Journal of Medical Engineering & Technology, Vol. 34, No.

2, February 2010, 97107

Review Conrmation of spectral jitter: a measured shift in the spectral distribution of intense pulsed light systems using a time-resolved spectrometer during exposure and increased uence
C. ASH*{, G. TOWN{ and M. CLEMENT{ {School of Medicine, Swansea University, Swansea, SA2 8PP, UK {GCG Healthcare, Haywards Heath, RH16 2LT, UK
(Received 16 June 2009; revised 19 September 2009; accepted 9 October 2009)

High quality intense pulsed light (IPL) systems can oer simple, safe and eective treatments for long-term hair reduction, skin rejuvenation and removal of benign vascular and pigmented lesions. Considerable dierences in clinical ecacy and adverse eects have been recorded amongst dierent IPL systems despite comparable display settings. This study examines the variation in pulse structures exhibited by several popular professional IPL systems that can cause a spectral change within the broadband output depending on the pulse structure chosen by the system designers. A fast spectrometer was used to capture IPL spectral outputs. A spectral distribution shift that occurs both within a pulse and between pulses is clearly demonstrated and is more prominent with uncontrolled free discharge systems than with square pulsed technology, which provides a constant spectral distribution throughout the pulse duration. Keywords: Pulse duration; Spectral distribution; IPL; Time-resolved spectroscopy

1. Introduction Typically, intense pulsed light (IPL) systems discharge radiation between 510 nm and 1100 nm, with wavelengths ltered as required, depending upon the condition being treated [1]. These visible and near-infrared wavelengths penetrate skin and are absorbed by target chromophores via selective photothermolysis [2] for treatments such as hair reduction [35] and skin rejuvenation [69]. Transient post-treatment erythema, peri-follicular oedema and hyperpigmentation [10] are common treatment endpoints in IPL treatments. There are safety issues regarding such intense light sources on human skin but they are generally considered safer than laser systems [11]. However, cases of permanent side eects [12] and ocular damage [13] have been reported. As the technology is relatively new, long-term eects of IPL treatment are still *Corresponding author. Email: caerwynash@yahoo.co.uk

unknown, with many physicians and scientists recommending further investigation into biological changes and malignant lesions [14,15]. Knowledge of the optical dosimetry characteristics of an IPL device is essential to establish a scientic basis for applications involving lighttissue interaction using an IPL device. There is only limited published literature on the measurement of IPL devices [1622]. Town et al. [23] identied ve key IPL measurement parameters: pulse duration, radiant exposure (uence), spatial prole, spectral output and time-resolved spectral output. Standardization of measurement introduces consistency into a system, lowering the risk of adverse reactions from device malfunction and improving treatment ecacy and reliability [22]. Time-resolved spectral measurement is signicant in optimizing treatment parameters and assessing clinical

Journal of Medical Engineering & Technology ISSN 0309-1902 print/ISSN 1464-522X online 2010 Informa UK Ltd. http://www.informaworld.com/journals DOI: 10.3109/03091900903402089

98

C. Ash et al.

and ocular hazards [2022]. Dierent absorption characteristics of chromophore targets in skin (gure 1) require that sucient energy be delivered in the wavelength range that is most likely to be biologically eective, whilst minimizing adverse reactions by ltering potentially damaging wavelengths being delivered to the tissue matrix. Information about the true spectral footprint during IPL exposure makes it possible for an ocular hazard assessment to be completed. Ocular hazard assessment determines the risk of exposure directed onto the human eye, which is predominantly sensitive to wavelengths in the range 380 1400 nm [28]. A light source that emits a spectrum that changes within a pulse and from one pulse to another, is unlikely to generate an ecient reproducible response with each discharge in human tissue. At increased current density within the ashlamp the spectral distribution is shifted towards lower wavelengths. Vaynberg et al. [24] was the rst to state that the spectral distribution of an IPL system is proportional to the input current. The ability to control the ashlamp current facilitates more accurate management of temperature dierences between target and surrounding tissue, thus improving selectivity and potentially improving clinical outcomes. IPLs can be categorized into two main types by the method used to generate and deliver the energy required for

