REVIEW

CURRENT
OPINION Cataract grading systems: a review of past
and present
Helena E. Gali, Ruti Sella, and Natalie A. Afshari

Purpose of review
To provide a comprehensive summary of past cataract grading systems, how they have shaped
current grading systems, and the developing technologies that are being used to assess and grade
cataracts.
Recent findings
This summary of cataract grading systems examines the development and limitations that existed in past
grading systems and how they have shaped the grading systems of present time. The Lens Opacities
Downloaded from http://journals.lww.com/co-ophthalmology by BhDMf5ePHKbH4TTImqenVIiuKVF7qTxsO7egTcR85zhfbHf7wqjCQCHIL2A3JkMwaD1TuyJxyDU= on 01/10/2019

Classification System III (LOCS III) system is currently used both clinically and for research purposes. Recent
advancements in imaging technologies have allowed researchers to create automatic systems that can
locate lens landmarks and provide cataract grading scores that correlate well with LOCS III clinical grades.
Utilizing existing technologies, researchers demonstrate that fundus photography and optical coherence
tomography can be used as cataract grading tools. Lastly, deep learning has proved to be a powerful tool
that can provide objective and reproducible cataract grading scores.
Summary
Cataract grading schemes have provided ophthalmologists with a way to communicate clinical findings
and to compare new developments in diagnostic technologies. As technologies advance, cataract grading
can become more objective and standardized, allowing for improved patient care.
Keywords
automatic systems, cataract grading, deep learning, Lens Opacities Classification System III, optical
coherence tomography

INTRODUCTION research findings, providing a uniform way to

Cataract grading schemes have evolved since their
implementation over 30 years ago. West and Tay-
lor called for a standardized way to characterize
and grade cataracts in 1986, and since then, the
Oxford Clinical Grading System and early itera-
tions of the Lens Opacities Classification System
(LOCS) have contributed to the formation of the
LOCS III system that is currently used today
&&
[1,2 ]. Grading systems are used both clinically
and for the purposes of research, and the use of a
grading schema plays an important role in both
settings. Grading of cataracts may aid in the com-
munication between physician and patient, as
cataract grades can provide patients with under-
standable information about the diagnosis, pro-
gression, and importance of treatment for their
cataracts. Additionally, grading a cataract allows
the ophthalmologist to follow the cataract over a
period of time to assess progression and to share
data with other physicians in a clear and standard-
ized way. A grading scheme also standardizes
communicate findings and compare whereas a
certain grade of cataract does not result in the
same degree of visual impairment for each patient,
it does provide insight into how severe vision
impairment may be. As grading systems for cata-
racts have evolved, so has technology that allows
clinicians to better assess, grade, and monitor
cataract formation and progression. Grading sys-
tems combined with advancements in technology
provide patients with increased information and
improvements in their care.
Department of Ophthalmology, Shiley Eye Institute, University of Cal-
ifornia San Diego, La Jolla, San Diego, California, USA
Correspondence to Natalie A. Afshari, MD, Department of Ophthalmol-
ogy, Shiley Eye Institute, University of California San Diego, 9415
Campus Point Dr #0946, La Jolla, San Diego, CA 92093-0946, USA.
Tel: +1 858 822 1569; fax: +1 858 822 1514;
e-mail: naafshari@ucsd.edu
Curr Opin Ophthalmol 2019, 30:13–18
DOI:10.1097/ICU.0000000000000542

1040-8738 Copyright ß 2018 Wolters Kluwer Health, Inc. All rights reserved. www.co-ophthalmology.com

Copyright © 2018 Wolters Kluwer Health, Inc. All rights reserved.

Cataract surgery and lens implantation
KEY POINTS
 This review reflects the historical progression of past
cataract grading systems to the grading systems used
at present.
 The LOCS III grading scheme introduced in 1993 is still
used by clinicians and by investigators when
comparing newer cataract grading technologies.
