Академический Документы
Профессиональный Документы
Культура Документы
a r t i c l e i n f o
abstract
Article history:
Received 9 September 2010
Accepted 6 November 2011
Strabismus is a pathology that affects about 4% of the population, causing aesthetic problems, reversible
at any age; however, problems that can also cause irreversible muscular alterations, and alter the vision
mechanism. The Hirschberg test is one of the exams used to detect this pathology. The application of
high technology resources to help diagnose and treat ophthalmological conditions is, lamentably, not
commonly found in the sub-specialty of strabismus. This work presents a methodology for automatic
detection of strabismus in digital images through the Hirschberg test. For such, the work was organized
into four stages: (1) nding the region of the eyes; (2) determining the precise location of the eyes;
(3) locating the limbus and brightness; and (4) identifying strabismus. The methodology has produced
results on the range of 100% sensibility, 91.3% specicity and 94% for the correct identication of
strabismus, ensuring the efciency of its geostatistical functions for the extraction of eye texture and
for the calculation of the alignment between the eyes on digital images obtained from the
Hirschberg test.
& 2011 Elsevier Ltd. All rights reserved.
Keywords:
Medical image
Strabismus
Hirschberg test
Geostatistical functions
Image processing
Pattern recognition
Support vector machine
1. Introduction
Strabismus is an abnormal condition that makes the eyes lose
their parallelism between. While an eye stares at a frontal point,
the other turns aside, or even upwards and downwards. Because
of this, the brain receives two images with different focuses,
instead of two images that converge into a single spot. There are
several types of strabismus: the affected eye can be yawed toward
the nose (convergent strabismus); it can turn aside (divergent
strabismus); or it turns upwards or downwards (vertical strabismus). There can be a combination of horizontal and vertical
yaw in the same patient, as, for example, toward the nose and
upwards.
In general, it can be said that the mechanical component of
strabismus, in other words, the esthetic aspect of the yaw, can be
treated at any age. On the other hand, the sensorial disturbances
are more signicant, and are only treatable at a certain period in
ones lifethe stage of plasticity of the visual system, which
lingers on till the age of nine. Thus, as the main sensorial
complication of a yaw is the strabismic amblyopya, its treatment
must be initiated as soon as a strabismus condition with amblyogenic characteristics is detected [1,2].
To diagnose strabismus, the following exams are performed:
visual acuity, eye background, external examination of the eyes
(cornea, sclera, conjunctiva, iris, lens, etc.), and eye movement
exam, obtained by means of the Cover test and the Hirschberg
test. The Hirschberg test consists basically of sending a thin beam
of light into the patients eyes in order to verify if the reection in
each eye is located at the same place on both corneas. Besides
these exams, there are the devices called electronic synoptophores, which measure strabismus via the projection of two
separate and dissimilar images in the same position in space.
Despite the increasing use of cutting-edge resources to help
with the diagnosis and the treatment of various ophthalmological
conditions, the sub-specialty of strabismus has not been given the
same importance. Considering the fact that it is not easy to nd
professionals with enough experience in this sub-area away from
large urban centers a fact that makes precocious diagnoses
more difcult these technologies have become essential in cities
father away from those more advanced centers.
In 1998, the Brazilian government created the Unied Health
System (SUS). This system provides health assistance from
simple ambulatory assistance to organ transplantations ensuring integral, universal and cost-free health benets for the entire
population. Within this program, and in harmony with the
principles and dictums of the SUS, the Health Program at Schools
136
J. Dallyson Sousa de Almeida et al. / Computers in Biology and Medicine 42 (2012) 135146
2. Theoretical basis
This section presents the theoretical basis necessary for the
understanding of the proposed methodology.
2.1. Strabismus
Strabismus, one of the commonest ophthalmologic alternations in childhood, can be dened as an abnormal binocular
interaction between the eyes, where the same image does not
reach the fovea2 of both eyes at the same time; consequently, the
eyes do not xate on the same image.