light-based treatments, i.e. free discharge and square pulse. A free discharge system applies a large electrical charge to a capacitor or a number of capacitors in parallel, then discharges the entire stored energy directly though the ashlamp; this discharge prole is characterized by a rising/ falling slope. It was theorized that as the energy supplied to the ashlamp varies, the emission characteristics change through a shift in the emitted wavelengths [25]. This eect of spectral jitter was rst proposed by Clement et al. [25] where the spectral output varies during a pulse or pulse train and between energy levels. Thus, by the application of a constant current, improved targeting of IPL treatments based on selective photothermolysis is achieved. A constant (square) current spectral emission over the duration of the pulse controls the temperature dierence between the target and surrounding tissue more eciently than can be achieved with a conventional free-discharge IPL treatment. Whilst the role of measurement of pulse duration and pulse prole for lasers and intense pulsed light sources has been recognized recently [2,18,26], only a few studies to date have attempted to document methods for measuring IPL pulse durations or to examine in detail the timeresolved spectral output of IPLs across each millisecond of the pulse duration. The objective of this study is to show the eect of spectral jitter with evidence of spectral energy

Figure 1. Absorption coecient of melanin, oxyhaemoglobin, water and porphyrin with a typical IPL discharge emission spectra overlaid for reference. Note: logarithmic axis.

Spectral distribution shift in intense pulsed light systems

99

distribution and spectral stability during an IPL emission using time resolved spectroscopy. 2. Material and methods The authors performed measurements over a 6-month period on three constant-current and 16 free-discharge systems, which were all in daily use in private dermatology clinics and salons in the UK. These included StarLux (Palomar Medical Technologies, Burlington, MA), iPulse (CyDen, Swansea, UK), NovaLight (Ultramed, Geneva, Switzerland), Chromolite (Chromogenex, Llanelli, UK), Crystal 512 (Active Optical Systems, Petach-Tikva, Israel), EllipseFlex/EllipseLight (Ellipse, Hrsholm DK, Denmark), Harmony (Alma Lasers, Caesarea, Israel), ULTRA (Energist, Swansea, UK), BBL (Sciton, Palo Alto, CA), Lumina600 (Lynton Lasers, Cheshire, UK), GPFlash1 (General Project, Montespertoli, Florence, Italy), Plasmalite (Medical Biocare Sweden, Vastra Frolunda, Sweden), Ecolite (Greenton London, London, UK), Quantum/Aculight (Lumenis, Santa Clara, CA), Freedom IPL (Freedom Beauty, Leicester, UK), Trinity (Espansione Marketing Spa, Bologna, Italy) and SkinStation (Radiancy (Israel), Yavne, Israel). The authors previously grouped these individual systems into four distinct categories by their delivery pulse pattern namely, square pulse, free discharge, close pulse stacking, and spaced pulse stacking [21]. One example of each category was selected, analysed and presented as part of this study. Conventional spectrometers need a relatively long sample time, rather like the exposure time on a camera. This averaging eect dampens or eliminates the variations in spectral peaks. Time-resolved spectral measurements make it possible to assess variations in spectral composition during the light pulse, and hence evaluate the quality and consistency of sequential ashes of the IPL. These assessments may then form the basis for hypotheses on improvements that may be made to the pulse duration and spectral pattern of an IPLs output characteristics to produce improved clinical outcomes. The time-resolved spectra in this study were produced using an Ocean Optics HR2000 spectrometer and its counterpart SpectraSuite software (OceanOptics, Dunedin, FL). This software has the capability of sampling a spectrum of light with a minimum integration time of 1 ms by generating 1000 full spectral scans per second. Time-resolved spectral data of IPL outputs were captured and stored with an optical resolution of a monochromatic source measured as full width half maximum (FWHM) resolution of 0.035 nm. This fast spectrometer uses a Sony ILX511 2048-element linear silicon CCD-array detector to capture data into memory every millisecond interfaced to a PC via a USB 2.0 port for later analysis. The HR2000 spectrometer has the facility of stray light correction to compensate for ambient light, which could

