Open Access
Add to BibSonomy
Abstract
We report a handheld, smartphone-based spatial frequency domain imaging device. We first examined the linear dynamic range of the smartphone camera sensor. We then calculated optical properties for a series of liquid phantoms with varying concentrations of nigrosin ink and Intralipid, demonstrating separation of absorption and scattering. The device was then tested on a human wrist, where optical properties and hemoglobin-based chromophores were calculated. Finally, we performed an arterial occlusion on a human hand and captured hemodynamics using our device. We hope to lay the foundation for an accessible SFDI device with mass-market appeal designed for dermatological and cosmetic applications.
© 2022 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
Non-invasive imaging of skin has been an area of significant interest in dermatology and cosmetology. In particular, dermatoscopes have seen widespread adoption as a relatively low-cost and easy-to-use option by professionals and individuals [1,2]. In dermatology, dermatoscopes have been used for imaging melanocytic lesions as well as other skin abnormalities [2,3]. Even personal usage of dermatoscopes may ease economic and logistical barriers by allowing patients to capture and upload images of skin lesions for further review [4,5]. In cosmetology, facial imaging by dermatoscopy and photography have been used to grade skin texture, wrinkles, and acne and have been moving towards more personalized solutions [6,7]. Introduction of bespoke skincare routines and evaluation of dermocosmetic products have been a developing topic in the last few years [8–11].
However, current skin imaging techniques have several key limitations. Dermatoscopic imaging can provide high-resolution, color snapshots of skin, but are subjective in nature [12,13]. Multi-spectral techniques have been implemented in smartphone formfactors [14,15]. While these devices can estimate chromophore concentrations, tissue scattering quantification still remains a challenge. Other skin imaging technologies excel at structural imaging, such as optical coherence tomography and high-frequency ultrasound, but these methods do not examine tissue optical properties or metabolic parameters, metrics shown to be correlated with skin condition and wound healing [1,16,17]. In the area of skin imaging, there is a need for an inexpensive and easy-to-use, yet quantitative handheld imaging device.
Spatial frequency domain imaging (SFDI) has demonstrated potential as an emerging skin imaging technology, providing quantitative metrics such as tissue absorption (µa), reduced scattering (µs’), oxyhemoglobin (HbO2), and deoxyhemoglobin (HbR) [17,18]. SFDI has been used in a number of research topics such as monitoring chronic ulcers, surgical skin flap viability, and skin burns [19–21]. Yet, while SFDI can provide valuable compositional and functional insights about skin, typical SFDI device size, cost, and complexity has remained challenges for larger market appeal.
In this report, we present a handheld, low-cost, and accessible SFDI device in a smartphone-attached form factor. To evaluate the device, we first test the smartphone camera response to near-infrared (NIR) light and determine the linear dynamic range. Extraction of optical properties were then tested using liquid-based phantoms, as well as on a human wrist. Finally, we track hemodynamics of a human hand during arterial occlusion and release.
2. Material and methods
2.1 Device design
SFDI measurements were performed using a custom designed (SolidWorks, Dassault Systèmes) 3D-printed smartphone case fitting a Samsung Galaxy S8 (Samsung Electronics Co. Ltd.) with the infrared camera filter removed, as shown in Fig. 1. Light-emitting diodes (LED) in a multi-package case were utilized as light sources (Shenzhen Best LED Opto-electronic Co., Ltd.) with center wavelengths of 665 nm, 830 nm, and 930 nm. The case was connected to a shroud, which provided a fixed measurement distance and minimized ambient light. LEDs were driven and controlled using an Arduino Nano 33. Light from the LED projected through a polarizer (58-649, Edmund Optics Inc.) and then through a glass slide with a static sinusoidal pattern (SF-1.0-90-TM-G, Applied Image Inc.). The resulting sinusoidal projection had a spatial frequency (SFX) of 0.21 mm−1. Images were then collected by the smartphone camera through a crossed polarizer. The device size was approximately 9 × 15 × 7 cm and 260 grams in weight.