Once the position of each eye (center of the pupil) is determined, relative to a reference (either the observed point or the
observation point), i.e. the directions of each axis (either
the visual or the pupillary point), strabismus may be dened as
the difference between the expected alignments, i.e. the angle
between the ocular directions, corresponding to a disturbance of
2
The fovea is located in the optical axle of the eye, on which is projected the
image of the focused object, and the image formed on it is very sharp [12].
J. Dallyson Sousa de Almeida et al. / Computers in Biology and Medicine 42 (2012) 135146
the binocular positional relation, relative to a given point (normally, the object toward which the sight is directed) [13]. Fig. 1
illustrates the occurrence of strabismus.
The symptoms and the consequences of strabismus differ
according to the age at which it appears and the way it manifests
itself. Strabismus that appears before the age of 6 has an
adaptation mechanism that makes the image created in the
yawed eye be suppressed, and, as a result, the patient does not
present diplopia.3 However, sight diminishment occurs (amblyopya or lazy sight) in the yawed eye. On the other hand, if a
person becomes strabic after 6 years of age, then this person will
present diplopia: each eye will focus the image on to different
positions, relative to the yaw. In a child, diplopia is periodical and
leads to suppression. This suppression consists of a cortical
mechanism of elimination of the image caught by the yawed
eye, something that occurs only in children who still have
cerebral plasticity.
Several techniques can be applied to the treatment of strabismus with the objective of restoring muscular balance and solving
the problem of amblyopya. The medical treatment commonly
used is: prescription of glasses, execution of orthoptical exercises,
and obstruction of the xating eye, alternating with the other eye.
If the medical treatment does not sufce, surgery may be
recommended to ensure the retrocession of the weakened ocular
muscles.
2.2. The Hirschberg method
In order to evaluate the strabismus yaw by using luminous
focus, one should initially describe the Hirschberg test that
calculates the approximate magnitude of the yaw relative to
the luminous reection displacement of the cornea in the
non-xating eye, taking into account the center of its ocular
globe. Depending on the reection location incidence, with
respect to the complex limbus-iris-pupil, one can infer the
magnitude of the yaw. Alternatively, in order to avoid the
variations resulting from the size of the pupil, one may correlate
the luminous reection to the center of the cornea and the limbus
[14]. The term corneal luminous reection is unsuitable, for it is
not a reection from outside the cornea. What we can rst see as
a luminous reection is actually the reection of Purkinjes image,
which is a virtual image located behind the pupil [15].
When examining an individual by means of the Hirschberg
test in order to diagnose strabismus, one must observe that the
xating eye has the rst Purkinjes image aligned to its optical
center, and consequently the other eye, the non-xating eye, is
the eye in which the yaw must be observed. The yaw is inferred
by comparing the reection of the light in the anterior surface of
the cornea with its optical center and by detecting whether there
is a misalignment. As it is difcult to determine on a non-xating
eye its precise location, the yaw will be evaluated in relation to
the anatomic center of the eye, or, in other words, in relation to
the center of the pupil. One can notice from this description the
existence of another variable interfering with the observation of
3
Diplopia consists in the perception of the same object in two different
spatial locations (in retina).
137
the yaw; and that is the Kappa angle.4 This angle must be
measured for that eye and must be taken into consideration
when examining the reex. However, other factors interfere
with the relative positioning of the luminous reection on the
non-xating eye in relation to the position the reex assumes in
the xating eye. These factors are: corneal curvature, the size of
both cornea and eye, and refraction. If the data obtained from
both eyes are too dissimilar, then these can disturb the evaluation; so much so that, when attempting to analyze or quantify the
yaw by means of the Hirschberg method, one must take all of
these factors into consideration [16,17].
2.3.1. Semivariogram
The curve relating the semivariance as a function of the
distance of a point is called Semivariogram. The greater the
distance between the samples, the greater will be the semivariance; and the smaller the distance between them, the smaller
will be the semivariance.