otherwise create a slight oset in the results. Every result was recorded with this facility enabled. The spectrometer is externally triggered using a breakout box and because of the relatively short pulse duration of an IPL system, the sampling was taken over an extended time period to ensure capture of the data. The source of the intense light from the IPL system and the spectrometer optical bre was separated by a distance of 150180 cm to prevent saturation of exposed light upon the CCD array within the spectrometer. During testing, suitable broadband protective eyewear was worn by all persons present within the enclosed room. The data was used to provide the spectral contribution of a number of wavelength points from 300 nm to 1000 nm in 50-nm intervals. The spectral distribution was plotted across the pulse duration with a 1-ms resolution, or in the case of multiple pulses the spectral distribution at the peak of the pulse is taken and plotted with pulse number. 3. Results The results are discussed with reference to the category of pulse delivery and the eects of increasing uence. 3.1. Square pulse Figures 2(a) and 2(b) show the corresponding sequence of time-resolved spectral emission views for each millisecond of exposure (iPulse i200, CyDen) exhibiting a sharp cuto lter at 530 nm employed to reduce epidermal absorption. The pulse prole measured is a single pulse of 25 ms duration. Figure 2(b) records the spectral distribution of key wavelengths (3001000 nm) during the pulse duration showing wavelengths proportional and consistent to their neighbouring wavelengths with respect to time. A consistent distribution of wavelength with time may predict more consistent treatment outcomes. 3.2. Free discharge Figures 3(a) and 3(b) present the spectral analysis of Chromolight (Chromogenex) showing a varying output with time that is poorly ltered with 17% of spectral energy below 500 nm, potentially posing a risk to skin and ocular safety. Figure 3(b) shows the spectral distribution of wavelengths changes with respect to time. The 650 nm wavelength increases by 84% whereas 500 nm, 550 nm, 600 nm and 700 nm all decrease by 42% on average from start to nish. 3.3. Close pulse stacking Figures 4(a) and 4(b) record spectral analysis of Ellipse Light (Ellipse) showing a sharp cut-o lter at 600 nm

100

C. Ash et al.

Figure 2. (a) Time-resolved square pulsespectral analysis of a square pulse system with a 530-nm lter for each millisecond of exposure (iPulse i200, CyDen). (b) The spectral distribution of key wavelengths (3001000 nm) showing stability during the pulse duration. employed to reduce epidermal absorption. The pulse prole on the Ellipse Light is of seven sub-pulses stacked closely, decaying with increasing uence. Figure 4(b) shows changes of wavelength contribution during the dierent sub-pulses with respect to time of key wavelengths between 300 nm and 1000 nm. The 950 nm wavelength noticeably increases by 93% whereas 600 nm, 650 nm and 700 nm wavelengths all decrease on average by 23% from rst subpulse to the last. 3.4. Spaced pulse stacking Figures 5(a) and 5(b) show the spectral analysis of a Lumina600 (Lynton Lasers) with a 650-nm cut-o lter

Spectral distribution shift in intense pulsed light systems

101

Figure 3. (a) Free discharge pulsespectral analysis of a poorly ltered free discharge system (Chromolight, Chromogenex). (b) The spectral distribution of key wavelengths (3001000 nm) changing during the pulse duration. employed to reduce epidermal absorption. This IPL system uses ve separate capacitors to discharge the energy in short high-energy pulses. Figure 5(b) shows wavelengths that are consistent with respect to the pulses of ve sequential sub-pulses. In this case, due to the relatively short on-times and long dwell periods any changes to the ashlamp characteristics reverse before the next subpulse. 3.5. Increasing radiant exposure (uence) To explain why the spectral emission changes during delivery of free discharge systems and not with square pulse systems, a test was devised where we took timeintegrated spectral captures of the two respective systems with set pulse duration and varied the uence. A spectral shift in some wavelengths occurs for both systems. Thus the

102

C. Ash et al.

Figure 4. (a) Close pulse stackingspectral analysis of a close pulse stacking system with a 600 nm (Ellipse Light, Ellipse). (b) The spectral distribution of key wavelengths (3001000 nm) vary during the dierent pulses.

