The dark art of light measurement: accurate radiometry for low-level light therapy

Electromagnetic radiation can be considered as both a wave and a particle (depending on how it is measured), which transports energy through space. This energy can be absorbed by physical objects and converted into other forms such as thermal or electrical energy (solar cells). For example, in photographic light meters, incident visible light causes electric current flow when the radiant light energy is transferred to electrons as kinetic energy, from which light power can be inferred (‘Photodiodes and power meters’). Similarly, for LLLT, light is transferred to cells as radiant energy that modulates cellular responses via CCO absorption and is analogous to photosynthesis, whereby the action of light, and light alone, directly stimulates cell responses. Whilst radiant energy is important [112], the parameter alone is not enough to determine treatment efficacy since an infinite combination of irradiance and exposure could lead to similar radiant energies. Radiant energy is denoted as Q and expressed in Joules (J). Spectral radiant energy accounts for monochromatic (such as a single wavelength laser) and polychromatic sources (for example, lights which emit over a range of wavelengths) and is defined as radiant energy per unit wavelength interval at wavelength, λ:

‘Flux’ or ‘power’ describes the time rate of flow of radiant energy. This is a parameter that is usually reported within LLLT literature through manufacturers’ information or measured using power/energy meters (Tables 1 and 2). Metaphorically, this describes the ‘potency’ of the light and although important, does not provide adequate information concerning spectral and spatial distribution of the energy or the actual irradiance delivered to the target site. Lack of spatial information assumes uniform irradiance over the output area [113, 114], which can be far from accurate, especially considering the true irradiance across an active beam area (Fig. 2). This may differ according to the type of light source, e.g. the common elliptical profile of lasers (Fig. 2b) or the non-unform irradiance distribution of LEDs (Fig. 2c) and the distance from the light tip to the target area. Although power or radiant flux should be reported, it only partially describes irradiation parameters that are necessary for complete information relevant to LLLT research. Radiant power or flux has units of Joules per second (J/s) or watts (W) and is defined as:

The radiant flux per unit wavelength interval at wavelength, λ is given the term spectral radiant flux and is defined as:

Beam area is often referred to as ‘spot-size’ in the LLLT literature [39, 62, 91, 103]. However, the term beam area should be preferred over spot-size and reported in square centimetres. The term spot is usually descriptive of a circular shape and size is ambiguous although units may remove ambiguity. As mentioned previously, lasers may emit an elliptical beam, which would significantly affect the area calculation (Fig. 2) and lead to misinterpretation of irradiation parameters. In cases where the beam area is non-circular, or of circular Gaussian, the beam area and/or diameter can be accurately determined using techniques such as beam profilometry which will be discussed in ‘Light measurement/detectors’. Nonetheless, this review finds that a significant number of LLLT studies fail to report beam area (Tables 1 and 2), a key parameter that should be reported in all LLLT studies.

The radiant beam area acting on a target site is likely to significantly influence biological response in both in vivo, in vitro and clinical studies. Although systemic and local responses to LLLT irradiation have been reported [115] in vivo, beam area is also important for dosing and radiometric calculations. In vitro, a localised effect of light irradiation is likely to result in a significant biological response. Consequently, if the beam area is much smaller than the target culture area, then only a proportion of the host cells will be irradiated, attenuating the measured biological response and possibly resulting in a false-negative result. Therefore, a suggested good practice would be to ensure the whole culture well is irradiated evenly with a round, flat top beam.

The radiant flux per unit area received by a surface from any direction can be termed irradiance (Fig. 3a). However, sometimes this is confusingly termed power density or intensity in the LLLT literature, and this does not distinguish between irradiance or light arriving (irradiance) or that leaving a surface (exitance). Irradiance is defined as:

where dϕ is the radiant flux and dA is differential area. The measured flux can also be that leaving the surface from any direction due to emission and/or reflection (Fig. 3b) and is given the term radiant exitance and defined as:where dϕ is the radiant flux leaving and dA is differential area where dϕ is leaving from. A possible use of this terminology in LLLT could be to describe reflection off biological tissue such as dentine, bone and other tissues.

Commonly, irradiance is not verified by researchers and only reported from manufacturers’ quoted values or is calculated using values gained from inadequate measurement tools (Tables 1 and 2). Manufacturers’ values are typically measured at the aperture or fibre tip, but because the beams are highly divergent, they are not useful for studies where the beam may be projected to a target such as in an in vitro study. Manufacturers’ specific methods of measurement are also rarely divulged. Irradiation, in vivo or in in vitro cultures, likely occurs through tissue or culture plastics at distances greater than 0 mm. Therefore, replicating and measuring key light parameters at, or through, relevant targets at specific distances and geometry is far more accurate and clinically and experimentally relevant than using manufacturers’ data alone.