Semivariogam is dened by
Nh
gh
1 X
x y 2
2Nh i 1 i i
2.3.2. Semimadogram
Semimadogram is the mean of the absolute difference measured in the pairs of the sample, as a function of distance and
4
Angle formed by the visual line and the axle of the pupil.
138
J. Dallyson Sousa de Almeida et al. / Computers in Biology and Medicine 42 (2012) 135146
"
sh
to
ec
v
ion
)
r (h
Angular tolerance ct
e
Dir
BW h
t
wi
#1=2
8
2.4. Validation
d
an
Lag tolerance
Nh
1 X 2
y m2 h
Nh i 1 i
Lag2
Lag1
Lag3
Lag4
Lag increment
X
Fig. 2. Parameters used in the calculation of geostatistical functions [10].
1 X
9x y 9
2Nh i 1 i i
mh
2.3.3. Covariogram
The covariogram measures the correlation between two variables.
In Geostatistics, covariance is calculated as the variance of the sample
minus the value of the variogram. The covariance function tends to
increase as the variables values are closer to each other i.e., when
h0; and tends to decrease as these values are farther away from
each other, or nearer to the limit. Covariogram is dened by
Nh
Ch
1 X
x y mh m h
Nh i 1 i i
where mh is the mean value of the vectors origins, and m h is the
mean value of the vectors extremities
Nh
mh
1 X
x
Nh i 1 i
mh
1 X
y
Nh i 1 i
Nh
2.3.4. Correlogram
The correlation function (correlogram) is the normalized
version of the covariance function. The coefcients of correlation
range from 1 to 1. The correlation is expected to be higher for
units close to each other (correlation 1 for distance zero) and it
tends to zero when the distance between the units increases [10].
Correlation is dened by
rh
Ch
sh s h
sh
Nh
1 X 2
x m2h
Nh i 1 i
#1=2
7
J. Dallyson Sousa de Almeida et al. / Computers in Biology and Medicine 42 (2012) 135146
139
140
J. Dallyson Sousa de Almeida et al. / Computers in Biology and Medicine 42 (2012) 135146
12
The radiuses intervals considered in this work were determined through an
analysis done on the image base used in tests.
J. Dallyson Sousa de Almeida et al. / Computers in Biology and Medicine 42 (2012) 135146
13
141
The angle formed by the optical axle and the xating line.
Computerized exam to measure the curvature of the corneal surface.
18
The computerized ceratoscopy or topography of the cornea is the exam
through which a qualitative and quantitative analysis of the corneal astigmatism
can be done.
17
142
J. Dallyson Sousa de Almeida et al. / Computers in Biology and Medicine 42 (2012) 135146
Semimadogram
Semivariogram
Correlogram
Covariogram
All
%
TP
TN
FP
FN
SE
SP
PPV
NPV
AC
408
392
383
369
411
99
90
76
70
103
8
16
29
31
5
25
42
52
70
21
94.23
90.32
88.05
84.06
95.14
92.53
84.96
72.38
69.30
95.38
98.08
96.08
92.96
92.25
98.78
79.84
68.19
59.38
50.00
83.07
93.89
89.26
85.00
81.30
95.19
Fig. 8. Location of the eyes. (a) Correct location of the eyes, (b) failure in the
location of the eyes and (c) eye region classied as non-eye.
Fig. 7. Automatic detection of the region of the patients eyes. (a) Patient without glasses, (b) region of the patients eyes (a) detected. (c) Patient wearing glasses, (d) region
of the patients eyes (b) detected.
J. Dallyson Sousa de Almeida et al. / Computers in Biology and Medicine 42 (2012) 135146
Fig. 9. Examples of images where the location of the eyes was performed
correctly.
143
limbus of the right eye (RE), we can see that its radius is smaller
than that of the limbus in the left eye (LE), with a difference above
2 pixels. Thus, considering that the methodology takes the
smallest limbus for reference, in the case of this possibility, it
would be impossible to locate the limbus in the LE correctly.
Considering the location of brightness on the 40 images taken
of the correct location of the limbus, we have obtained a matching
rate of 100% for location of brightness on both eyes. In Fig. 17a
and b we show some examples of the correct location of brightness by using the methodology.