 Advancements in automatic grading systems have
proved to be a reproducible and objective way of
grading cataracts, although they do face some
limitations and require further refinement.
 Deep learning and neural networks are proving to be a
promising way to further advance the field of cataract
diagnosis, grading, and treatment.
HISTORICAL GRADING SYSTEMS
To understand cataract grading systems at present, it
is helpful to reflect on the development of grading
systems of the past. In the mid-1980s, a combina-
tion of resolution target projection ophthalmoscopy
(acuity scope), the Oxford Clinical Grading system,
and photographs were recommended to be used to
assess cataracts [3]. Although resolution target pro-
jection ophthalmoscopy is no longer used and pho-
tographic technologies have improved greatly,
grading systems have remained a standard clinical
tool. The Oxford Clinical Grading system of 1983
was created to be a comprehensive way to score a
variety of cataract features and included the follow-
ing elements: anterior clear zone thickness, anterior
subcapsular opacity, posterior subcapsular opacity,
cortical spoke opacity, waterclefts, vacuoles, retro-
dots, focal dots, nuclear brunescence and white
nuclear scatter [3]. The features listed were given a
grade on a scale from 0 to 5. There existed complex-
ity behind each numerical grade; for example,
‘vacuoles [were] graded according to their frequency
within an 8 mm diameter area by comparison with
standard diagrams showing frequency on a logarith-
mic scale’ [3]. Results of using the Oxford Clinical
Grading system were mixed when tested among four
physicians, as there was varying degree of agreeabil-
ity among the various elements tested. Inter-
observer mean weighted kappa values ranged from
0.87 (excellent) to 0.36 (poor), prompting authors of
the 1987 article to caution users that there may be
variability in the system [3].
Perhaps in contrast to the Oxford Clinical Grad-
ing System, several groups of researchers around this
time set out to create a simplified grading scheme to
increase efficiency in eye surveys and epidemiologi-
cal studies of cataract patients, especially in rural
14 www.co-ophthalmology.com
Copyright © 2018 Wolters Kluwer Health, Inc. All rights reserved.
communities. Mehra and Minassian [4] had four
ophthalmologists and trained ophthalmic assistants
grade lens opacity on a scale from 0 to 5 based on the
amount of red reflex visualized by direct ophthal-
moscope. Kappa values for this study ranged from
1.0 to 0.60, and there was good correlation between
visual impairment and the grades given for cata-
racts, most likely owing to the simplicity of the
grading system [4]. Another study at the time also
set out to create a model to take standardized slit-
lamp photographs and to use a simplified 0–4 grad-
ing scheme for both nuclear and cortical cataracts
[5]. The resulting kappa values of this study also
proved high reproducibility with this model [5].
Lastly, the Beaver Dam Eye Study, referred to by
most as the Wisconsin cataract grading system,
grades nuclear sclerosis on a scale from 0 to 5 and
cortical cataracts in a sum of nine separate lens
segments [6]. The Wisconsin cataract grading sys-
tem is often used at present when comparing clini-
cally used grading systems with new automated
techniques in cataract grading [6]. These studies
provided an example that although much less com-
plex than the Oxford Clinical Grading system, a
simple grading scheme could correlate well with
the patient’s experience of their cataract. But while
visual experience is a patient’s primary concern,
what did clinicians need out of their grading scheme
in order to enhance patient care and treatment?
GRADING SCHEMES FROM PAST
TO PRESENT
The LOCS was created with the goal to provide a
reliable and reproducible grading scheme that was
simpler for clinicians to use while also accounting
for important cataract characteristics [7]. Intro-
duced in 1988, the LOCS I set out to classify cataracts
using a slit lamp examination or with slit lamp and
retroilluminated photographs [7]. The LOCS I sys-
tem evaluated the following: opacification in the
cortical and posterior subcapsular zones and inten-
sity of opalescence in the nucleus. Nuclear color and
opalescence were evaluated separately, as it was
found that color had less to do with cataract severity
than previously thought [7]. Of note, LOCS I marked
a shift in focus from grading cataract severity based
on visual acuity and instead basing it on cataract
morphology, as unaggregated changes in the cortex
had little role in visual acuity changes [7]. Instead,
clustered aggregation provided clinicians a way to
tell a patient that they have an early cataract, inde-
pendent of visual symptoms.