spectral distribution is dependant on the current density within the ashlamp. As the current density of free discharge systems change with time, so does the spectral emission. Figures 6(a) and 6(b) present the spectral analysis of a square pulse system with various uence values in the range 618 J cm72 (iPulse i200, CyDen). The pulse duration used during measurements was a single pulse of

25 ms in duration. The graph in gure 6(b) shows that the spectral distribution of wavelengths changes with increasing uence. The 550 nm, 600 nm, 650 nm and 700 nm wavelengths all increase, whereas other wavelengths remain consistent with respect to uence. Figures 7(a) and 7(b) display the spectral analysis of a free discharge system with various energy level values ranging from settings 110

Spectral distribution shift in intense pulsed light systems

103

Figure 5. (a) Spaced pulse stackingspectral analysis of a spaced pulsed stacking system with a 650 nm lter (Lumina, Lynton Lasers). (b) The spectral distribution of key wavelengths (3001000 nm) is stable during the ve sub-pulses.

(Chromolight, Chromogenex). The graph in gure 7(b) shows the spectral distribution of key wavelengths 300 1000 nm with increasing energy level. The spectral distribution of wavelengths changes with increasing uence as 450 nm and 500 nm wavelengths increase whereas 650 nm decreases with respect to increased energy level. For both gures 6(b) and 7(b) the wavelengths most noticeable of

change are within the region of the plasma energy used to create the broadband energy and are solely dependant on the magnitude of uence. Figures 2(a), 3(a), 4(a) and 5(a) show the compromise in spectral distribution systems made to provide pulses of energy that are therapeutic and cautious. The spaced stacking system has the highest cut-o lter at 650 nm,

104

C. Ash et al.

Figure 6. (a) Increasing uence/square pulsespectral analysis of a square pulse system with various uence values ranging 618 J cm72 (iPulse i200, CyDen). (b) The spectral distribution of key wavelengths 3001000 nm with increasing uence (J cm72).

set higher than the other systems to prevent adverse reactions from the high energy interspaced pulses. As a consequence the system is inecient as the excluded wavelengths result in excessive heat being extracted by bulky and costly cooling systems. The example free

discharge system had the lowest cut-o lter, presumably to maximize delivered uence. The four very dierent spectral distributions are important in assessing ocular hazards as the spectral content is weighted against an eye response function.

Spectral distribution shift in intense pulsed light systems

105

Figure 7. (a) Free dischargespectral analysis of a free discharge system pulse with various energy level values ranging 110 (Chromolight, Chromogenex). (b) The spectral distribution of key wavelengths 3001000 nm with increasing energy level. is shown to provide a more consistent spectral distribution during the pulse duration compared with free discharge. Thus the spectral distribution is dependant on the current density within the ashlamp. Although this eect occurs, it is unlikely to be clinically signicant, as uence and pulse dosimetry are much greater dependant factors aecting treatment outcome.

4. Discussion This study has veried a spectral distribution shift occurring both within a pulse and between pulses that is more prominent with uncontrolled free discharge systems. A change in spectral distribution is shown with increasing uence for both systems. However, square pulse technology