Spectral irradiance is the radiant flux per unit wavelength interval at wavelength λ and can be defined for both irradiance and radiant exitance with the following equations:

Although the most straightforward irradiation parameter to measure, exposure time (s) is not always reported in the literature. Sixteen per cent (12/74) of articles reviewed in this study failed to report exposure duration (Tables 1 and 2), which is likely due to a common misconception that wavelength, flux (radiant energy) and fluence are all that are necessary to replicate a successful treatment [29]. Exposure duration is a key component of ‘dose’, which is the product of irradiance and exposure time and should always be separately defined (Section ‘Radiant exposure and exposure reciprocity’).

In addition to reporting exposure time, when multiple exposures are performed, the number of treatment sites, the number of exposures and the interval between exposures should also be reported in order to fully describe the treatment protocol [29, 30].

The energy delivered per unit area of cells during light stimulation for LLLT is also an important parameter since the efficacy of the treatment would depend on the irradiance delivered over a given area. The quantity areal, measured in joules per square centimetre and is often incorrectly termed energy density, should only be used for volumetric energy deposition (J cm⁻³). The proper terminology for the total amount of energy delivered per unit area is ‘radiant exposure’, or more commonly termed fluence in the LLLT literature, where H (the radiant exposure or fluence) is defined as the integral of the irradiance from Eq. 4

However, many researchers merely quote radiant exposure (sic. energy density) as an expression of dose within the literature with missing irradiance (W/cm²) or exposure duration (s) values (or even both; Tables 1 and 2). This is potentially unreliable, as it assumes an inverse correlation between the effects of irradiance and exposure duration.

The Bunsen-Roscoe ‘Law of Reciprocity’ states that photochemical reactions will be independent of irradiance and exposure time with the effects being directly proportional to the total energy delivered [116‐118]. Although it can be assumed that this law is valid for photochemical reactions within a certain dose range, photobiological responses of cells and tissue usually involve a sequence of interacting biological reactions making a linear dose-time relationship less likely. A true reciprocal relationship between irradiance and time would achieve similar therapeutic effects regardless of how radiant exposure was achieved (e.g. 20 s at 100 mW/cm² would exhibit similar therapeutic effects compared with 200 s at 10 mW/cm² or 80 s at 25 mW/cm²). However, although an effective radiant exposure for a specific cell type is an important, and a largely unknown quantity in LLLT, the individual parameters (irradiance and time) are critical and should also be defined. Notably, if the irradiance is too low and/or the delivery time too short, any significant beneficial effect may not be realised or even reduced [19, 81]. Furthermore, if the irradiance is too high or the irradiation time is too long, any significant benefit may also be attributed to heat, or even sometimes produce inhibitory, rather than therapeutic effects [19, 68]. Thus, any useful concept of exposure reciprocity may not be applicable in biological systems such as LLLT, since treatment modalities may only be effective within a window of specific irradiation parameters [19]. However, using similar radiant exposure by varying the combination of irradiance and time, and its effect on stimulatory/inhibitory cell responses is not fully understood and warrants a systematic approach to further understanding of the photobiomodulation of different cell types.

The pulse frequency is the number of pulses of a repeating signal in a specific time frame and is usually measured pulse per second (Hz), thus pulse operation of lasers or LEDs is not classified as a continuous wave. This type of operation is beneficial for heat dissipation and to achieve high peak irradiances, but since there is an on/off period, dosing parameters such as radiant energy and radiant exposure are affected which may affect the efficacy of LLLT [52, 64, 68]. For example, if the irradiation was pulsed to deliver light at 0.5 s intervals, then only half the energy would be delivered compared with continuous delivery at similar irradiance and exposure time. Thus, when pulsing regimes are utilised, the peak irradiance should be defined along with pulse frequency and the on/off durations as previously recommended [29, 30].