Fig. 1. Design details of the smartphone-based SFDI device. (a) The light-emitting diode and patterned slide projected the spatial pattern onto the sample, while the phone camera collected the resulting image. Crossed polarizers were placed in front of the light source and camera in order to minimize capturing specular reflections. The measurement area was approximately 4 × 4 cm. (b) The 3D printed attachment consisted of the phone case, the projection unit, which housed the light source with spatial pattern slide, and the shroud which provided a fixed measurement distance and blocked ambient light. (c) The handheld device was approximately 9 × 15 × 7 cm and weighed 260 grams.
For this prototype, the majority of material expense stemmed from the smartphone and sinusoidal-patterned glass slide, as shown in Table 1. The approximate total cost was 1250 in United States Dollars (USD).
Table 1. Approximate cost of components of our SFDI device.
2.2 Extraction of optical properties
SFDI is an imaging modality capable of non-invasively quantifying tissue optical properties and has been thoroughly discussed in literature. Briefly, extraction of optical properties using SFDI is based on measurements of the diffuse modulation transfer function [18,22]. Traditional SFDI measurements are performed by projecting 2-D sinusoidal patterns at three phases (i.e., each shifted by 120 degrees) onto a turbid medium and then imaged using a camera. A three-phase SFDI measurement captures a separate frame for each phase at each wavelength. This necessarily increases the acquisition time as well as instrument complexity, with many groups utilizing a digital micromirror device (DMD) or a mechanical translator to perform the phase shifts [18,23,24].
In this work, we follow an alternative method to capture SFDI data involving only a single phase measurement, sometimes referred to as “single-snapshot optical properties” (SSOP) [25,26]. After imaging the sinusoidal projection, by Fourier transformation, one can separate the image into its modulated (AC) and unmodulated (DC) constituents. This has greatly increased the potential acquisition speed of SFDI, but at the cost of spatial resolution and artifacts [25]. Recently, a technique was presented to improve the image quality of SSOP using 2D anisotropic filters during the demodulation step [26]. Diffuse reflectance images can then be obtained by calibration to a well-characterized phantom. Optical properties are calculated at each pixel by fitting with the diffusion approximation to the radiative transport equation or by utilizing data generated through Monte Carlo methods [18,27]. Tissue chromophore concentrations are finally estimated from absorption values using least-squares fitting with known extinction coefficients [28].
2.3 Smartphone camera evaluation
Smartphone camera capturing software may utilize post processing filters to enhance the final image tuned for human perception. However, these types of filters tend to change the output pixel intensities, thus may not be appropriate for quantitative imaging where unmodified frames are desirable. While on some smartphones, post-processing can be disabled, these cameras also typically employ compression methods such as the Joint Photographic Experts Group format for still images and Moving Picture Expert Group-4 (MP4) format for videos in order to reduce the file size of digital media [29]. Image compression affects the color accuracy as well as the dynamic range and dynamic range linearity.
To demonstrate the impact of digital media compression and assess the dynamic range of our smartphone, we utilized the apparatus shown in Fig. 2. Approximately half of the light from the 665 nm LED was directed to the camera sensor using a 50% beam splitter (CCM1-BS014, Thorlabs Inc.) while the transmitting light was focused on to an optical power meter (Starbright, Ophir Optronics Solutions Ltd.). Optical power was regulated using a variable neutral density filter (NDC-50C-4-B, Thorlabs Inc.). Videos were recorded as digital negatives (DNG), a raw format without postprocess filters [30]. A compressed format (MP4) was also captured for comparison. The camera ISO was set to the lowest available (ISO 100) to minimize sensor noise while the exposure time was maintained at 0.25 seconds.
Fig. 2. Optical apparatus used to image light from the LED using the smartphone camera as well as measure the relative optical power. In this setup, the light emitting diode (LED) output was collimated using a collimator lens (CL) and attenuated by the variable neutral density filter (VND). The light was then split using a beam splitter (BS) with approximately half of the output directed to the smartphone (SP) and the other half to the power meter (PM).