4.4. Detection of strabismus
In this section, we present the results obtained on the detection stage of strabismus. We will consider the 40 patients whose
limbus and brightness were located on the previous stage.
Fig. 11. Analysis of Fig. 10a. (a) Output image of the accumulation vector after
application of HT to the image of the eye region. (b) Right eye candidate wrongly
classied as non-eye by the SVM.
Fig. 12. Analysis of Fig. 10b. (a) Output image of the accumulation vector after
application of HT to the image of the region of the eyes. (b) and (c) left and right
eye candidates, respectively, classied correctly as non-eye by the SVM.
Fig. 13. Analysis of Fig. 10c. (a) Output image of the accumulation vector after the
application of the HT to the image of the region of the eyes. (b) Right eye
candidates correctly classied as eye. (c) Left eye candidates erroneously classied
as eye by the SVM.
Fig. 14. Analysis of Fig. 10d. (a) Output image of the accumulation vector after the
application of the HT to the image of the region of the eyes. (b) Left eye candidates
erroneously classied as eye by the SVM.
144
J. Dallyson Sousa de Almeida et al. / Computers in Biology and Medicine 42 (2012) 135146
Fig. 15. Examples of images where the methodology correctly found the region of limbus.
Table 2
Result obtained from the processing of the 40 patients images for verifying the
alignment of the eyes compared to the specialists analysis. Ppatient, Sspecialist and Mmethodology.
P
Fig. 16. Image where the methodology failed in locating the limbus.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
RE
LE
CPC
RR
distY
distX
LR
distY
distX
34
26
33
25
30
34
31
32
34
27
28
34
31
25
33
26
20
32
27
21
20
24
21
18
23
20
25
19
21
23
16
26
22
21
22
24
22
25
24
19
4
2
1.87
3
3.73
4
2.80
6.78
0.97
3.85
1.92
0
3.87
1.92
5.82
0
3.0
4
1
2
2
2
1.90
2
2
1
2
2
1
1.91
2
1
1
2
3.82
0.95
1
2
2.87
1
4
5
1.88
1
0
2
3.74
1.93
1.94
0
0.96
2
0
1.92
0
4
1.0
1
1
1
1
1
0
0
4
2
2
0
0
0.95
0
0
1
0
0
0
0
1
0.95
2
34
28
31
25
28
34
29
31
33
26
27
34
30
24
32
26
20
33
27
21
20
25
20
18
23
20
25
19
21
22
16
26
23
21
21
23
24
25
23
20
0
0.92
1
0
3
4
0
3
6
3
1
5
2
2
3
2
1.0
2
3
1
3
1.92
1
3
1
1
1
0
1
2
8
1
1.91
2
5
0
1.83
0
2
0
8
1.85
0
3
0
1
4
1
1
0
0
0
1
1
0
1
0
0
0
0
2
0.96
1
2
2
1
2
0
0
3
0
1
0
1
2
0
0
1
1
0.95
1
0.93
0.93
1
0.93
1
0.93
0.96
0.97
0.96
0.96
1
0.96
0.96
0.97
1
1.0
0.97
1
1
1
0.96
0.95
1
1
1
1
1
1
0.95
1
1
0.95
1
0.95
0.96
0.91
1
0.95
0.95
HDIF
4
1.07
1.87
3
0.73
0
2.80
3.78
5.02
0.85
0.92
0
1.87
0.08
2.82
2
2.0
2
2
1
1
0.08
0.90
1
1
0
1
2
0
0.08
6
0
0.91
0
1.18
0.96
0.83
2
0.87
1
VDIF
4
3.14
0.87
2
0
1
0.25
0.93
0.94
0
0.96
1
1
0.92
0
3
1.0
1
1
1
1
0.04
1
2
2
1
0
0
0
2.04
0
1
1
1
2
0
0
0
0.04
1.2
Result
S
Yes
Not
Not
Yes
Not
Not
Yes
Not
Not
Not
Yes
Not
Not
Not
Yes
Yes
Not
Not
Not
Not
Not
Not
Yes
Yes
Yes
Not
Not
Not
Not
Not
Yes
Yes
Not
Not
Yes
Yes
Not
Yes
Not
Yes
Yes
Yes
Not
Yes
Not
Not
Yes
Yes
Yes
Not
Not
Not
Not
Not
Yes
Yes
Not
Not
Not
Not
Not
Not
Not
Yes
Yes
Not
Not
Not
Not
Yes
Yes
Not
Not
Not
Yes
Not
Not
Not
Not
Yes
Fig. 18. Images where the application of the Hirschberg test failed to: identify the
normal patient (a and b) and the strabic patient (c and d).