106

C. Ash et al.

Our results show that the spectral distribution of free discharge and close stacking systems vary during pulses. Figure 3(b) shows 650 nm increasing by 84% and four other wavelengths decrease by 42% on average; gure 4(b) shows 950 nm increasing over the pulse duration by 93% and three other wavelengths decreasing by 23% on average. Figures 2(b) and 5(b) show spectral stability from square pulse and spaced pulse systems. From the measurements taken, the spectral shift of wavelengths during the period of exposure generally increases in the region of 900950 nm and decreases in the range 500 700 nm. Many systems incorporate high cut-o lters to prevent undesirable epidermal absorption, thus inadvertently concealing this decaying eect in shorter wavelengths. 4.1. Melanin The eect on epidermal melanin during pigmentation lesion treatment where shorter wavelengths greatly absorb has a potential clinical impact as, during exposure, the dissipated wavelengths shift to longer wavelengths that are less well absorbed. The treatment of hair removal is however unlikely to be aected by a change in spectral distribution due to the broad absorption range of melanin. 4.2. Acne Porphyrin absorption has ve distinct peaks, the largest at 424 nm (Sorec band) and four smaller peaks in the Q-band at 500 to 625 nm (gure 1). The decay of emitted wavelengths in this region of the electromagnetic spectrum may negatively impact the treatment of supercial acne where the target bacterium is destroyed by photons delivered in the shorter wavelength region where light penetration is at its shallowest. 4.3. Haemoglobin The largest absorption peak occurs around 578585 nm, this peak being around the pivot pointing spectral distribution where shorter wavelengths decrease and the longer wavelengths increase. Photons need to penetrate the tissue matrix through the epidermal barrier where vessels are located in the upper dermis. Although the exact clinical extent of spectral jitter is currently unknown, a deliberate swing in spectral distribution during exposure could inuence treatment parameters of specic applications. There is one manufacturer of IPL systems advocating a spectral shiftcontrolling property, however the eect of spectral variation is clinically unmeasurable in this case as the pulse duration is drastically altered with respect to spectral shift [27].

5. Conclusions It has been ve years since Clement and colleagues [25] rst proposed a theoretical model involving the physics of the time resolved pulse structure inuencing the output dosimetry of photo therapeutic treatments. This paper presents quantitative results of spectral shift both within pulses and between pulses, and with increasing uence. This spectral shift is more prominent with free discharge systems. A square pulse IPL with a consistent release of therapeutic wavelengths with time suggests a more ecient and consistent treatment outcome with fewer adverse reactions than with other pulse structures. Optical spectral footprints provide valuable information regarding IPL system performance, clinical eciency and patient safety. IPL manufacturers should provide time resolved spectroscopy graphs to users. These measurements could be helpful in determining whether there is a potential impact on ecacy of absorption of light by the primary skin chromophore targets of interest. Acknowledgements The authors would like to thank CyDen Ltd, Swansea, UK for part-funding and provision of equipment used in this study. In addition, we thank Dr Susanna Town, University of Calgary, Canada for review of the manuscript. Declaration of interest: Godfrey Town receives consultancy fees and travel grants from CyDen Ltd. Caerwyn Ash is a PhD student at Swansea University and receives travel grants from the university. He also receives salary from CyDen Ltd.

References
[1] Raulin, C., Greve, B. and Grema, H., 2003, IPL technology: a review. Lasers in Surgery and Medicine, 32, 7887. [2] Anderson, R.R. and Parish, J.A., 1983, Selective photothermolysis: Precise microsurgery by selective absorption of pulsed radiation. Science, 220(4596), 524527. [3] Toosi, P., Sadighha, A., Sharian, A., Gita, M., 2006, A comparison study of the ecacy and side eects of dierent light sources in hair removal. Lasers in Medical Science, 21, 14. [4] Bedewi, A.F., 2004, Hair removal with intense pulsed light. Lasers in Medical Science, 19, 4851. [5] Tse, Y., 1999, Hair removal using a pulsed-intense light source. Dermatologic Clinics, 17, 373385. [6] Sadick, N.S. and Weiss, R., 2002, Intense pulsed-light photorejuvenation. Seminars in Cutaneous Medicine and Surgery, 21, 280287. [7] Bitter, P.H., 2000, Non-invasive rejuvenation of photodamaged skin using serial, full-face intense pulsed light treatments. Dermatologic Surgery, 26, 835842. [8] Kulick, M.I., 2001, Photorejuvenation: using intense pulsed light technology in a cosmetic surgery practice. Aesthetic Surgery Journal, 21, 255258.

