Hand-held pulsed photothermal radiometry system to estimate epidermal temperature rise during laser therapy

Background/purpose: During laser therapy of port wine stain (PWS) birthmarks in human skin, measurement of the epidermal temperature rise ( D T epi ) is important to determine the maximal permissible light dose. In order to measure D T epi on a speciﬁc PWS skin site, we developed an AC-coupled hand-held pulsed photothermal radiometry (PPTR) system, which overcomes the in vivo measurement limitations of bench-top systems. Methods: The developed hand-held PPTR system consists of an infrared (IR) lens, AC-coupled thermoelectrically cooled IR detector, laser hand-piece holder, and positioning aperture. The raw AC-coupled signal was integrated to obtain a higher signal-to-noise ratio (SNR). The experimental temperature difference ( D T ) calibration was compared with theoretical computations. In vitro and in vivo measurements of D T were performed with a tissue phantom as a function of radiant exposure and human subject as a function of melanin concentration, respectively. Results: The integrated AC-coupled signal provided higher SNR as compared with the raw AC-coupled signal. The experimental D T calibration resulted in good agreements with the theoretical results. The in vitro and in vivo results also presented good agreements with theory. Conclusions: A ﬁber-free, hand-held AC-coupled PPTR system is capable of accurate epidermis temperature rise ( D T epi ) measurements of human skin during pulsed laser exposure.

T O TREAT hypervascular dermatologic conditions such as port wine stain (PWS) birthmarks, laser therapy is commonly performed. The goal of such a therapeutic procedure is to selectively heat a subsurface target (e.g., abnormal blood vessels). As epidermal melanin also absorbs at laser wavelengths typically used for these procedures (e.g., 585-600 nm), a maximum safe radiant exposure (H max ) can be defined, above which epidermal thermal damage would occur. For laser pulse durations on the order of milliseconds, a threshold epidermal temperature (T thresh ) of 70 1C has been assumed (1).
A measure of epidermal heating would provide clinicians with an objective means to determine H max . Pulsed photothermal radiometry (PPTR) (2)(3)(4)(5) can provide accurate measurements of epidermal heating. In PPTR, time-resolved blackbody emission from a sample is measured with a mid-infrared (IR) detection system after pulsed laser exposure. Algorithms have been developed to convert the acquired IR signal to a depth profile of the initial temperature distribution immediately after pulsed laser exposure (2,4); this profile provides information on the depth and degree of heating of targeted chromophores (e.g., epidermal melanin and hemoglobin molecules in blood). PPTR can be used to estimate the epidermal heating and the temperature rise (DT epi ) at a given subtherapeutic radiant exposure H o . If ambient skin temperature is assumed to be 30 1C, H max is approximately equal to (40 1C) Â (H o /DT epi ), where 40 1C is the difference between the assumed T thresh for epidermal damage (70 1C) and ambient skin temperature (30 1C).
A key component of typical PPTR systems is a relatively bulky liquid nitrogen cooled detector or focal plane array. Katzir and colleagues (6)(7)(8) have investigated the use of IR fibers to deliver blackbody emission to a benchtop detector, but these fibers tend to be expensive. To provide the clinician with a user friendly PPTR system that can be used to estimate H max at selected skin sites, we designed a small, fiber-free, hand-held system. Inasmuch as currently available thermoelectrically cooled detectors are AC coupled, the goal of this study was to determine the feasibility and accuracy of an AC-coupled PPTR system for DT epi measurements. We characterized AC-coupled PPTR system performance, and then compared measurements of DT epi with quantitative skin melanin content measurements to assess system accuracy.

Materials and Methods
AC-coupled PPTR system Figure 1 shows a photograph of the small, fiberfree, hand-held PPTR system developed in this study. The key component of the system was a customized thermoelectrically cooled HgCdZnTe IR detector (Oriel, Stratford, CT, USA) operating in photovoltaic mode. Specifications of the detector include a 2 Â 2 mm 2 active area, detectivity D* of 1.2 Â 10 10 at 6 mm, and spectral sensitivity of 2-6 mm. A built-in preamplifier with a bandwidth of 0.01-140 kHz was used to amplify the detected signal. An integrated power supply/controller box regulated the detector temperature.
A holder was designed to position in a repeatable fashion the handpiece of a clinically used pulsed dye laser (PDL) (Candela Corp., Wayland, MA, USA). The holder was positioned to direct PDL light through on open area in a fixed positioning aperture. Time-resolved blackbody emission from a 2 Â 2 mm 2 area on a sample placed at the positioning aperture was collected with a biconvex CaF 2 IR lens (f 5 25.4 mm, F/# 5 1). To operate in a wavelength band over which the blackbody emission attenuation coefficient is uniform, the spectral sensitivity of measured IR radiation was restricted to 4.5-5 mm by placing an optical filter in front of the detector window (5).
As the detector preamplifier was designed for AC-coupled signal detection, a reference background temperature measurement was required for calibrated DT epi measurements. A shutter system was constructed to provide a variable reference background temperature. The shutter system consisted in part of a thin, thermoelectrically cooled copper plate. To simulate blackbody emission from human skin, one side of the plate was coated with a uniform layer of highthermal emissivity black paint. A custom-built temperature controller was used to maintain the shutter temperature at a user-specified level.