The LOCS II was developed the following year,
improving upon LOCS I by implementing ways to
differentiate degrees of cortical, subcapsular, and
Volume 30  Number 1  January 2019

nuclear opacification in addition to adding color
photographs to be used as standards for comparison
[8]. Visual acuity was again left out of the grading
scheme, as there are other factors that contribute to
changes in vision that could be unrelated to lens
abnormality. LOCS II demonstrated good intra-
observer and inter-observer agreements at the slit
lamp and in photographic readings [8]. The slit lamp
exam proved to be slightly more sensitive at detect-
ing opacification than photographs, likely owing to
limitations in photographic technologies at the
time [8].
The LOCS III was created in 1993 to improve
upon limitations in LOCS II. LOCS III expanded the
scales used in LOCS II to better capture the early
stages of cataract formation in the grading scheme
&&
[2 ]. It examines nuclear opalescence (NO) and
nuclear color (NC) on a scale from 1 to 6, cortical
cataracts (C) on a scale from 1 to 5, and posterior
subcapsular cataracts on a scale from 1 to 5 (Fig. 1)
&&
[2 ]. Although a finer grading scheme lowers con-
cordance among measurements, the resulting
increase in sensitivity was chosen as a worthy reason
for the change. Of note, the original 1993 article
highlighting improvements of LOCS III over LOCS II
compared photographic evaluation of cataracts as
opposed to in-vivo evaluation based on the ability to
standardize photographic protocols and to reduce
FIGURE 1. The Lens Opacities Classification System III Standards introduced in 1993. Data from [2 ]. &&
1040-8738 Copyright ß 2018 Wolters Kluwer Health, Inc. All rights reserved.
Copyright © 2018 Wolters Kluwer Health, Inc. All rights reserved.
Cataract grading systems Gali et al.
&&
user variation in the slit lamp examination [2 ].
This comparison marked an important change in
welcoming the idea that photographic assessment
could be superior to the slit lamp exam, perhaps
creating a gateway for digital analysis to provide a
more objective measure of cataract than the subjec-
tive clinician assessment. At present, the LOCS III
system is still used in varying degrees in clinical
practice, however, its clinical impact on decision-
making and timing of surgery is questionable.
CORRELATION BETWEEN GRADE AND
VISUAL SYMPTOMS FOR THE PATIENT
Before discussing cataract grading systems further, an
important question to ask is how it relates to the
symptoms experienced by the patient. If cataract
surgery is to be performed regardless of cataract
severity and instead based on patient experience of
their cataract, what benefit does a comprehensive
grading scheme provide? From an epidemiological
standpoint, it is important to characterize cataract
type and severity to better understand population
health and treatment need, but how do cataract
grades perform on a patient-to-patient basis? In a
2012 study examining LOCS III grading and visual
functioning, it was noted that vision-specific func-
tioning decreased in a statistically significant way for
www.co-ophthalmology.com 15
Cataract surgery and lens implantation
different cataract types at different grades using
&
the LOCS III [9 ]. Overall, the trend makes sense in
that as grade increased, vision-specific functioning
decreased. The nuance, however, lies in the fact that
each cataract type produced symptoms at different
grades. Nuclear cataracts affected functioning at
grades 4 for opalescence and 5 for color, cortical
cataracts affected functioning at grade 3, and poste-
rior subcapsular cataracts affected functioning at all
&
grades [9 ]. Although it may make sense to a clinician
that a posterior subcapsular cataract produces visual
symptoms in a more significant way than does a
nuclear cataract, from a patient perspective it may
be confusing to find that a cataract grade of P1 is
much more visually bothersome than a cataract grade
of NC3 if the patient is unfamiliar with the nuances of
cataract disease. The study raises an important con-
sideration of whether or not vision-specific function-
ing should be incorporated back into the assessment
of cataracts if the grading scheme is not only meant to
incorporate physician findings but also patient per-
spective and understanding.