2.4 Imaging samples
To test the ability of our device to separate absorption and scattering, we modified the optical properties of two Intralipid-based (Fresenius Kabi) liquid phantoms: 1) increased absorption by addition of a nigrosin solution (Sigma-Aldrich) while maintaining scattering 2) decreased scattering of the Intralipid solution through a series of dilutions while maintaining absorption. The phantom volume was kept at approximately 500 mL throughout the experiment. To dilute the phantom, 10% of the total phantom volume was removed and subsequently replaced with water. While increasing absorption, the total volume of nigrosin introduced was <1% of the total phantom volume and was assumed negligible. After each step, as summarized in Table 2, the phantom was gently mixed using a glass stir bar.
Table 2. Two Intralipid-based liquid phantoms were constructed. In one phantom, the absorption was increased by the addition of nigrosin ink. For the other phantom, scattering was reduced by diluting the phantom volume with water at each step.
To perform measurements, the SFDI device was fixed in place by optical posts with the shroud in contact with the surface of the liquid phantom. Measurements were automated by having the device continuously capture while flashing the 830 nm LED. To avoid recording partial frames (i.e., if the LED turned on or off in the middle of a frame), 10 frames were recorded. Partial frames were detected and discarded. The rest of the frames were averaged to mitigate the effects of random pixel noise. A 10 by 10 pixel median filter was also applied. The smartphone had a native resolution of 3024 × 4032 pixels. Images were cropped to exclude direct specular spotlight from the LED as well as parts of the shroud that were visible in the camera field of view.
2.5 In-vivo measurements
The human wrist of a healthy male subject was measured using our device. Procedures similar to those for the phantom measurements were followed. The device was held on the subject’s wrist while capturing data for the 665, 830, and 930 nm wavelengths. The overall measurement speed including instrumental delays was approximately 10 seconds per measurement for our 3-wavelength system.
The hand of the same subject was also occluded and measured using all three wavelengths. An occlusion cuff was placed around the bicep of a healthy male subject. After a 30 second baseline, the occlusion cuff was inflated using a programmable pump (moorVMS-PRES, Moor Instruments Inc.) until 210 mmHg was reached. This pressure was maintained for 2 minutes before rapidly releasing. We then observed for a 1-minute recovery period.
2.6 Data processing
All phantom and tissue data was handled by exporting DNG files from the internal phone storage to a micro secure digital (micro-SD) card. Data was then copied from the micro-SD to the solid-state drive of a laptop computer. This method of data transfer proved much faster than copying files directly from the phone to the computer using a USB connection. MATLAB 2020b (MathWorks, Inc.) was used for data processing and visualization.
3. Results
3.1 Smartphone dynamic range
The dynamic range of the smartphone camera was tested using two different file saving formats, as shown in Fig. 3. A collimated spot from the LED was imaged using the phone camera and monitored using an optical power meter. Saved file formats were either DNG or MP4. Data from both file formats were normalized to their respective maximum pixel intensity. Optical power was increased until approximately 147 µW. Camera sensor saturation was recognized as a plateau region, wherein an increase in optical power did not result in any increase in pixel intensity.
Fig. 3. Dynamic range measurements using uncompressed and compressed file formats. The uncompressed method provided the greatest dynamic range as well as the most linear dynamic range. The compressed method had significantly reduced dynamic range as well as nonlinearity. Dashed lines indicate linear fits.
The DNG format had the greatest dynamic range, saturating at approximately 127 µW. In contrast, the compressed format saturated at approximately 55 µW. Linear lines were fit to the DNG data excluding points in the saturation region resulting in a coefficient of determination of R2 = 0.996. Another linear line was fit to the MP4 data in the same manner yielding a coefficient of determination of R2 = 0.858.
3.2 Liquid phantoms
Optical properties maps at 830 nm imaged using our smartphone-based SFDI for the liquid phantoms can be seen in Fig. 4. Absorption can be seen steadily increasing with the amount of nigrosin introduced. Inversely, reduced scattering maps can be seen reducing in intensity as the Intralipid solution was diluted. These characteristics were also observed for the other wavelengths in the system, shown in supplementary Fig. S1. While the camera was operated using the lowest available ISO, grain noise can still be seen in the images due to the relatively low sensitivity of the sensor to NIR light sources.