PPTR system calibration
Temperature calibration of the PPTR system was performed with a blackbody calibration source (Omega, Stamford, CT, USA). The blackbody source was placed at the positioning aperture of the system and the temperature was initially adjusted to the user-specified shutter temperature (i.e., 22 or 33 1C in this experiment) to obtain a temperature difference (DT) of zero between the shutter and the source. The DT range used in this study was 0-70 1C, in 2 1C increments.
At each blackbody temperature setting, the shutter was closed and opened and a raw ACcoupled IR signal (DS raw ) was acquired by a digital oscilloscope (Tektronix, Beaverton, OR, USA) at a sampling rate of 1 kHz. A representative example is shown in Fig. 2a, for DT of 35 1C. At to0, the shutter was closed. At t 5 0, the shutter was opened to allow measurement of the blackbody emission. DS raw was integrated over time to arrive at a signal DS similar to that shown in Fig. 2b. The resultant maximum value of DS (e.g., DS max 5 2.8 V in Fig. 2b) then represented the specified value of DT. This procedure was repeated for each DT value to generate a calibration curve.
The experimentally derived calibration curve was compared with a theoretical simulation using Planck's law (9): where W b (l, T) is the spectral emissive power, e b (l) is the emissivity of the object, K 1 5 3.743 Â 10 4 (W Á (mm) 4 /cm 2 ), and K 2 5 1.4387 Â 10 4 (mm Á K). For a Lambertian emitter, the spectral emission power is given by (10) where y is the viewing angle relative to the normal surface. In a simulation, blackbody temperatures of 295-365 and 306-376 K were used with shutter temperatures of 295 and 306 K, respectively. Other parameters included e b 5 1, spectral range of 4.5-5 mm, and y 5 261. The total emissive power in the selected spectral range was computed at each blackbody temperature. The total emissive power of the shutter was subtracted from each blackbody emissive power value to generate a theoretical DT calibration curve.
In vitro and in vivo experiments to evaluate PPTR system accuracy The relationship between PDL radiant exposure and DT was studied using an in vitro skin simulating gel model. An agar gel stained with Direct Red 81 (Sigma, St Louis, MO, USA) to absorb incident 585 nm PDL light was prepared. The gel was irradiated with progressively higher radiant exposures, from 8 to 20 J/cm 2 , in 1 J/cm 2 increments. The reference temperature (22 1C) of the gel before PDL irradiation was monitored with a patch-type Omega thermocouple during the experiment and DT calibration was performed with a shutter temperature of 22 1C.
As a preliminary test of PPTR system accuracy, an in vivo human skin temperature measurement was acquired from the dorsal side of a subject's hand. For reference temperature measurements, patch-type Omega thermocouples were placed on the dorsal side of the subject's hand (33 1C) and the shutter (22 1C) surface.
As a first demonstration of the feasibility of our PPTR system to estimate H max , we correlated measurements of DT epi with epidermal melanin content based on skin color measurements. Initial skin surface temperature (33 1C) was measured with a patch-type thermocouple and DT calibration was performed with a shutter temperature of 33 1C. Epidermal melanin content was indirectly evaluated with a CR-200 chromameter (Minolta, Osaka, Japan), in which b* values (i.e., of the L*a*b* color space) were utilized as an index (11). Experiments in our laboratory suggest that b* is a more accurate melanin index than L*, another popular metric for melanin content (unpublished data). Higher b* values indicate higher melanin content. Finally, 10 matched pairs of DT epi and b* were acquired from the forearms of five normal subjects. The relationship between DT epi and b* was investigated using regression analysis.
All in vivo human skin measurements were acquired under a protocol approved by the Institutional Review Board at University of California, Irvine.