CORRELATION BETWEEN GRADE,
SURGICAL APPROACH, AND OUTCOME
Not only is it important to consider the patient’s
experience of their cataracts and how it relates to
cataract grade, it is also important to consider how
the surgical approach and outcome are determined
by cataract grading. Will surgical approach be dif-
ferent if a cataract is given a grade of NO3 and NC3
versus NO4 and NC4? A study by Davidson and
Chylack investigating the use of LOCS III and pha-
coemulsification performance in 2364 cases found
that nuclear cataract phacoemulsification time cor-
related well with LOCS III but that cortical and
posterior subcapsular phacoemulsification had little
&&
relation to LOCS III classes [10 ]. Of note, an expo-
nential increase in phacoemulsification energy was
used intraoperatively as nuclear grades increased,
proving LOCS III to be a useful tool in creating an
&&
operative plan for nuclear cataract procedures [10 ].
Cataract scores can play a useful role in determining
case complexity during cataract surgery and post-
operative outcomes. This preoperative knowledge
can better allow clinicians to allocate certain cases
to surgeons in training and to provide realistic risks
and potential outcomes to patients.
NEW WAYS TO ASSESS AND GRADE
CATARACTS
Since the development of the LOCS III system, there
have been a variety of ways presented to better
evaluate and grade cataracts. Shortly after the LOCS
16 www.co-ophthalmology.com
Copyright © 2018 Wolters Kluwer Health, Inc. All rights reserved.
III system was introduced, Hall et al. set out to
provide a new way to evaluate nuclear cataracts
using a laser slit lamp in an attempt to standardize
an objective way to visualize cataract complexities.
The team created a device using a laser light slit,
viewing arm with a beam splitter, and a charge
coupled device camera to illuminate the anterior
segment with laser light [11]. The images obtained
with the laser were analyzed with a computer pro-
gram that calculated mean pixel intensity of the area
of interest in the nucleus as compared with a darker
background portion of the anterior segment in the
same image to measure the amount of light back-
scattered from the nucleus [11]. This method proved
to correlate well with the LOCS III grading scheme,
demonstrating a linear relationship between
nuclear opalescence LOCS III scores and the pixel
intensities imaged by the laser slit lamp and ana-
lyzed with computer software [11]. This highlighted
the use of a computer program to assist with
cataract grading.
Following the work of Hall, Babizhayev et al.
[12] set out to use computer generated analysis of
lens imaging to measure cataract severity in combi-
nation with a glare disability test. An automated
computer program was used to measure the inten-
sity of individual image pixels, and using serial 2D
images, 3D topography images of the lens were
created to provide a better understanding of the
characteristics of the lens [12]. The computer’s
analysis of the lens was compared against a clinical
slit lamp examination using the same grading
scheme laid out by Taylor and West. This analysis
technique demonstrated correlation between the
existing LOCS III scoring system and a new way
to assess cataracts [12].
Additional teams of researchers further refined
automatic systems for classifying cataracts using slit
lamp photography. Fan et al. [13] created a method
to grade nuclear sclerotic cataracts based on land-
marks in the visual axis from slit lamp photographs.
Another group from Johns Hopkins also used photo-
graphs from the visual axis and analyzed a variety of
features of the lens [14]. In these methods, a neural
network was employed to analyze the images and
produce a grade of nuclear cataract severity based on
the lens landmarks noted in the visual axis [14]. At
the time, these models focused only on the visual
axis, whereas a clinical exam evaluated the entirety
of the lens.
