Content uploaded by Alexey Lihachev
Author content
All content in this area was uploaded by Alexey Lihachev on Jul 11, 2016
Content may be subject to copyright.
Autofluorescence imaging of basal cell
carcinoma by smartphone RGB
camera
Alexey Lihachev
Alexander Derjabo
Inesa Ferulova
Marta Lange
Ilze Lihacova
Janis Spigulis
Autofluorescence
imaging of basal
cell carcinoma by
smartphone RGB camera
Alexey Lihachev,a,*Alexander Derjabo,b
Inesa Ferulova,aMarta Lange,aIlze Lihacova,aand
Janis Spigulisa
aUniversity of Latvia, Institute of Atomic Physics and Spectroscopy,
Biophotonics Laboratory, Raina Boulevard 19, Riga LV-1586, Latvia
bRiga East University Hospital, Oncology Centre of Latvia, Hipokrata
Street 4, Riga LV-1079, Latvia
Abstract. The feasibility of smartphones for in vivo skin
autofluorescence imaging has been investigated. Filtered
autofluorescence images from the same tissue area were
periodically captured by a smartphone RGB camera with
subsequent detection of fluorescence intensity decreasing
at each image pixel for further imaging the planar distri-
bution of those values. The proposed methodology was
tested clinically with 13 basal cell carcinoma and 1 atypical
nevus. Several clinical cases and potential future applica-
tions of the smartphone-based technique are discussed.
©2015 Society of Photo-Optical Instrumentation Engineers (SPIE) [DOI: 10
.1117/1.JBO.20.12.120502]
Keywords: autofluorescence; photobleaching; RGB imaging;
smartphone.
Paper 150558LRR received Aug. 20, 2015; accepted for publication
Nov. 12, 2015; published online Dec. 11, 2015.
1 Introduction
Incidences and mortality from skin cancer are still increasing.1
Depending on the melanin concentration, skin tumors are
broadly classified into two types—malignant melanomas (MM)
and nonmelanoma skin cancers (NMSC). MM is the most
aggressive skin cancer modality with death rate ∼80% of all
fatal skin cancer cases.1The most common NMSC are basal cell
carcinoma (BCC, about 80% of new cases) and squamous cell
carcinoma (SCC, about 20% of new cases) derived from the
basal and squamous cells of the epidermis, respectively.2BCC
is characterized by very slow growth tendency, low mortality
rate, and high risk of recurrence, while SCC is more aggressive
and associated with the risk of metastasis.3,4
Early detection and removal of skin cancers can significantly
increase the survival time. Noninvasive methods in primary
oncological diagnostics of skin tumors are still topical for
dermatologists and oncologists worldwide. One of those is
skin autofluorescence (AF) imaging and spectroscopy, based
on differences of AF specific information (intensity, spectral
shape, and lifetime) in the tumor and surrounding normal
skin.5–8
The feasibility of AF spectroscopy for BCC diagnostics and
differentiation has been studied extensively over this most recent
decade. AF spectra from BCC lesions excited in UV/blue region
(337 to 450 nm) were broadly characterized by decreased
fluorescence intensity in comparison with surrounding healthy
skin,8most often attributed to the shift in the levels of NADH/
NAD+ (reduced form and oxidized form of nicotinamide
adenine dinucleotide) and reduced elastin and collagen, affected
by malignant process. In some cases, especially in late tumor
growing stages, a weak red fluorescence peak of the endogenous
porphyrins has been observed.8
Tissue AF usually shows the photobleaching effect,9which
may be helpful in biomedical applications.10–14 Under continu-
ous wave (cw) excitation, skin AF intensity mainly drops during
the first 15 to 20 s, followed by a slow decrease. Photobleaching
kinetics can be well described by empirical double-exponential
function with subsequent extraction of time constants τ1and τ2
that characterize the rate of fast and slow phases of the AF
decrease.9Our previous research has demonstrated that each
skin pathology as well as healthy skin has its own specific
AF intensity decrease kinetics depending on excitation, locali-
zation, melanin content, and blood perfusion. Furthermore, the
analysis of AF decrease kinetics during the first 15 to 20 s of cw
laser excitation seems to be most suitable for clinical
implementation.15,16
Currently available smartphones equipped with high-resolu-
tion RGB cameras in combination with good light sensitivity
and color representation mainly satisfy the required technical
properties for adequate image acquisition.17 This technology
may become a useful diagnostic tool for dermatologists and
oncologists thanks to wide accessibility, convenient use, and
low cost.18,19 However, the ability to switch off the “embedded”
automatized settings such as exposure time, white balance, and
ISO is crucial for the skin parametric imaging. Our latest studies
have shown that smartphones such as Galaxy and Nexus are
suitable for mapping of skin chromophores.17
So far use of smartphones in skin pathology diagnostics has
been mainly related to dermatoscopy—specifically magnified
image acquisition under white or color illumination with sub-
sequent analysis based on ABCD rules, fractal image analysis
or other algorithms established in dermatoscopy.20–24 In this
paper, we present a smartphone-compatible technique for
acquisition and analysis of 405 nm light-emitting diode (LED)
excited skin autofluorescence images.