Fig. 4. Optical property maps of liquid phantoms measured by the smartphone-based SFDI device. Images on each row were displayed using the common scale shown to the right. (Top) Absorption maps from the liquid phantom experiment in which nigrosin was added to the solution at each step (Bottom) Reduced scattering maps as the Intralipid solution was diluted with water at each step.
Average optical properties were then calculated for each step. Absorption was seen increasing linearly (R2 = 0.997) with the amount of nigrosin added, as shown in Fig. 5. A small decrease in reduced scattering was observed from step 1 to step 10 as the phantom volume was slightly increased by the addition of nigrosin. Reduced scattering can be seen as linearly (R2 = 0.989) correlated to the amount of Intralipid in the solution. Absorption was observed to be slightly higher at low Intralipid percentages due to an increased water fraction. In either scenario, the device was capable of separating absorption and scattering effects.
Fig. 5. Mean and standard deviation values shown for the absorption and reduced scattering dynamics at 830 nm during the liquid phantom experiment. Dash lines indicate linear fits. (a) Variable absorption and nearly constant scattering were measured as a result of increasing nigrosin concentration. (b) Nearly constant absorption and variable scattering were measured as a result of diluting the Intralipid solution.
Fig. 6. In vivo measurement of a human wrist. Horizontal black bars represent a 1 cm scale. (a) Approximate measurement area on the wrist (b) Absorption maps for three wavelengths (c) Reduced scattering maps for the three wavelengths (d) Calculated HbO2 and HbR chromophore maps.
Fig. 7. Recovered hemodynamics during arterial occlusion and release of a subject’s hand. Measurements were performed using wavelengths 665, 830, and 930 nm. Camera grain noise persistent throughout the data contributed greatly to the standard deviation.
3.3 In vivo wrist measurement
Optical properties at 665, 830, and 930 nm LED wavelengths were measured on a human wrist using our device (Fig. 6). Based on these three wavelengths, tissue chromophores HbO2 and HbR were also calculated. Subsurface blood vessels were visible in the absorption maps, as well as the chromophore maps. Wrist tissue optical properties measured by our SFDI device were similar to those previously reported in literature [31].
3.4 Arterial occlusion measurement
Tissue hemodynamics of a human hand were monitored during an arterial occlusion challenge. Shown in Fig. 7, after inflation of the blood pressure cuff, HbO2 can be seen decreasing, while HbR increased. Immediately after arterial release, a rapid increase in HbO2 and decrease of HbR was observed before finally normalizing. This is a typical post-occlusive hyperemic response as seen by other groups utilizing SFDI [32,33].
4. Discussion
Our goal for this work was to integrate SFDI with several advantages offered by smartphones, which have become near-ubiquitous in recent times. Combining the imaging capabilities of SFDI with the intuitiveness of a smartphone platform could enable greater adoption of the technique. In addition, the computing power of smartphones have significantly evolved over the past two decades, making SFDI data acquisition and data processing on a single handheld device an increasingly feasible endeavor. We envision processed images could then be interpreted by trained professionals or machine learning algorithms, similar to screening methods previously reported with digital dermatoscopy [34].
Our smartphone camera was first evaluated by dynamic range testing. Postprocessing filters such as smoothing, white balancing, and color saturation as well as image compression could adversely affect the accuracy of captured images. Raw formats save unedited intensity information for each pixel on the camera sensor. For DNG images, the camera sensor had a linear response to the input optical power until near the saturation point. For all phantoms and tissue measurements, the exposure time was maintained such that the sensor operated exclusively in the linear region below saturation and above the noise floor. We found that the most significant factor of the linear dynamic range test was actually the output file format.
We observed two main differences when comparing MP4 to DNG results. The MP4 data had a significantly truncated dynamic range, due to bit depth limitations of the file format, as well as a non-linear response, likely a result of tone-mapping [35]. However, one drawback of utilizing a raw format was the file size. For example, during a typical measurement, 115 megabytes of data was generated for each second of recording. Expanding to a larger population could place significant burden on network bandwidth and storage limitations. Towards future implementations of this device, maximizing raw data handling on the smartphone itself to minimize large data transfers to external processing computers should be a high priority.
