Journal of the Korean Physical Society, Vol. 51, No. 5, November 2007, pp. 1663 1667 Evaluation of a Chip LED Sensor Module at 770 nm for Fat Thickness Measurement of Optical Tissue Phantoms and Human Body Tissue Dong-Su Ho and Beop-Min Kim Department of Biomedical Engineering, Yonsei University, Wonju 220-842 In Duk Hwang and Kunsoo Shin Bio & Health Laboratory, Samsung Advanced Institute of Technology, Suwon 440-600 (Received 20 March 2007) We measured the fat thickness noninvasively by using a diffuse optical method for tissuesimulating phantoms and human tissues. A light source module composed of 770 nm low-power chip LEDs and a photo-detector were used in this study. The optical tissue phantoms were composed of a fat and a muscle layer made with gels with appropriate absorption/scattering coefficients. The fat thickness was varied from several to 30 mm. After a proper calibration procedure, we used this system to conduct human studies. Based on this preliminary study, noninvasive fat thickness measurements are possible with a simple curve-fitting procedure. PACS numbers: 42.62.Be Keywords: Fat thickness, Diffuse optics, LED I. INTRODUCTION II. EXPERIMENTS There are various techniques used to study the lean body mass and subcutaneous fat distribution [1]. Imaging methods such as magnetic resonance imaging (MRI), computer tomography (CT), dual-energy X-ray absorptionmetry (DEXA), and ultrasound imaging are precise and accurate techniques. However, these methods are applicable in very limited cases due to the high cost and the risk of radiological burden (CT, DEXA). A noninvasive optical approach, such as diffuse optical spectroscopy, has many advantages over these technologies because it costs less and poses no radiation toxicity [2 4]. More importantly, it can be constructed in a small size and incorporated with mobile devices. If a portable, miniaturized optical sensor module is to be developed for human fat thickness measurements, a hardware layout, including an appropriate source-detector-distance (SD) and LED power, has to be configured. For this purpose, we analyzed the diffusely reflected light intensity at different SD s and LED powers by using tissue-simulating phantoms that had optical properties similar to those of human tissues at a wavelength of 770 nm. E-mail: indhwang@samsung.com; Fax: +82-31-280-8277 The optical properties of the fat and the muscle of skin at 770 nm are shown in Table 1, which shows that the fat layer is highly scattering while the muscle is highly absorbing [2 4]. To accurately simulate the human skin, it is necessary to include the epidermal and the dermal layers, too. However, considering that those layers are very thin and they may contribute only to a linear attenuation of the diffusely reflected light for all SD s, we decided to construct a simple skin model with two layers (fat and muscle). The fat and the muscle phantoms were made separately from gelatin (Sigma Corp.) mixed with intralipid solution as a scatterer and neutral-red powder as an ab- -1663-1. Phantom Preparation Table 1. Optical properties of human fat and muscle at 770 nm [2,3,6]. Tissue Layer µ a (1/cm) µ s (1/cm) Fat Phantom (Intralipid 20 %) 0.064 10.50 Muscle Phantom 0.3 7
-1664- Journal of the Korean Physical Society, Vol. 51, No. 5, November 2007 Fig. 3. LED optical property acquisition module. 2. Experimental Setup Fig. 1. Photograph of fabricated fat (left) and muscle (right) phantoms. Fig. 2. Schematic of the two-layer tissue phantom experiment. sorber. The optical properties of the intralipid solution and the neutral-red powder were pre-determined by using an inverse adding-doubling (IAD) method from the transmittance and the reflectance measurements using an integrating sphere [5]. All measurements were made at 770 nm wavelength. Biological tissues are well known to be less opaque in the wavelength range, called the optical window of 600 1,100 nm. We picked the wavelength of 770 nm simply because less expensive and effective source module were available at that wavelength [7 14]. The final fat and muscle phantoms had optical properties similar to those of human tissues, as shown in Table 1. Once the phantoms had been made, we measured the optical properties of the phantoms separately to make sure that the phantoms had the same optical properties as designed. After each phantom had been made, a sliced fat layer with a specific thickness was placed over the absorber layer to obtain a two-layered skin model. The thickness of the fat layer were varied from 2 mm to 30 mm while the muscle layer was fixed at a thickness of 35 mm. The muscle thickness is not important because light cannot reach the bottom surface of the absorber phantom anyway. Figure 1 shows a photograph of the fabricated fat and muscle phantoms. The width and the