Li et al. [15] expanded on previous groups’ work
to create an automatic diagnosis system for nuclear
cataracts by training the system to locate the entire
lens and to extract features such as intensity, color,
and entropy inside the lens and nucleus. They based
the lens structure detection ability of their program
Volume 30  Number 1  January 2019

on their prior work with the Active Shape Model
method that they described in previous articles, and
they built upon it to better identify the lens nucleus
in addition to the lens itself [16,17]. The lens and
feature detection system was tested on over 5800
images with a 96.8% lens location detection rate and
a 95% lens structure detection rate [15]. When
compared against clinical diagnoses of the same
images, the automatic diagnosis system only varied
by greater than 1 clinical grade in only 3.37% of
images reviewed, proving its agreement with clinical
grading schemes and clinical diagnostic methods
[15]. Xu et al. [18] set out to create an automatic
grading system that refined the work of Li by build-
ing on the success of the lens structure detection and
identifying the most effective elements from the
image. This bag-of-features (BOF) method allows
for location-independent representation of features
that can be assessed for qualities such as intensity
and scale. Combining the BOF method with regres-
sion model training, the system achieved a 69.0%
exact agreement ratio when compared with clinical
grading and a 98.9% decimal grading error less than
1.0 [18]. These results demonstrated that clinical
diagnosis could be facilitated with the use of auto-
matic diagnosis systems.
Improving upon these automatic diagnosis sys-
tems, Srivastava et al. [19] developed what they call
the Automatic Cataract Screening from Image Anal-
ysis-Nuclear Cataract, Version 0.10, ACASIA-
NC_v0.10. The system builds upon the work done
by Li et al. in that it uses image gradients on edges
produced by landmarks in the lens. At higher grades
of nuclear cataract, these landmarks tend to be less
visible as the landmarks become less distinct. The
ACASIA-NC_v0.10 instead focuses on the gradient
between areas of the lens as a means to measure the
difference in grades of cataract and to overcome the
loss of visible landmarks in severe cataracts [19].
Although the ACASIA-NC_v0.10 did not prove to
be useful in grading very low or very high grade
nuclear cataracts, it improved upon prior methods
that used color and intensity and demonstrated that
gradient information is useful in automatic cataract
grading programs [19].
BEYOND SLIT LAMP PHOTOGRAPHY:
NEW TECHNIQUES IN CATARACT
ASSESSMENT AND GRADING
Although slit lamp examination and slit lamp pho-
tographs have been the mainstay of cataract grading
systems, different ways to assess for cataracts have
been proposed. These methods may vary in the
exactness of which they can assess cataracts as com-
pared with standard slit lamp photography analysis,
1040-8738 Copyright ß 2018 Wolters Kluwer Health, Inc. All rights reserved.
Copyright © 2018 Wolters Kluwer Health, Inc. All rights reserved.
Cataract grading systems Gali et al.
but they may provide alternatives to cataract grad-
ing when resources or time are limited. Using imag-
ing modalities in addition to slit lamp photographs,
&&
Xiong et al. [20 ] suggest that retinal images can be
used to screen for cataracts. They employed the scale
used by Wang et al. [21] to grade blurriness in retinal
images. Previous work on the topic was performed
with Fourier analysis by Abdul-Rahman et al. [22].
Their automated method of quantifying optical
degradation in retinal images proved to be useful
in detecting the presence of cataract and correlated
well with LOCS III [22]. However, automatic grading
of the cataract was not performed. Additionally,
vitreous opacity could not be evaluated separately
from lens opacities, potentially influencing grading
[22]. With these limitations in mind, Xiong et al.
developed a successful method to detect vitreous
opacity and separate it from the retinal structure
detection so as to improve cataract grading accu-
racy. This system graded cataracts with an 81.1%
accuracy and kappa value of 0.7435 when compared
&&
with clinical grading [20 ]. As technologies in tele-
medicine and artificial intelligence advance, using
retinal images to grade cataract may prove an effi-
cient and useful tool to utilize fundus photos that
may already be taken for diabetic retinopathy or age-
related macular degeneration monitoring.
In addition to fundus photography being used
as a novel method to grade cataracts, optical coher-
ence tomography (OCT) has also been suggested.