2 Materials and Methods
2.1 Experimental Setup
For parametric mapping of skin AF intensity decrease rates, a
sequence of AF images under continuous 405 nm LED (model
LED Engin LZ1-00UA00-U8, spectral band half-width 30 nm)
excitation for 20 s at a power density of ∼20 mW∕cm2with a
frame rate 0.5 fr∕swas recorded and analyzed. Four battery-
powered violet LEDs were placed within a cylindrical light-
shielding wall that also ensured fixed distance (60 mm) between
the smartphone camera and evenly irradiated a spot (diameter
40 mm) of the examined tissue. A long pass filter (>515 nm)
was placed in front of smartphone camera to prevent detection of
the LED emission. The recorded RGB images were further sep-
arated to exploit R- and G-images for imaging of skin tissue AF
*Address all correspondence to: Alexey Lihachev, E-mail: aleksejs.lihacovs@
gmail.com 1083-3668/2015/$25.00 © 2015 SPIE
Journal of Biomedical Optics 120502-1 December 2015 •Vol. 20(12)
JBO Letters
in the red and green spectral bands, respectively. Due spectral
cutoff by 515-nm long pass filter B band images in further cal-
culations are not used. The Samsung Galaxy Note 3 smartphone
comprising integrated CMOS RGB image sensor with resolu-
tion of 13 MP was used for image acquisition. All images
were taken using the following settings: ISO—100, white bal-
ance—daylight, focus—manual, exposure time—fixed 200 ms.
2.2 Image Processing
In order to visualize the skin AF intensity decrease rates during
the photobleaching, the following image processing expression
was applied:
EQ-TARGET;temp:intralink-;e001;63;399NðCÞ¼½It0ðCÞ−ItðCÞ∕It0ðCÞ;(1)
where NðCÞrepresents normalized AF intensity decrease map
for each pixel (or pixel group) during the excitation period,
It0½C—AF image at the excitation start moment, It½C—AF
image after 20 s of continuous excitation. C—color component
of the RGB image—red (R), green (G), and blue (B), respec-
tively. The values of RGB components were defined from the
image data by a special program developed in MATLAB®.
Overall 50 patients with 150 different skin neoplasms
(or suspicious) were inspected in the clinic. For the detailed
image analysis 13 BCC and 1 atypical nevus were selected.
This study was approved by the Ethics Committee of the
Institute of Experimental and Clinical Medicine, University of
Latvia. All involved volunteers were informed about the study
and signed required consent.
3 Results and Discussion
A total of 10 solid and 3 ulcerating BCCs were selected for the
study. In all BCC cases (confirmed by cytological examination)
the AF images showed lowered AF intensity in malignant tissue
as compared with the healthy surrounding skin, which may be
attributed to decreased levels of fluorophores and increased
blood perfusion caused by the malignancy process.2,5,7AF spec-
tra from in vivo BCC under 405 nm excitation are characterized
by broad (450 to 750 nm) emission spectrum with maximum in
green spectral region (510 to 530 nm). In comparison with sur-
rounding healthy skin, the intensity of AF from malignant tissue
usually are lower, while the shape of the spectrum remains
unchanged. Moreover, the intensity of AF is strictly correlated
with the tumor pigmentation, specifically, the higher the pig-
mentation, the lower is the intensity of fluorescence.2,8In all
BCC cases, the G-band (corresponding to the AF maximum)
AF intensity images in comparison with R-band images showed
the more pronounced intensity contrast within the tumor tissue
Fig. 1 Images of ulcerating basal cell carcinoma (BCC). Filtered AF color image (a) at excitation start
moment, (b) the corresponding G-band image, and (c) normalized AF intensity decrease map. The
pseudo color scale represents (b) the G-band intensity range and (c) the normalized AF intensity variation
range.
Fig. 2 Images of solid BCC. Filtered AF color image (a) at excitation start moment, (b) the corresponding
G-band image, and (c) normalized AF intensity decrease map. The pseudo color scale represents (b) the
G-band intensity range and (c) the normalized AF intensity variation range.