Our device successfully demonstrated separation of absorption and scattering on a series of liquid phantoms. This allows SFDI to quantify tissue chromophores without scattering assumptions. In addition, scattering also has particular importance for monitoring biological tissues. While hemoglobin concentrations are closely related to the functional status of tissue, scattering can be a sensitive parameter for compositional changes as seen on burn wounds [36,37]. During the experiment, the device was lowered to touch the surface of the liquid phantom and then fixed by optical posts. However, a minor tilt of the imaging plane (with respect to the surface of the liquid) during placement caused an overestimation in absorption due to a change in SFX on one side of the measurement area. Due to a divergent projector, the device is particularly sensitive to height and tilt changes. This can result in inaccurate optical properties and residual sine projection artifacts if imaging conditions are not closely matched with the calibration phantom [38]. On the liquid phantom series with varied scattering, the device was placed without the tilt, resulting in more uniform optical properties throughout the experiment. Height and tilt variations can be corrected by profilometry methods [38,39], which will be applied to future renditions of the device.
Our device exclusively utilized SSOP. Advantages of utilizing SSOP were twofold: 1) The first was due to the relatively close measurement distance. For typical skin imaging, such as usage of a dermatoscope, the device is placed in close proximity to the tissue in order to provide a magnified view. The smartphone-based SFDI was designed to be operated in a similar manner. While SSOP can cause some resolution degradation in comparison to three-phase SFDI resulting in fine-detail loss at a distance [25], this can essentially be mitigated by bringing the device closer to the measurement area. For our in vivo wrist measurement, blood vessels 1.5 mm in diameter were clearly visible for absorption and chromophore maps on the human wrist. 2) Phase shifting a 2D sinusoidal projection requires additional instrumentation complexities such as a DMD or motor as other groups have utilized [23,40]. While effective, these phase modulators rely on additional power and control circuits, which ultimately increases the overall size of the device. In contrast, SSOP can be performed using a single static, transparent or reflective patterned surface.
SSOP also does not require the use of moving parts. Theoretically, this would result in a stable instrument state over time. By utilizing a shroud to control the distance between the device and tissue, the projection and imaging foci also remain consistent. This has allowed us to measure optical properties of a phantom while using a calibration profile from several days prior (Fig. S2). Although a longer-term study is needed, we envision that a sufficiently stable device could reduce or eliminate the need for users to perform calibration.
Our device was applied during an arterial occlusion of the human hand. The classic response of tissue deoxygenation during occlusion followed by reactive hyperemia was successfully observed. However, camera grain noise remains a challenge for our smartphone-based SFDI. This noise can result in an undesirable variance even on homogeneous media. Phone camera sensors typically have low responsivity in the far NIR. Currently, additional frames are captured to average out the noise, but at the cost of device acquisition speed. Phone models with cameras more sensitive in the NIR may be utilized in the future.
For future applications, we are also exploring spatio-temporal noise reduction techniques [41]. For cost-saving measures, the sinusoidal-patterned glass slide was the most expensive component in our system. Other groups have shown that printed patterns on low-cost transparent paper could be utilized for SFDI, and may be implemented in future iterations of our device [23,38]. Lastly, newer smartphones models commonly include additional cameras, potentially allowing for multiple modality imaging. For example, one camera could be reserved for SFDI, while another could be utilized for traditional dermatoscopy.
5. Conclusions
A smartphone-based SFDI device for portable, accessible imaging of skin was presented. We validated the device by first testing the phone camera linear dynamic range. Our device was then applied to a series of liquid phantoms with varying optical properties. The device demonstrated capability to separate absorption and scattering. A human wrist measurement was presented, where tissue optical properties and hemoglobin parameters were calculated. We also show hemodynamics of a human hand during occlusion, where the device was sensitive to tissue oxygenation states and post-occlusive reactive hyperemia. We hope our work will facilitate further advancements of handheld SFDI devices, enabling wider adoption of the technology for dermatological and cosmetic applications.