opposite signal, annulling the appearance of the yaw and giving the
notion of an adequate binocular position, in spite of a yaw [13].
Analyzing Table 2, without considering the patient on whom
the Hirschberg test failed, we have obtained the values of TP 10,
J. Dallyson Sousa de Almeida et al. / Computers in Biology and Medicine 42 (2012) 135146
Fig. 19. Images where the methodology failed in determining the precise location
of the limbus. (a) and (b) RE and LE of patient 2. (c) and (d) RE and LE of patient 30.
FP 2, TN21 and FN0. Thus, we can verify that the methodology achieved 100% of sensitivity, 91.3% of specicity and 94% of
matching for the 33 remaining images. The two patients
considered strabic, even if they were not strabic, have been
considered as such because of the precision error that occurred
when locating the limbus.
Analyzing Fig. 19a, which presents the right eye of patient 2 from
Table 2, we have noticed that the region of the limbus was not
precisely located, because the center of the limbus was to be closer
to the center of the pupil. This resulted in the vertical misalignment
of the right eye in relation to the left eye (Fig. 19b), making,
therefore, the result of VDIF, of 3.14, indicating the presence of
strabismus, which contradicts the specialists diagnosis.
In Fig. 19d, representing the left eye of patient 30 from Table 2,
one can see that, similar to the last error, the region of the limbus
was not correctly located, leaving a small remainder of the limbus
outside the region located. This caused an increase in the vertical
misalignment of the left eye in relation to the right eye (Fig. 19c),
making the result of VDIF, of 2.04, reveal the presence of
strabismus, contradicting, in this way, the specialists diagnosis.
145
5. Conclusion
References
This work recommends the use of image processing techniques, geostatistical functions and support vector machines for the
automatic detection of strabismus on digital images by using the
Hirschberg method. Along with this study, some other contributions can be veried. The rst one may be seen to occur on the eye
location stage, where an innovating combination of techniques to
locate the eyes on human faces is proposed, using homomorphic
ltering, the Hough Transform, geostatistical functions, stepwise
discriminant analysis, SVM and EMA similarity measurements.
Other contributions have to do with the features extraction stage,
where geostatistical functions are used to extract texture information
from the eyes, allowing then the discrimination of eye regions from
the other regions with a fair amount of precision. The third and main
contribution is the creation of a methodology that gives support to
the automatic identication of strabismus on digital images.
On the stage of automatic detection of eye region by using
projections, the gradient magnitude exhibited a matching rate of
100% for the patients images. On this stage, we concluded that
the use of homomorphic ltering contributed to the overall
matching, since the technique of illumination adjustment made
it more uniform on the image.
On the second stage, the methodology obtained a matching
rate of 95.19%, with the combination of four geostatistical functions, such as texture descriptors in the extraction of features.
Nevertheless, although such results may be seen as very promising, it is still necessary to increase the diverseness of the sample
faces, so that a more robust and generic methodology can be
developed. However, we can conclude that the obtained results
do give the matter its due importance regarding the new
approaches based on geostatistical functions for describing the
texture of eye regions on digital pictures of faces. Still, on the
[1] G.V. Noorden, E. Campos, Binocular Vision and Ocular Motility: Theory and
Management of Strabismus, Mosby Inc, 2001.