Journal of Biomedical Optics 120502-2 December 2015 •Vol. 20(12)
JBO Letters
and surrounding healthy skin. Moreover, G-band AF decrease
maps showed more structured compositions at tumor area in
comparison with R-band AF decrease maps. In the cases of
solid BCCs the AF images showed clearly bordered tumor
areas with relatively low AF intensity in comparison with sur-
rounding healthy skin. Whereas the ulcerating BCCs can be
characterized by the high AF intensity in the ulcerating part sur-
rounded by clearly bordered tissue emitting relatively low AF
intensity. Furthermore, in all BCC cases the normalized AF
intensity decrease maps showed high AF intensity decrease
rates at the tumor areas with low AF intensity in comparison
with the surrounding healthy skin and the internal ulcerating
area.
Figure 1represents images of ulcerating BCC: (a) filtered AF
image at the excitation start moment, AF intensity G-band
image (b), and parametric map of normalized AF intensity
decrease rates in the green band (c). AF intensity image
[Fig. 1(b)] shows high AF intensity in the ulcerating part (path-
ology center), surrounded by clearly bordered tissue emitting
relatively low AF intensity. Furthermore, the AF intensity
decreasing is more intensive at the external tumor area expo-
sures in comparison with the surrounding healthy skin and the
internal ulcerating area [Fig. 1(c)].
Another case of solid BCC is presented in Fig. 2. G-band AF
intensity image [Fig. 2(b)] shows relatively low intensity within
the tumor area with clear margins between tumor and surround-
ing healthy skin. The tumor area also shows higher AF intensity
decrease rate [Fig. 2(c)] in comparison with the surrounding
healthy skin. Besides, the images presented in Fig. 2reveal a
small area with lowered fluorescence located at “five o’clock”
from the main tumor. The low AF intensity and high AF inten-
sity decrease rate in that skin area similarly indicates to cancer-
ous process, which is probably determined by multicentric
tumor growing process.
In addition, one atypical nevus was selected for the study
(Fig. 3). The nevus before surgical excision was suspected as
melanoma; histological analysis of the removed tissue samples
had confirmed three different types of tissue cells within the
lesion area. Specifically, the upper part of the pathology mostly
prevailed by intradermal nevus, the middle part by dysplastic
nevus, and the lower part by junctional nevus. Normalized AF
decrease distribution map [Fig. 3(c)], on the other hand, showed
the fastest intensity decrease in the lower (junctional nevus) and
upper side (intradermal nevus), while the middle part (dysplastic
nevus) of lesion photobleached slower. The observed different
AF photobleaching rates most probably are associated with dif-
ferent tissue fluorophore concentration, melanin content, locali-
zation, and metabolic state.
4 Conclusions
Smartphone AF imaging has shown potential for remote pri-
mary evaluation of cancerous or suspicious skin tissues. The
proposed noninvasive technique and method adequately (with
respect to the available literature data) represented planar distri-
bution of AF intensities in malignant and healthy tissues. More-
over, the temporal analysis of AF intensity during the photo-
bleaching showed a potential to be used as an additional indi-
cator for demarcation of suspicious tissues. It may find clinical
implementation, e.g., for primary evaluation of BCC, such as
determination of precise excision margins prior to surgery,
adequate selection of treatment method (in the case of multicen-
tric growing process the nonsurgical methods are more desirable
for the patients), as well as in selective application of immune
response modulators for BCC therapy. The proposed excitation
band around 405 nm covers the absorption maxima of several
porphyrines and may find applications in photodynamic diag-
nostics of superficial nonmelanoma lesions. The most intriguing
result of this research was the fact that AF photobleaching rate
map showed quantitative correlation with the histology tests in
the case of atypical dysplastic nevus. To explain, one can
assume that specific tissue fluorophores might have individual
bleaching kinetics features, which eventually could provide
information on fluorophore concentration and environmental
factors. The increased photobleaching rates in the tumor area
most probably indicate different fluorophore content composi-
tion affected by the tumor growing process, e.g., destruction of
collagen elastin cross-links along with decrease in NADH lev-
els. Undoubtedly, this phenomenon requires additional studies
to clarify the exact mechanism of uneven photobleaching of skin
fluorophores under continuous optical excitation.
Acknowledgments
This work was supported by the European Regional Develop-
ment Fund project “Innovative technologies for optical skin
diagnostics”(No. 2014/0041/2DP/2.1.1.1.0/14/APIA/VIAA/015).
Fig. 3 Color filtered AF image of skin atypical nevus (a) at the excitation start moment, (b) the corre-
sponding G-band image and (c) normalized AF intensity decrease map. The color scale at (b) image
represents G-band intensity range and at (c) image—range of normalized AF intensity variations.
Journal of Biomedical Optics 120502-3 December 2015 •Vol. 20(12)
JBO Letters
References
1. American Cancer Society, “Cancer facts & figures,”(2009), http://www.
cancer.org.