A spectrometer is an instrument used to measure the properties of light over specific portions of the electromagnetic spectrum and provides a useful system to analyse spectral characteristics critical for LLLT research. Spectrometers are coupled with flexible, transparent optical fibres of varying diameters made from high-quality glass that function as waveguides or light guides to transmit light between the two fibre ends. Opaline cosine correctors are usually attached, which have diffusing material apertures allowing light measurement normal to its surface with 180° field of view (Fig. 4a). Whilst cosine corrector probes provide a cheap, versatile, robust and reliable method of light measurement, the measurement accuracy is limited when analysing large light sources due to its small collection area and its 180° field of view. Alternatively, integrating spheres of varying diameter and port size (dependent upon the source size) can be used, which consist of hollow spherical cavities covered with diffuse white reflective coating. Spheres can be used to capture and measure light radiated in all directions from the light source as light scattered by the interior of the integrating sphere is evenly distributed over all angles (Fig. 4b). However, measurements using integrating spheres are limited by the size of the sphere and the size of the light source intended to be measured. Nevertheless, the fibres and cosine corrector (or integrating sphere) collectively become an optical probe, which can be calibrated using a photometric standard or calibrated light source to National Institute of Standards and Technology (NIST) standards providing an accurate measurement system known as a spectroradiometer.

Light is captured through the cosine corrector or integrating sphere and travels through the optical fibre into the spectrometer. The core of the spectrometer is formed by a diffraction grating which splits radiant light into its spectral components and projects the diffracted elements onto a detector. Computer software is used to calculate all radiometric, photometric and colourimetric quantities from spectral data. Two types of spectrometers exist, an array type, which has a fixed diffraction grating and a detector array, and a scanning spectrometer, which has a single detector and a rotating diffraction grating (Fig. 4c). Spectrometers are popular light measurement systems for many aspects of photonics research although rarely used in LLLT studies (Table 2). However, there are other limitations of fibre-coupled spectrometers which use cosine correctors and integrating spheres, primarily that power uniformity within the incident beam is assumed and the power distribution of light across the exit diameter of large light sources cannot be measured [114]. For example, using a typical cosine corrector diameter of 4 mm to measure an incident beam diameter of 10 mm, the outer 6 mm of the beam will not be captured by the sensor. Light sources used for LLLT typically have a Gaussian distribution and therefore if the irradiance is measured centrally, power is assumed to be equal over the whole area and the measured irradiance would be overestimated. Error is increased with the increasing ratio of beam diameter to probe diameter (or vice versa) and researchers need to cautiously interpret data in such situations. Ideally, researchers should employ methods that will adequately measure all of the light by considering the projected beam area on a target and the distance that the beam is applied from. For example, if LLLT studies are performed by irradiating culture dishes or tissue samples from a specific distance, then the experimental light measurement methods should simulate this to accurately analyse light properties at the target site (irradiance), i.e. measure light received by cells, not what the light outputs. The effects of absorption, scattering and reflection by media, cell culture plasticware and other materials/tissue on spectral irradiance at the target site are critical and should be carefully considered, i.e. for cell culture work, the irradiance delivered on the culture area through plasticware should be measured to accurately determine the irradiance delivered to cells.

Spectrometers can also be used to measure absorption characteristics of specific cellular chromophores or photorecepetors localised in the mitochondria that are responsible for the absorption of light. A light source emitting multiple wavelengths is focused on to a sample which attenuates light through absorption, scattering and reflection of the incident light. The action spectra, a plot of relative effectiveness of different wavelengths, which is believed to mimic the absorption spectrum of CCO, has been reported by Karu et al. [119] and indicates several effective bands relating to the copper complexes of CCO. Thus, by recording this attenuation of light for various wavelengths, an absorption spectrum can be obtained and potential therapeutic windows for LLLT can be identified for specific tissue.

A photodiode (detector) is responsive to optical input from UV to near infrared radiation and operates as a photoelectric converter generating a current that is proportional to the incident light. A photon of sufficient energy creates an electron hole pair by a mechanism known as the inner photoelectric effect that is dependent upon the efficiency of the photodiode. Quantum efficiency is dependent upon many factors, but in general if the energy of the photon is greater than the energy gap of the device, these photons will be absorbed very near the surface where the recombination rate is high and will contribute to a photocurrent. Thus, the photocurrent produced by the photodiode is proportional to the power of the light which can be measured directly by a ‘power meter’ which uses an operational amplifier circuit known as a transimpedance amplifier.