Funding
National Research Foundation of Korea (NRF-2018K1A4A3A02060572, NRF-2019R1I1A3A01062141, NRF-2020H1D3A1A04080958, NRF-2022R1I1A3073688).
Acknowledgements
We acknowledge the OpenSFDI project as a generally helpful resource during the early stages of this research. This work was supported by the National Research Foundation of Korea (NRF) funded by MSIT and MOE (NRF-2018K1A4A3A02060572, NRF-2019R1I1A3A01062141, NRF-2020H1D3A1A04080958, and NRF-2022R1I1A3073688).
Disclosures
J.K. is an employee of MEDiThings Co. Ltd., which is developing a commercial NIRS device.
Data availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
Supplemental document
See Supplement 1 for supporting content.
References
1. B. P. Hibler, Q. Qi, and A. M. Rossi, “Current state of imaging in dermatology,” Semin. Cutaneous Med. Surg. 35(1), 2–8 (2016). [CrossRef]
2. R. Giuffrida, C. Conforti, N. Di Meo, T. Deinlein, S. Guida, and I. Zalaudek, “Use of noninvasive imaging in the management of skin cancer,” Curr. Opin. Oncol. 32(2), 98–105 (2020). [CrossRef]
3. G. Micali, F. Lacarrubba, D. Massimino, and R. A. Schwartz, “Dermatoscopy: Alternative uses in daily clinical practice,” J. Am. Acad. Dermatol. 64(6), 1135–1146 (2011). [CrossRef]
4. J. M. Goulart, J. Malvehy, S. Puig, G. Martin, and A. A. Marghoob, “Dermoscopy in skin self-examination: a useful tool for select patients,” Arch. Dermatol. 147(1), 53–58 (2011). [CrossRef]
5. L. C. Deda, R. H. Goldberg, T. A. Jamerson, I. Lee, and T. Tejasvi, “Dermoscopy practice guidelines for use in telemedicine,” npj Digit. Med. 5(1), 55 (2022). [CrossRef]
6. R. Abdel Hay, R. Hegazy, M. Abdel Hady, and N. Saleh, “Clinical and dermoscopic evaluation of combined (salicylic acid 20% and azelaic acid 20%) versus trichloroacetic acid 25% chemical peel in acne: an RCT,” J. Dermatolog. Treat. 30(6), 572–577 (2019). [CrossRef]
7. G. O. Cula, P. R. Bargo, A. Nkengne, and N. Kollias, “Assessing facial wrinkles: Automatic detection and quantification,” Ski. Res. Technol. 19(1), e243–e251 (2013). [CrossRef]
8. Y. Lee, J. Choi, and S. Shin, “A Study on the direction of evaluation indicators for personalized beauty self-care,” J. Fash. Bus. 24(6), 120–134 (2020). [CrossRef]
9. Y. J. Lee, J. W. Choi, H. W. Nam, and S. Y. Shin, A Study on the Direction of Beauty Tech Reflecting the Skin Characteristics of Koreans: Focused on Case Studies (Springer International Publishing, 2022), 1003 SCI.