Results and Discussion
Sensitivity of IR radiometry system Calculation of DS resulted in a higher sensitivity, as compared with DS raw in the measurement of DT (Fig. 3). The signal at t40 in Fig. 3a represents the DS raw (dotted lines) that resulted with a 1 1C temperature rise of blackbody radiation (23 1C) over the shutter temperature (22 1C). In Fig. 3a, it is difficult to identify the difference between blackbody and shutter temperatures. However with integration of DS raw (DS), the SNR was improved (Fig. 3b) and a 1 1C temperature difference can be discerned after integration of DS raw , with 0.2 1C error.
The signals at to0 in Figs. 2b and 3b are DS of the shutter. The peak-to-peak value of average DS of the shutter was AE 0.015 V across the blackbody temperatures for a DT calibration data set. By using the DT calibration curve, the corresponding DT was determined to be AE 0.5 1C, suggesting that the IR radiometric system can resolve a DT of AE 0.5 1C between the reference and sample temperatures.
PPTR system calibration Experimental and theoretical calibration curves were similar to one another. As an example, Fig.  4a shows a DT calibration curve as a function of DS max for a shutter temperature of 33 1C, to which a cubic regression resulted in the best fit (R 2 5 0.99) to the measurements. The experimental result in Fig. 4a was compared with theoretical simulation results of Planck's law ( Eqs.1 and 2). The experimental and theoretical results showed a strong linear correlation (R 2 5 0.99) as shown in Fig. 4b.
In vitro and in vivo experiments to evaluate PPTR system accuracy In theory, the laser pulse duration (1.5 ms in this study) is shorter than the epidermal thermal relaxation time ($ 20 ms) and thus we expected DT epi generated by laser exposure to be directly proportional to the incident radiant exposure (9). In agreement with this hypothesis, a high-positive correlation does exist between DT epi and the incident radiant exposure (Fig. 5).   Figure 6 shows a measurement of a DT value (10.6 1C) between the dorsal side of the subject's (33 1C) hand and the shutter (22 1C). Thus, the measured absolute skin temperature was 32.6 1C, with the following equation: where T sample and T shutter represent the temperatures of the subject's hand and shutter, respectively. There was a temperature discrepancy of 0.4 1C between the actual (33 1C) and measured temperatures (32.6 1C), demonstrating the accuracy of the AC-coupled PPTR system. Figure 7 shows the regression analysis between DT epi and b* in which DT epi linearly increases with melanin content (R 2 5 0.99). This result agrees with the theoretical modeling study by Gabay et al. (12), in which skin surface temperature rise linearly increased as a function of epidermal absorption coefficient (i.e., melanin content).
Use of an integrated AC-coupled PPTR signal (DS) resulted in higher SNR as compared with a raw AC-coupled PPTR signal (DS raw ) because of a reduction in the high-frequency noise (Fig. 2). To the best of our knowledge, this technique has not been applied to AC-coupled IR signals. Sade et al. (10) relied on DS raw alone in characterization of their AC-coupled PPTR system. Moreover, when using DS raw , the shutter speed must be constant during both the temperature calibration process and the sample temperature measurement because the time-resolved intensity of the AC-coupled PPTR signal depends on shutter speed. This dependency can be a limitation in PPTR measurements that involve variable laser pulse durations. Even if the DT calibration is performed with a constant shutter speed, the PPTR measurement of DT may have additional error if the laser pulse duration, which essentially serves as a shutter, is not identical to the shutter speed used for DT calibration. We addressed this issue by using an integrated AC-coupled signal (DS), which is not sensitive to shutter speed.
The developed fiber-free, hand-held ACcoupled PPTR system is compact and easy-to-use. By using a positioning aperture, the measurement distance is held constant. In addition, the PPTR system uses a thermoelectrically cooled IR detector that requires infrequent DTcalibration. We next plan to apply this system to in vivo measurement of epidermal damage to investigate further relationship between DT epi and H max .

Conclusions
The fiber-free, hand-held AC-coupled PPTR system is capable of accurate DT epi measurements. The system has a measured minimum resolvable DT of AE 0.5 1C. The experimental and theoretical calibration curves showed a strong linear correlation. The temperature measurement accuracy was demonstrated with in vivo skin measurements on the dorsal side of a subject's hand, with a temperature error of 0.4 1C. In a PPTR experiment with a gel phantom, DT showed a strong positive correlation with incident radiant exposure, in agreement with our prediction. Finally, measurements of in vivo DT epi as a function of melanin content showed the potential of the AC-coupled PPTR system for accurate DT epi measurements.