Although this type of measurement system is most popular within LLLT literature (Table 1), measurements from these devices should be interpreted cautiously. The spectral sensitivity differs with wavelength due to the quantum efficiency of the photodiode and generally has a better response at longer wavelengths. Thus, if broadband light sources are measured, the power emitted at short and longer wavelengths maybe be under- or over-estimated, respectively. Photodiodes also assume power uniformity across the beam and do not therefore effectively characterise the distribution of light for the same reasons described previously (‘Spectro(radio)meters’). Although reproducible measurements can be made at the sub-picoampere regime, the response time is limited by the sensor size, which slows as surface area increases. Furthermore, the detectors are usually made from fragile or sensitive materials such as silicon (Si; 190–1100 nm), germanium (Ge; 400–1700 nm), indium gallium arsenide (InGa; 800–2600 nm), lead(II) sulphide (<1000–3500 nm) or mercury cadmium telluride (400–14000 nm) which can be prone to damage and therefore measurements usually must be made without contact with the detector. Consequently, even small distances are likely to result in a loss of power due to divergence which will reduce the measured power. More so and unlike spectrometers, photodiodes do not provide spectral information and are usually only limited to power readings (W) and crude irradiance measurements (W/cm²) based on the sensor size or inputted tip or beam area values. However, photodiodes are relatively insensitive to temperature fluctuation (not critical for LLLT as low powered sources are used) and their main and unique advantage lies in their ability to measure very small optical powers which is specifically useful for basic light characterisation in LLLT studies.

Thermopiles are essentially thermal sensors, which are best suited for measuring constant wave (CW) laser power, average power in pulsed lasers or the energy of long pulses. Thermopiles are robust, reliable and are a well-established method to measure light energy. They can be considered as an array of miniature thermocouple junctions connected in series as differential pairs. These differential pairs make up cold and hot junctions that are connected by alternating n-type and p-type materials. Thermopiles operate by using temperature differences to create a voltage, which is correlated to the temperature gradient between the hot and cold junctions and proportional to the light energy. These systems are particularly useful for measuring high powered sources (>1 Watt) which can damage other types of sensors. Thermopiles are made from materials such as antimony (Sb), bismuth (Bi), poly-silicon gold (Au) or alumininium (Al) and operate over a broad spectral range (200–20000 nm). Thermopiles tend to be more accurate than photodiodes but measurement sensitivity is reduced at low power. Since thermopiles work by using temperature gradients, they can be used to map irradiance distribution and offer uniform spatial response that is unaffected by changes in beam size, position or uniformity unlike photodiodes and spectrometer based systems which rely on inputted beam area values for irradiance calculation. However, although this type of sensor validly characterises the distribution of light and provides an accurate measure of important light properties such as irradiance and beam area, like the photodiode, it does not provide spectral information. Furthermore, response times are slow (generally a few seconds), which could be problematic in time-dependent experiments and thus are only really capable of measuring average powers. In addition, since the measurement is based on heat exchange, rapid fluctuations in housing temperature will decrease accuracy.

A charge coupled device (CCD) is an integrated circuit etched onto a silicon surface forming light sensitive elements called pixels. Photons incident on this surface generate charge that is converted into a digital copy of the light pattern. Following appropriate calibration, beam profilometry is very useful for characterisation and quantification of power distribution and irradiance of a given light source and has the advantages of both photodiodes (good response time and unaffected by temperature) and thermopiles (unaffected by beam diameter, good sensitivity and spatial distribution of power and irradiance, which can be used over a large range of power outputs). Light can be collected through lenses and directed onto the CCD sensor which then creates a digital image of the beam. Calibration using pre-determined power values may then be used to calculate the average power delivered to each pixel within the defined beam area to create a mapped irradiance image [120]. This is known as the top-hat factor and can be used to characterise the degree of spatial [114, 120‐122] and spectral [123] uniformity of the power distribution. For system calibration, if the total measured power is calculated (using, for example, a photodiode or thermopile), the power received by each pixel in the detector’s diode array can be calibrated to generate a 2- or 3-dimensional map of irradiance distribution across the active beam area. Therefore, the beam area can be accurately calculated rather than only measuring the light delivery tip diameter by crude methods such as calipers. For example, if the tip size of a LLLT laser device is much larger (7.5 mm diameter) than the actual beam diameter (~0.11 mm diameter; Fig. 5), erroneous irradiance values are inevitably obtained if the active beam area is assumed to be the same as the tip diameter. Thus, the need for standardised beam area calculation is required in LLLT, preferably using the ISO standard method (D4σ or second moment width; ISO 11145 3.5.2 [122]) or 1/e² as suggested previously [29, 30].

Although CCD cameras and beam profilers are widely used for a variety of applications including dental research [114], this method has only been utilised in a limited number of LLLT studies [82, 120]. However, whilst beam profilers provide a relatively accurate measurement system, they are also unable to readily provide spectral information and are sensitive to spectral variation. Accurate and reliable test centres for LLLT research would have a suite of complimentary equipment including spectrometers or integrating spheres, photodiodes and beam profilers.