10. K. Oosthuizen, E. Botha, J. Robertson, and M. Montecchi, “Artificial intelligence in retail: The AI-enabled value chain,” Australas. Mark. J. 29(3), 264–273 (2021). [CrossRef]
11. M. M. Vergilio, L. I. Vasques, and G. R. Leonardi, “Characterization of skin aging through high-frequency ultrasound imaging as a technique for evaluating the effectiveness of anti-aging products and procedures: A review,” Ski. Res. Technol. 27(5), 966–973 (2021). [CrossRef]
12. H. Kittler, H. Pehamberger, K. Wolff, and M. Binder, “Diagnostic accuracy of dermoscopy,” Lancet Oncol. 3(3), 159–165 (2002). [CrossRef]
13. O. Reiter, I. Mimouni, M. Gdalevich, A. A. Marghoob, A. Levi, E. Hodak, and Y. A. Leshem, “The diagnostic accuracy of dermoscopy for basal cell carcinoma: A systematic review and meta-analysis,” J. Am. Acad. Dermatol. 80(5), 1380–1388 (2019). [CrossRef]
14. J. Spigulis, I. Oshina, A. Berzina, and A. Bykov, “Smartphone snapshot mapping of skin chromophores under triple-wavelength laser illumination,” J. Biomed. Opt. 22(9), 091508 (2017). [CrossRef]
15. S. Kim, D. Cho, J. Kim, M. Kim, S. Youn, J. E. Jang, M. Je, D. H. Lee, B. Lee, D. L. Farkas, and J. Y. Hwang, “Smartphone-based multispectral imaging: system development and potential for mobile skin diagnosis,” Biomed. Opt. Express 7(12), 5294 (2016). [CrossRef]
16. M. M. Vergilio, S. A. M. E. Silva, R. M. Jales, and G. R. Leonardi, “High-frequency ultrasound as a scientific tool for skin imaging analysis,” Exp. Dermatol. 30(7), 897–910 (2021). [CrossRef]
17. S. Gioux, A. Mazhar, and D. J. Cuccia, “Spatial frequency domain imaging in 2019: principles, applications, and perspectives,” J. Biomed. Opt. 24(7), 071613 (2019). [CrossRef]
18. D. J. Cuccia, F. Bevilacqua, A. J. Durkin, F. R. Ayers, and B. J. Tromberg, “Quantitation and mapping of tissue optical properties using modulated imaging,” J. Biomed. Opt. 14(2), 024012 (2009). [CrossRef]
19. A. Yafi, F. K. Muakkassa, T. Pasupneti, J. Fulton, D. J. Cuccia, A. Mazhar, K. N. Blasiole, and E. N. Mostow, “Quantitative skin assessment using spatial frequency domain imaging (SFDI) in patients with or at high risk for pressure ulcers,” Lasers Surg. Med. 49(9), 827–834 (2017). [CrossRef]
20. A. Yafi, T. S. Vetter, T. Scholz, S. Patel, R. B. Saager, D. J. Cuccia, G. R. Evans, and A. J. Durkin, “Postoperative Quantitative Assessment of Reconstructive Tissue Status in a Cutaneous Flap Model Using Spatial Frequency Domain Imaging,” Plast. Reconstr. Surg. 127(1), 117–130 (2011). [CrossRef]
21. J. Q. Nguyen, C. Crouzet, T. Mai, K. Riola, D. Uchitel, L.-H. Liaw, N. Bernal, A. Ponticorvo, B. Choi, and A. J. Durkin, “Spatial frequency domain imaging of burn wounds in a preclinical model of graded burn severity,” J. Biomed. Opt. 18(6), 066010 (2013). [CrossRef]
22. N. Dögnitz and G. Wagnières, “Determination of tissue optical properties by steady-state spatial frequency-domain reflectometry,” Lasers Med. Sci. 13(1), 55–65 (1998). [CrossRef]
23. M. Erfanzadeh, S. Nandy, P. D. Kumavor, and Q. Zhu, “Low-cost compact multispectral spatial frequency domain imaging prototype for tissue characterization,” Biomed. Opt. Express 9(11), 5503 (2018). [CrossRef]
24. M. B. Applegate, K. Karrobi, J. P. Angelo, W. M. Austin, S. M. Tabassum, E. Aguénounon, K. Tilbury, R. B. Saager, S. Gioux, and D. M. Roblyer, “OpenSFDI: an open-source guide for constructing a spatial frequency domain imaging system,” J. Biomed. Opt. 25(01), 1 (2020). [CrossRef]
25. J. Vervandier and S. Gioux, “Single snapshot imaging of optical properties,” Biomed. Opt. Express 4(12), 2938 (2013). [CrossRef]
26. E. Aguénounon, F. Dadouche, W. Uhring, and S. Gioux, “Single snapshot of optical properties image quality improvement using anisotropic two-dimensional windows filtering,” J. Biomed. Opt. 24(7), 1–21 (2019). [CrossRef]
27. J. Angelo, C. R. Vargas, B. T. Lee, I. J. Bigio, and S. Gioux, “Ultrafast optical property map generation using lookup tables,” J. Biomed. Opt. 21(11), 110501 (2016). [CrossRef]
28. T. H. Pham, O. Coquoz, J. B. Fishkin, E. Anderson, and B. J. Tromberg, “Broad bandwidth frequency domain instrument for quantitative tissue optical spectroscopy,” Rev. Sci. Instrum. 71(6), 2500–2513 (2000). [CrossRef]