Anterior segment OCT has been used to assess a
variety of features such as corneal thickness, but
Wong et al. thought to compare the anterior seg-
ment OCT nucleus density measurement with LOCS
III grading of nuclear opalescence and nuclear color.
They found that there was a significant correlation
between the two measurements and that nuclear
opalescence scores had a slightly higher association
with anterior segment OCT than did nuclear color
[23]. Their use of anterior segment OCT to grade lens
density proved to be an objective, reliable, and fast
assessment using a frequently used clinical tool that
requires less training than mastering the LOCS III.
THE ROLE OF DEEP LEARNING IN
CATARACT GRADING
A discussion of cataract grading would be incom-
plete without touching on recent advancements in
the use of deep learning with machine learning and
artificial intelligence to assess lens opacity.
Although the existing automatic methods to grade
cataracts discussed previously rely on predefined
landmarks and features to recognize structures
and grade cataracts, systems can be trained to learn
grading features, filter them, and feed them into a
www.co-ophthalmology.com 17
Cataract surgery and lens implantation
neural network to analyze them further. Gao et al.
&&
[24 ] used this approach with a 5378 slit lamp image
data set and obtained a 70.7% exact integral agree-
ment ratio and a 99.0% decimal grading error less
than 1.0 when compared with clinical integral grad-
ing. They demonstrated that deep learning can be an
effective and powerful tool to grade cataracts.
CONCLUSION
The shift towards using computer-aided analysis of
cataract imaging to better standardize cataract grad-
ing is an exciting and rapidly developing field. The
role for automatic grading of cataracts is growing as
new technologies are developed to image the eye
and to better view structures that were previously
hard to capture on imaging. With the grading sys-
tems of past in mind, researchers have been able to
build on the classical ways that cataracts have been
assessed through slit lamp examination and use the
features of the lens that are typically assessed as a
platform for automated systems to grade cataracts in
an objective and standardized way.
Acknowledgements
None.
Financial support and sponsorship
None.
Conflicts of interest
There are no conflicts of interest.
REFERENCES AND RECOMMENDED
READING
Papers of particular interest, published within the annual period of review, have
been highlighted as:
& of special interest
&& of outstanding interest
1. West SK, Taylor HR. The detection and grading of cataract: an epidemiologic
perspective. Surv Ophthalmol 1986; 31:175–184.
2. Chylack LT, Wolfe JK, Singer DM, et al. The lens opacities classification
&& system III. Arch Ophthalmol 1993; 111:831–836.
The LOCS III system defined in this publication is a fundamental clinical diagnostic
tool and is used frequently as the basis of comparison for developing technologies
that diagnose and grade cataracts.
3. Sparrow JM, Bron AJ, Brown NA, et al. The Oxford Clinical Cataract Classi-
fication and Grading System. Int Ophthalmol 1986; 11:151–157.
18 www.co-ophthalmology.com
Copyright © 2018 Wolters Kluwer Health, Inc. All rights reserved.
4. Mehra V, Minassian DC. A rapid method of grading cataract in epidemiolo-
gical studies and eye surveys. Br J Ophthalmol 1988; 72:801–803.
5. Taylor HR, West SK. The clinical grading of lens opacities. Aust N Z J
Ophthalmol 1989; 17:81–86.
6. Klein BEK, Klein R, Linton KLP, et al. Assessment of cataracts from photo-
graphs in the Beaver Dam Eye Study. Ophthalmology 1990; 97:1428–1433.
7. Chylack LT, Leske MC, Sperduto R, et al. Lens opacities classification system.
Arch Ophthalmol 1988; 106:330–334.
8. Chylack LT, Khu P, Kashiwagi T, et al. Lens opacities classification system II
(LOCS II). Arch Ophthalmol 1989.
9. Chew M, Chiang PPC, Zheng Y, et al. The impact of cataract, cataract types,
& and cataract grades on vision-specific functioning using rasch analysis. Am J
Ophthalmol 2012; 154:29.e2–38.e2.