length of the phantoms are 110 mm and 80 mm, respectively. Figure 2 shows the experimental setup for testing our sensor module with a two-layer phantom, which consists of a fat layer on top of a muscle layer. Two kinds of LEDs with a center wavelength of 770 nm were used in our experiment. One was a lamp LED (viewing angle 20 degrees), and the other was a miniaturized chip LED of the surface mount type (viewing angle 120 degrees). Figure 3 shows a picture of the chip LED module system. The frequency of the LED pulse operation was about 4.8 khz, and its duty was less than 10 %. In the case of the lamp LED, the source distances from the detector were 5, 10, 15, 20 and 25 mm, respectively. The applied current to each LED was 80 ma. In the case of the chip LED, the light sources were separated from the detector by 5 mm (S1), 10 mm (S2) and 20 mm (S3). As the SD increases, the detected intensity decreased exponentially. To compensate for this, multiple LEDs were used at longer distances from the detector. Figure 3 shows three LEDs were used at a distance of 10 mm and 5 LEDs were used at 20 mm. III. RESULTS 1. MonteCarlo Simulation In our Monte Carlo simulation, each layer of the model structure was assumed to be infinitely wide. The thickness of the fat layer was varied from 2 mm to 30.0 mm, in a step of 2.0 mm. The muscle layer was placed underneath the fat layer, and the muscle thickness was set large enough to guarantee that no light reached the bottom surface of the muscle layer. The refractive indices of the fat and the muscle layers were both set as 1.37, the diameter of the source beam was 2.5 mm, and the diffuse reflectance was measured up to a radial position 40 mm from the source. The number of photons was 105. Figure 4 show the simulation results on the detected diffuse reflectance as a function of the fat thickness at 770 nm. As expected, it shows an exponential decay of the diffuse reflectance as a function of the source-detector (SD) distance [6]. It also shows that the overall diffuse reflectance (integrated area under the curve) increases
Evaluation of a Chip LED Sensor Module at 770 nm for Fat Thickness Dong-Su Ho et al. -1665- Fig. 4. Simulation results of the reflected intensity for different fat thicknesses as a function of the source-detector (SD) distances. Fig. 6. Detected intensity for various source-detector distances (SDs) SDs as a function of the fat thickness by using a llamp LED. sults are presented below. 2. Phantom Experiment Using Lamp LED s Fig. 5. Simulation results of the detected intensity for various source-detector distances (SDs) as a function of the fat thickness for thicknesses from 0.2 cm to 3 cm. as the thickness of the fat increases because the fat has much a lower absorption coefficient than muscle does. As the fat layer thickness increases, more light has a chance to travel through the fat and to be diffuse reflected on the surface. Figure 5 shows the simulation result for the relationship between the fat thickness and the diffuse reflectance for various source-detector (SD) separations. The diffuse reflectance increases gradually as the fat thickness increases for all SD s. The signal is ultimately saturated because most photons pass through the fat layer as the fat layer becomes thicker. For larger SD s, the photons pass though the deeper region before detection, and the signal is saturated for larger fat thickness. We verified our observation via a phantom experiment, and the re- We first investigated the sensor module, which consists of one detector and four lamp LED s (770 nm, divergence angle 20 degrees, 3 mm diameter) placed at various SD s. Starting 10 mm from the detector, the LED s are placed 5 mm apart from each other. The maximum distance between the detector and the source is 25 mm. The same current and amplifier gain were applied to all the LED s. Because we collected all photons backscattered with the same gain, the magnitudes at SD s of 20 mm and 25 mm were much lower than those of SD s of 10 mm and 5 mm. In Figure 6, the detected intensity at different SD s as a function of the fat thickness, shows trends similar to those of the simulated results (Figure 5). From these results, it is obvious that the fat thickness can be well-estimated only up to a thickness of 10 mm or so when a single LED is placed at a SD of 10 mm. It is possible, but difficult, to measure a fat thickness thicker than 10 mm. Especially, if the SD becomes too large, the signal magnitude becomes too small to be measured. It is important to notice that the SD has to be larger so that a thicker fat layer can be studied. Therefore, a large SD is needed, and more optical power is needed for that large SD. 3. Phantom Experiment Using Chip LED s Based on what we observed in the lamp LED experiment, we built and tested another sensor module, which consist of one detector and a series of chip LED s (770 nm). The light sources (770 nm, chip LED s, divergence