Spectral distribution shift in intense pulsed light systems

[9] Fodor, L., Peled, I.J., Rissin, Y., Ramon, Y., Shoshani, O., Eldor, L., Gaiman, A., Ullmann, Y. 2004, Using intense pulsed light for cosmetic purposes: our experience. Plastic Reconstruction Surgery, 113, 1789 1795. [10] Moreno-Arias, G.A., Castelo-Branco, C. and Ferrando, J., 2002, Side eects after IPL photodepilation. Dermatologic Surgery, 28, 11311134. [11] Clarkson, D.M., 2004, Hazards of non coherent light sources as determined by the framework of IEC TR-60825-9. Journal of Medical Engineering & Technology, 28, 125131. [12] Friedman, P.M., Jih, M.H., Burns, A.J., Geronemus, R., KimyaiAsadi, A., Goldberg, L., 2004, Nonphysician practice of dermatologic surgery: the Texas perspective. Dermatologic Surgery, 30, 857863. [13] Sutter, F.K. and Landau, K., 2003, Ocular complication of PhotoDerm VL therapy for facial port-wine stain. Dermatologic Surgery, 29, 111112. [14] Sorg, O., Janer, V., Antille, C., Carraux, P., Leemans, E., Masgrau, E., Saurat, J., Salomon, D., 2007, Eect of intense pulsed light exposure on lipid peroxides and thymine dimers in human skin in vivo. Archives of Dermatology, 143, 363366. [15] Chan, H.H., Yang, C.H., Leung, J.C., Wei, W.I., Lai, K.N., 2007, An animal study of the eects on p16 and PCNA expression of repeated treatment with high-energy laser and intense pulsed light exposure. Lasers in Surgery and Medicine, 39, 813. [16] Ash, C., Town, G.A. and Martin, G.R., 2007, Preliminary trial to investigate temperature of the iPulse intense pulsed light (IPL) glass transmission block during treatment of Fitzpatrick II, IV, V, and VI skin types. Lasers in Medical Science, 22, 49. [17] Clarkson, D.M., 2003, Dual channel light pulse exposure timer. Medical Engineering and Physics, 25, 329334. [18] Clarkson, D.M., 2004, The role of measurement of pulse duration and pulse prole for lasers and intense pulsed light sources. Journal of Medical Engineering & Technology, 28, 132136. [19] Miyake, R.K., Miyake, H. and Kauman, P., 2001, Skin temperature measurements during intense pulsed light emission. Dermatologic Surgery, 27, 549554.

107

[20] Clarkson, D.M., Gill, D. and Odeke, M., 2007, A time resolved intense pulsed light spectral analysis system. Lasers in Medical Science, 2008 Jan;23(1):5964. Epub 2007 Apr 28. [21] Ash, C., Town, G. and Bjerring, P., 2008, Relevance of the structure of time-resolved spectral output to light-tissue interaction using intense pulsed light (IPL). Lasers in Surgery and Medicine, 40, 8392. [22] Eadie, E., Miller, P., Goodman, T. and Moseley, H., 2009, Timeresolved measurement shows a spectral distribution shift in an intense pulsed light system. Lasers Med Sci, 24, 3543. [23] Town, G., Ash, C., Eadie, E. and Moseley, H., 2007, Measuring key parameters of intense pulsed light (IPL) devices. Journal of Cosmetic Laser Therapy, 9, 148160. [24] Vaynberg, B., Panl, S. and Epshtein, V., 2005, Spectrum controlled IPL. Photonic Therapeutics and Diagnostics 2005, SPIE, 5686, 119 125. [25] Clement, M., Kiernan, M., Ross Martin, G.D., Town, G., 2006, Preliminary clinical outcomes using iPulse intense ash lamp technology and the relevance of constant spectral output with large spot size on tissue. Australasian Journal of Cosmetic Surgery, 2, 5459. [26] Ash, C., Town, G., Clement, M., Donne, K. and Daniel, G., 2009, Measurement of wavelength dependant pulse durations and uence variation within a discharge of a broadband intense pulsed light (IPL) system using time-resolved spectroscopy. 29th Annual Conference of the American Society for Laser Medicine and Surgery, National Harbor, Maryland, USA, 15 April 2009; ePoster 508: published online. [27] Min-Wei, C., 2006, Infrared, programmable wavelength light source for permanent hair reduction in skin types IV, Cutera Inc., personal communication. www.cutera.com [28] Clarkson, D.M., 2006, Determination of eye safety lter protection factors associated with retinal thermal hazard and blue light photochemical hazard for intense pulsed light sources. Physics in Medicine and Biology, 51, N59N64.