In this study, the authors show that vision-specific functioning decreased in a
statistically significant way for different cataract types at different grades using the
LOCS III, demonstrating that each cataract type produced symptoms at different
grades.
10. Davison JA, Chylack LT. Clinical application of the Lens Opacities Classifica-
tion System III in the performance of phacoemulsification. J Cataract Refract
&&
Surg 2003; 29:138–145.
This study demonstrated a positive exponential correlation between phacoemul-
sification energy requirements during surgery and nuclear cataract grade using the
LOCS III but did not show a correlation for posterior subcapsular or cortical
cataracts.
11. Hall NF, Lempert P, Shier RP, et al. Grading nuclear cataract: Reproducibility
and validity of a new method. Br J Ophthalmol 1999; 83:1159–1163.
12. Babizhayev MA, Deyev AI, Yermakova VN, et al. Image analysis and glare
sensitivity in human age-related cataracts. Clin Exp Optom 2003;
86:157–172.
13. Fan S, Dyer CR, Hubbard L, Klein B. An automatic system for classification of
nuclear sclerosis from slit-lamp photographs. In: Ellis RE, Peters TM, editors.
Medical image computing and computer-assisted intervention - MICCAI
2003. Berlin, Heidelberg: Springer Berlin Heidelberg; 2003. pp. 592–601.
14. Duncan DD, Shukla OB, West SK, Schein OD. New objective classification
system for nuclear opacification. J Opt Soc Am A Opt Image Sci Vis 1997;
14:1197–1204.
15. Li H, Lim JH, Liu J, et al. An automatic diagnosis system of nuclear cataract using
slit-lamp images. Conf Proc IEEE Eng Med Biol Soc 2009; 2009:3693–3696.
16. Li H, Lim JH, Liu J, Wong TY. Towards automatic grading of nuclear cataract.
In: Annual International Conference of the IEEE Engineering in Medicine and
Biology – Proceedings. 2007.
17. Li H, Lim JH, Liu J, et al. Image based grading of nuclear cataract by SVM
regression. Proc SPIE 6915, Medical Imaging 2008: Computer-Aided Diag-
nosis, 691536 (17 March 2008).
18. Xu Y, Gao X, Lin S. Automatic grading of nuclear cataracts from slit-lamp lens
images using group sparsity regression. In: Lecture Notes in Computer
Science (including subseries Lecture Notes in Artificial Intelligence and
Lecture Notes in Bioinformatics). 2013.
19. Srivastava R, Gao X, Yin F, et al. Automatic nuclear cataract grading using
image gradients. J Med Imaging (Bellingham) 2014; 1:014502.
20. Xiong L, Li H, Xu L. An approach to evaluate blurriness in retinal images with
&& vitreous opacity for cataract diagnosis. J Healthc Eng 2017; 2017:5645498.
This study goes beyond slit lamp photography and proposes a method of detecting
and grading cataracts with high success using retinal images that may already
need to be obtained for other clinical purposes, perhaps improving clinical
efficiency.
21. Yang Chuan-wu, Xu Liang, Wang Shuang, et al. The evaluation of screening
for cataract needed surgery with digital nonmydriatic fundus camera. Ophthal-
mol China 2010; 19:46–49.
22. Abdul-Rahman AM, Molteno T, Molteno ACB. Fourier analysis of digital retinal
images in estimation of cataract severity. Clin Exp Ophthalmol 2008;
36:637–645.
23. Wong AL, Leung CK-S, Weinreb RN, et al. Quantitative assessment of lens
opacities with anterior segment optical coherence tomography. Br J Ophthal-
mol [Internet] 2009; 93:61–65.
24. Gao SL X, TYW. Automatic feature learning to grade nuclear cataracts based
&& on deep learning. IEEE Trans Biomed Eng 2015; 62:2693–2701.
This study highlights the use of deep learning as a powerful and highly reproducible
tool to grade cataracts.
Volume 30  Number 1  January 2019