29. M. Rabbani and R. Joshi, An Overview of the JPEG 2000 Still Image Compression Standard (2002), 17(1).
30. “https://helpx.adobe.com/camera-raw/digital-negative.html,”.
31. K. P. Nadeau, A. J. Durkin, and B. J. Tromberg, “Advanced demodulation technique for the extraction of tissue optical properties and structural orientation contrast in the spatial frequency domain,” J. Biomed. Opt. 19(5), 056013 (2014). [CrossRef]
32. M. Ghijsen, “Quantitative real-time optical imaging of the tissue metabolic rate of oxygen consumption,” J. Biomed. Opt. 23(03), 1 (2018). [CrossRef]
33. S. Gioux, A. Mazhar, B. T. Lee, S. J. Lin, A. M. Tobias, D. J. Cuccia, A. Stockdale, R. Oketokoun, Y. Ashitate, E. Kelly, M. Weinmann, N. J. Durr, L. A. Moffitt, A. J. Durkin, B. J. Tromberg, and J. V. Frangioni, “First-in-human pilot study of a spatial frequency domain oxygenation imaging system,” J. Biomed. Opt. 16(08), 1 (2011). [CrossRef]
34. S. M. Rajpara, A. P. Botello, J. Townend, and A. D. Ormerod, “Systematic review of dermoscopy and digital dermoscopy/ artificial intelligence for the diagnosis of melanoma,” Br. J. Dermatol. 161(3), 591–604 (2009). [CrossRef]
35. Y. Salih, W. Bt. Md-Esa, A. S. Malik, and N. Saad, “Tone mapping of HDR images: A review,” ICIAS 2012 - 2012 4th Int. Conf. Intell. Adv. Syst. A Conf. World Eng. Sci. Technol. Congr. - Conf. Proc.1, 368–373 (2012).
36. G. Ganesan, R. V. Warren, A. Leproux, M. Compton, K. Cutler, S. Wittkopp, G. Tran, T. O’Sullivan, S. Malik, P. R. Galassetti, and B. J. Tromberg, “Diffuse optical spectroscopic imaging of subcutaneous adipose tissue metabolic changes during weight loss,” Int. J. Obes. 40(8), 1292–1300 (2016). [CrossRef]
37. A. Mazhar, S. Saggese, A. C. Pollins, N. L. Cardwell, L. Nanney, and D. J. Cuccia, “Noncontact imaging of burn depth and extent in a porcine model using spatial frequency domain imaging,” J. Biomed. Opt. 19(8), 086019 (2014). [CrossRef]
38. J. P. Angelo, M. van de Giessen, and S. Gioux, “Real-time endoscopic optical properties imaging,” Biomed. Opt. Express 8(11), 5113 (2017). [CrossRef]
39. S. Gioux, A. Mazhar, D. J. Cuccia, A. J. Durkin, B. J. Tromberg, and J. V. Frangioni, “Three-dimensional surface profile intensity correction for spatially modulated imaging,” J. Biomed. Opt. 14(3), 034045 (2009). [CrossRef]
40. R. B. Saager, A. N. Dang, S. S. Huang, K. M. Kelly, and A. J. Durkin, “Portable (handheld) clinical device for quantitative spectroscopy of skin, utilizing spatial frequency domain reflectance techniques,” Rev. Sci. Instrum. 88(9), 094302 (2017). [CrossRef]
41. S. W. Lee, V. Maik, J. H. Jang, J. Shin, and J. Paik, “Noise-adaptive spatio-temporal filter for real-time noise removal in low light level images,” IEEE Trans. Consum. Electron. 51(2), 648–653 (2005). [CrossRef]
