, 4(5):559 – 567. Mito-

chondrial Medicine - Developing the Scientiﬁc Basis

to Medical Management of Mitochondrial Disease.

Feder, R. S., Olsen, T. W., Prum, B. E. J., Summers, C. G.,

Olson, R. J., Williams, R. D., and Musch, D. C.

(2016). Comprehensive adult medical eye evaluation

preferred practice pattern guidelines. Ophthalmology,

123(1):P209–P236.

Hammer, M. and Schweitzer, D. (2002). Quantitative reﬂec-

tion spectroscopy at the human ocular fundus. Physics

in Medicine & Biology, 47(2):179.

Hong, S. C., Wynn-Williams, G., and Wilson, G. (2017).

Safety of iphone retinal photography. Journal of

Medical Engineering & Technology, 41(3):165–169.

PMID: 27924670.

ISO 10940 (2009). Ophthalmic instruments - Fundus cam-

eras. Standard, International Organization for Stan-

dardization, Geneva, CH.

ISO 15004-2 (2007). Ophthalmic instruments - Fundamen-

tal requirements and test methods.Part 2: Light Haz-

ard Protection. Standard, International Organization

for Standardization, Geneva, CH.

Kelbsch, C., Maeda, F., Strasser, T., Blumenstock, G.,

Wilhelm, B., Wilhelm, H., and Peters, T. (2016).

Pupillary responses driven by iprgcs and classical

photoreceptors are impaired in glaucoma. Graefe’s

Archive for Clinical and Experimental Ophthalmol-

ogy, 254(7):1361–1370.

Malinen, J., K

ans

akoski, M., Rikola, R., and Eddison, C. G.

(1998). Led-based nir spectrometer module for hand-

held and process analyser applications. Sensors and

Actuators B: Chemical, 51(1):220 – 226.

Patrick, D. and Fardo, S. (2000). Industrial electronics: De-

vices and systems.

Preece, S. J. and Claridge, E. (2002). Monte carlo mod-

elling of the spectral reﬂectance of the human eye.

Physics in Medicine & Biology, 47(16):2863.

Rahman, M. Q., Rotchford, A. P., and Ramaesh, K. (2013).

Catatrac: a novel red light-emitting diode device for

screening cataracts in the developing world. Eye

(Lond), 27(1):37–41. 23099916[pmid].

Ready, J. F. (1971). Chapter 2 - measurement techniques.

In Ready, J. F., editor, Effects of High-Power Laser

Radiation, pages 33 – 65. Academic Press.

Rothem, E. Measuring led power and irradiance with cali-

brated photodiodes. Technical report, Ophir.

Rukmini, A. V., Milea, D., Baskaran, M., How, A. C., Per-

era, S. A., Aung, T., and Gooley, J. J. (2015). Pupil-

lary responses to high-irradiance blue light correlate

with glaucoma severity. Ophthalmology, 122(9):1777

– 1785.

Saltmarche, A. E., Naeser, M. A., Ho, K. F., Hamblin,

M. R., and Lim, L. (2017). Signiﬁcant improvement

in cognition in mild to moderately severe dementia

cases treated with transcranial plus intranasal photo-

biomodulation: Case series report. Photomed Laser

Surg, 35(8):432–441. 28186867[pmid].

Schubert, E. (2018). Light-Emitting Diodes (3rd Edition).

Senior, J. (1986). Optical Fiber Communications: Princi-

ples and Practice. Prentice Hall International (UK)

Ltd., Hertfordshire, UK, UK.

Light Hazard Measurement on an Ophthalmic Instrument

Shen, B. Y. and Mukai, S. (2017). A portable, inexpensive,

nonmydriatic fundus camera based on the raspberry

 computer. Journal of Ophthalmology, 2017(3):5.

Tran, K., Mendel, T. A., Holbrook, K. L., and Yates, P. A.

(2012). Construction of an Inexpensive , Hand-Held

Fundus Camera through Modiﬁcation of a Consumer

“ Point-and- Shoot ” Camera. 53(12):7600–7607.

Wang, Y., Zekveld, A. A., Naylor, G., Ohlenforst, B.,

Jansma, E. P., Lorens, A., Lunner, T., and Kramer,

S. E. (2016). Parasympathetic nervous system dys-

function, as identiﬁed by pupil light reﬂex, and its pos-

sible connection to hearing impairment. PLOS ONE,

11(4):1–26.

Yuvacı, s., Pangal, E., Yuvacı, S., Bayram, N., Atas¸,

M., Bas¸kan, B., Demircan, S., and Akal, A. (2015).

An evaluation of effects of different mydriatics on

choroidal thickness by examining anterior chamber

parameters: The scheimpﬂug imaging and enhanced

depth imaging-oct study. Journal of Ophthalmology,

2015:1 – 6.

BIODEVICES 2019 - 12th International Conference on Biomedical Electronics and Devices