-1666- Journal of the Korean Physical Society, Vol. 51, No. 5, November 2007 Fig. 7. Reflected intensity variation for the three different separations between the source and the detector as a function of the fat thickness. as the fat became thicker. For a SD of 10 mm, the maximum measurable thickness was approximately 20 mm. However, since the optical power at large SD (20 mm) was increased, a fat thickness up to 30 mm could still be estimated by using this sensor module. In a separate experiment, to investigate the influence of the LED power, we measured the reflected intensity by using three LEDs and five LEDs for the same SD of 20 mm and for the same amplifier gain. In this test, all hardware conditions, except for the numbers of chip LED s, were the same. The results are shown in Figure 8. We observed that even if the signals showed similar increasing patterns, higher power (more chips) could produce a clearer signal. From the 3rd-order polynomial curve-fitting results of Figure 8, the data can be shown to be nicely fitted to a simple polynomial. Thus, we conclude that this sensor module with a larger SD and more optical power can measure a large range of fat thicknesses up to 30 mm. IV. CONCLUSION Fig. 8. Reflected intensity and the fitting results for two different numbers of LEDs as a function of the fat thickness at a fixed SD of 20 mm. angle of 120 degrees) were separated from a single Si detector by different distances of S1 (5 mm), S2 (10 mm) and S3 (20 mm). Since the diffuse reflectance decreases exponentially with radial distance, we used multiple LED s at larger distances (3 chips for S2 and 5 chips for S3). The cycle of the LED pulse operation was approximately 4.8 khz, and its duty was less than 10 %. We adjusted the gain of the amplifier so that the detected intensity (V) at fat thicknesses up to 30 mm could be detected for a SD of 20 mm with a maximum intensity signal of 3 V. The amplifier gain for the SD s of 5 mm, 10 mm and 20 mm were 1, 1.5 and 5, respectively. All data were collected at a LED driving current of 60 ma. Figure 7 shows the reflected intensity as a function of the fat thickness at three different separations between the source and the detector. The result is very similar to those of the lamp LED experiment. For a SD of 5 mm, the detected intensity rapidly increased up to 10 mm of fat thickness. After a thickness of 10 mm, it saturated We built an optical phantom by using gelatin mixed with intralipid solution as a scatterer and neutral-red powder as an absorber. The fat and muscle phantoms were made using the data in other literature. A Monte-Carlo simulation was performed, and the results were verified using the phantom experiments. We showed that a simple sensor module composed of a single detector and multiple LED s could measure fat thicknesses up to 30 mm. In further studies, the same module will be used for fat thickness measurements in humans. ACKNOWLEDGMENTS This study was supported by a grant of the Korea Health 21 R & D Project, Ministry of Health & Welfare, Republic of Korea (02-PJ3-PG6-EV07-0002). This work was also supported by the Brain Research Center of the 21st Century Frontier Research Program funded by the Ministry of Science and Technology (M103KV010019-06K2201-01910). REFERENCES [1] S. J. Wallner, N. Luschnigg, W. J. Schnedl, T. Lahousen, K. Crailsheim, R. Möller, E. Tafeit and R. Horejsi, International J. Obesity 28, 1143 (2004). [2] Y. Yang, O. O. Soyemi, M. R. Landry and B. R. Soller, Opt. Express 13, 1570 (2005). [3] A. Kienle, L. Lilge, M. S. Patterson, B. C. Wilson, R. Hibst and R. Steiner, SPIE 2326, 212 (1995).
Evaluation of a Chip LED Sensor Module at 770 nm for Fat Thickness Dong-Su Ho et al. -1667- [4] W. Cui and L. E. Ostrander, IEEE Trans Biomedical Eng. 39, 194 (1992). [5] B.-M. Kim, J. Korean Phys. Soc. 44, 427 (2004). [6] T. J. Farrell, M. S. Patterson and B. C. Wilson, Med. Phys. 19, 879 (1992). [7] J. T. Seo, U. Hommerich, J. D. Mackenzie, C. R. Abernathy and J. M. Zavada, J. Korean Phys. Soc. 36, 311 (2000). [8] H. M. Park, I. O. Chae, M. R. Lee, E. C. Jung, Y. J. Rhee and J. M. Lee, J. Korean Phys. Soc. 39, 886 (2001). [9] J. P. Culver, V. Ntziachristos, M. J. Holboke and A. G. Yodh, Opt. Lett. 26, 10 (2001). [10] M. A. Franceschini and D. A. Boas, Neurolimage 21, 372 (2004). [11] D. A. Boas, A. M. Dale and M. A. Franceschini, Neuroimage 23, S275 (2004). [12] J. Kim, C. Choi, H.-S. Soh, D.-S. Ho and B.-M. Kim, J. Korean Phys. Soc. 47, 375 (2005). [13] O.-Y. Jeon, M.-J. Jin, H.-H. Lim, B.-J. Kim and M. S. Cha, J. Korean Phys. Soc. 49, 589 (2006). [14] S. M. Kim, K. Kyhm, K. S. Hong and H.-S. Yang, J. Korean Phys. Soc. 49, 688 (2006).