[2] J.P. Diaz, C.S. Dias, Strabismus, Butterworth Heinemann, Woburn, Massachusetts, EUA, 2000.
[3] A.S. Jolson, H.R. Myler, A. Weeks, Apparatus for evaluating eye alignment, US
Patent 5,094,521, 1992.
[4] E.M. Helveston, F.H. Orge, R. Naranjo, L. Hernandez, Telemedicine: strabismus
e-consultation, J. Am. Assoc. Pediatr. Ophthalmol. Strabismus 5 (5) (2001)
291296.
[5] M.W. Quick, R.G. Boothe, A photographic technique for measuring horizontal
and vertical eye alignment throughout the eld of gaze, Invest. Ophthalmol.
Vis. Sci. 33 (1) (1992) 234.
[6] I. Subharngkasen, Successful amblyopia therapy by using synoptophore,
J. Med. Assoc. Thailand Chotmaihet thangphaet 86 (2003) S556.
[7] H.J. Kim, W. Kim, Eye detection in facial images using zernike moments with
SVM, ETRI J. 30 (2) (2008) 335337.
[8] J. de Almeida, A. Silva, A. Paiva, Automatic eye detection using semivariogram
function and support vector machine, in: Seventeenth International Conference
on Systems Signals and Image Processing IWSSIP 2010, 2010, pp. 174177.
[9] O.S.S. Souza Jr., A.C. Silva, Z. Abdelouah, Personal identication based on iris
texture analysis using semivariogram and correlogram functions, Int. J.
Comput. Vis. Biomech. 2 (1) (2009) 121129.
[10] A. Silva, P. Carvalho, M. Gattass, Analysis of spatial variability using
geostatistical functions for diagnosis of lung nodule in computerized tomography images, Pattern Anal. Appl. 7 (3) (2004) 227234.
[11] J.D. Sousa de Almeida, Metodologia Computacional para Deteco Automtica de
Estrabismo em Imagens Digitais atravs do Teste de Hirschberg., http://www.
tedebc.ufma.br//tde_busca/arquivo.php?codArquivo=430, 2009.
[12] L.C. Junqueira, J. Carneiro, Histologia basica. 8 Edic- a~ o, Guanabara Koogan.
[13] H. Bicas, Estrabismos: da teoria a pratica, dos conceitos a s suas operacionalizac- oes, Arq. Bras. Oftalmol. 72 (5) (2009) 585615.
[14] J.L. Mims, R.C. Wood, Proportional (fractional) displacement of the Hirschberg corneal light reection (test): a new easily memorized aid for strabometry; photogrammetric standardization (calibration), Binocul. Vis.
Strabismus Q. 17 (3) (2002) 192.
[15] K. Wright, P. Spiegel, L. Thompson, Handbook of Pediatric Strabismus and
Amblyopia, Springer Verlag, 2006.
[16] P.E. Romano, Individual case photogrammetric calibration of the Hirschberg
Ratio (HR) for corneal light reection test strabometry, Binocul. Vis. Strabismus Q. 21 (1) (2006) 45.
146
J. Dallyson Sousa de Almeida et al. / Computers in Biology and Medicine 42 (2012) 135146
Anselmo Cardoso de Paiva received BSc in civil engineering from Maranha~ o State
UniversityBrazil in 1990, a MSc in civil engineering-Structures and a PhD in
Informatics from Pontiphical Catholic University of Rio de JaneiroBrazil in 1993
and 2002. He is currently a Professor at the Informatics department, Federal
University of Maranha~ oBrazil. His current interests include medical image
processing, geographical information systems and scientic visualization.
Aristofanes Corr
ea Silva received a PhD degree in Informatics from Pontiphical
Catholic University of Rio de JaneiroBrazil in 2004. Currently he is a Professor at
the Federal University of Maranha~ o (UFMA), Brazil. He teaches image processing,
pattern recognition and programming language. His research interests include
image processing, image understanding, medical image processing, machine
vision, articial intelligence, pattern recognition and, machine learning.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.