2. E. Drakaki et al., “Spectroscopic methods for the photodiagnosis of
nonmelanoma skin cancer,”J. Biomed. Opt. 18(6), 061221 (2013).
3. E. N. Marieb and K. Hoehn, Human Anatomy & Physiology, 7th ed.,
Pearson International Edition, San Francisco, pp. 152–170 (2006).
4. S. D. Rigel et al., Cancer of the Skin: Expert Consult, 2nd ed., pp. 99–
123, Elsevier Saunders, China (2011).
5. M. Panjehpour et al., “Laser-induced fluorescence spectroscopy for in
vivo diagnosis of non-melanoma skin cancers,”Lasers Surg. Med.
31(5), 367–373 (2002).
6. N. P. Galletly et al., “Fluorescence lifetime imaging distinguishes basal
cell carcinoma from surrounding uninvolved skin,”Br. J. Dermatol.
159(1) 152–161 (2008).
7. D. Chorvat and A. Chorvatova, “Multi-wavelength fluorescence life-
time spectroscopy: a new approach to the study of endogenous fluores-
cence in living cells and tissues,”Laser Phys. Lett. 6(3), 175–193
(2009).
8. E. G. Borisova, L. P. Angelova, and E. P. Pavlova, “Endogenous and
exogenous fluorescence skin cancer diagnostics for clinical applica-
tions,”IEEE J. Quantum Electron. 20(2), 7100412 (2014).
9. H. Zeng et al., “The dynamics of laser-induced changes in human skin
autofluorescence experimental measurements and theoretical model-
ing,”Photochem. Photobiol. 68(2), 227–236 (1998).
10. M. E. Darvin et al., “Optical methods for noninvasive determination of
carotenoids in human and animal skin,”J. Biomed. Opt. 18(6), 061230
(2013).
11. H. Wang et al., “Improving skin Raman spectral quality by fluorescence
photobleaching,”Photodiagn. Photodyn. 9(4), 299–302 (2012).
12. J. S. Goodwin and A. K. Kenworthy, “Photobleaching approaches to
investigate diffusional mobility and trafficking of Ras in living
cells,”Methods 37(2), 154–164 (2005).
13. G. Hennig, H. Stepp, and A. Johansson, “Photobleaching-based method
to individualize irradiation time during interstitial 5-aminolevulinic acid
photodynamic therapy,”Photodiagn. Photodyn. Ther. 8(3), 275–281
(2011).
14. C. Jayyosi et al., “Photobleaching as a tool to measure the local strain
field in fibrous membranes of connective tissues,”Acta Biomater. 10(6),
2591–2601 (2014).
15. J. Spigulis, A. Lihachev, and R. Erts, “Imaging of laser-excited tissue
autofluorescence bleaching rates,”Appl. Opt. 48(10), D163–D168
(2009).
16. A. Lihachev et al., “Investigation of in-vivo skin autofluorescence life-
times under long-term cw optical excitation,”Quantum Electron. 44(8),
770–773 (2014).
17. I. Kuzmina et al., “Study of smartphone suitability for mapping of skin
chromophores,”J. Biomed. Opt. 20(9), 090503 (2015).
18. Statista GmbH, “Number of smartphone users* worldwide from 2014 to
2019 (in millions),”(2015), http://www.statista.com/statistics/330695/
number-of-smartphone-users-worldwide/ (26 November 2015).
19. S. A. Lamel et al., “Application of mobile teledermatology for skin
cancer screening,”J. Am. Acad. Dermatol. 67(4), 576–581 (2012).
20. S. Kroemer et al., “Mobile teledermatology for skin tumour screening:
diagnostic accuracy of clinical and dermoscopic image tele-evaluation
using cellular phones,”Br. J. Dermatol. 164(5), 973–979 (2011).
21. J. Robert et al., “Early detection of malignant melanoma: the role of
physician examination and self-examination of the skin,”CA Cancer
J. Clin. 35, 130–151 (1985).
22. T. Maier et al., “Accuracy of a smartphone application using fractal
image analysis of pigmented moles compared to clinical diagnosis
and histological result,”J. Eur. Acad. Dermatol. Venereol. 29(4),
663–667 (2015).
23. S. Dreiseitl et al., “Computer versus human diagnosis of melanoma:
evaluation of the feasibility of an automated diagnostic system in a pro-
spective clinical trial,”Melanoma Res. 19(3), 180–184 (2009).
24. C. Massone et al., “Melanoma screening with cellular phones,”PLoS
ONE 2(5), e483 (2007).
Journal of Biomedical Optics 120502-4 December 2015 •Vol. 20(12)
JBO Letters
