IIIT

Hyd

erab

ad

A Hierarchical System Design for detection of Glaucoma from Color Fundus Images

Madhulika Jain, Jayanthi Sivaswamy

CVIT, IIIT Hyderabad, India

IIIT

Hyd

erab

ad

What is glaucoma?

Glaucoma is an eye disorder that causes irreversible loss of vision

It affects the Optic Nerve in retina

IIIT

Hyd

erab

ad

Why glaucoma detection needed?

– Leading cause of blindness worldwide • In India – 2nd leading cause of blindness• It is estimated to affect 79 million people in the world by the year 2020

– Prevalent in aging population India alone hosts 20% of glaucoma cases

– Irreversible loss of vision• Thus timely detection required

– lack of manpower in terms of skilled technicians• Thus computer aided solutions for screening

IIIT

Hyd

erab

ad

Color Fundus Images

• It provides 2-D projection of retina – structural information of optic disc(OD)– Information regarding retinal structures such as cup, rim and blood vessels

• In a retinal image, the region of interest is the Optic Disk (OD)

• Disk – is marked by the outer boundary of OD (white)

• Cup – is marked by the inner boundary of OD (black)

• Glaucoma manifests primarily as structural deformations in OD

IIIT

Hyd

erab

ad

Why color fundus images are used?

low costnon-invasiveness ease of use

higher penetrability to every section of society

IIIT

Hyd

erab

ad

Visual Symptoms of GlaucomaG

loba

lG

loba

l

• Rim Thinning – is caused by enlargement of cup with respect to optic disk (arrow)

• Peripapillary Atrophy (PPA) – is atrophy of retinal cells around optic disk (yellow)– a change in intensity is observed

adjoining the disk boundary

• Retinal Nerve Fiber Layer (RNFL) Defect – occurs due to the loss of the respective layer in retina (green)– most subtle indicator of glaucoma

RIM THINNING

PERIPAPILLARY ATROPHY

RNFL DEFECT

LocalLocal

IIIT

Hyd

erab

ad

More Examples of Atrophy

More Examples of rim thinning

IIIT

Hyd

erab

ad

Glaucoma Detection – BackgroundLocal ApproachesLocal Approaches

Aim at measuring the cup to disc ratio after segmenting the cup and disk regions

Joshi et al. (2011) Chan Vese model (CV model) with no shape constraints to segment disk R-bends (relevant bends) and pallor information for cup segmentation

Liu et al. (2009) level set method followed by ellipse fitting for disk segmentation level set based cup region segmentation followed by ellipse fitting

Tao et al. (2013)Superpixel based classification for cup segmentation

+ Morphological changes (rim thinning) are captured well, provided segmentation is accurate− Accurate identification of these ill-defined boundaries is a difficult task

Global ApproachesGlobal Approaches

+ Need to accurately identify boundaries is eliminated− Difficult to achieve robustness to significant intra-class variations

Aim at deriving global image features

Bock et al. (2007) compare and select from pixel intensity values texture using Gabor filters spectral features - FFT coefficients histogram model

Bock et al. (2010) pixel intensity values , FFT and B-spline coefficients to derive probabilistic output

Meier et al. (2010) uses same features as [Bock 2007] additional pre-processing to remove disease independent variations

Annu et al. (2013)Uses texture features

IIIT

Hyd

erab

ad

Glaucoma Detection – BackgroundLocal ApproachesLocal Approaches

Aim at measuring the cup to disc ratio after segmenting the cup and disk regions

Joshi et al. (2011) Chan Vese model (CV model) with no shape constraints to segment disk R-bends (relevant bends) and pallor information for cup segmentation

Liu et al. (2009) level set method followed by ellipse fitting for disk segmentation level set based cup region segmentation followed by ellipse fitting

Tao et al. (2013)Superpixel based classification for cup segmentation

+ Morphological changes (rim thinning) are captured well, provided segmentation is accurate− Accurate identification of these ill-defined boundaries is a difficult task

Global ApproachesGlobal Approaches

+ Need to accurately identify boundaries is eliminated− Difficult to achieve robustness to significant intra-class variations

Aim at deriving global image features

Bock et al. (2007) compare and select from pixel intensity values texture using Gabor filters spectral features - FFT coefficients histogram model

Bock et al. (2010) pixel intensity values , FFT and B-spline coefficients to derive probabilistic output

Meier et al. (2010) uses same features as [Bock 2007] additional pre-processing to remove disease independent variations

Annu et al. (2013)Uses texture features

IIIT

Hyd

erab

ad

Proposals

Exploring global motion pattern based features for glaucoma

detection

Hierarchical system design using global

analysis only

Hierarchical system design using both global and local

analysis

Extending this global approach to the

detection of atrophy

IIIT

Hyd

erab

ad

Review of the GMP representation

Original Image On rotation

:follows as generated is GMP a ,interest ofregion aGiven GMPIrI

rITS=IGMP

denotes a location in the 2-D ROI is a rigid transformation applied to generate a sequence of transformed images is the coalescing function

rTS

Result

Earlier used for detecting bright lesions present in diabetic macular edema by Deepak et al. [6] [7]

IIIT

Hyd

erab

ad

GMP for Glaucoma Detection

ChallengesChallenges • capturing subtle structural deformation• relative contrast of cup and disc not consistent

StrategyStrategy Coalescing function Pixelwise maximum

Exploring different motion variable Pivot of motion Type of motion Extent of motion Step size of motion

Translation Translation

RotationRotation

Rotation + TranslationRotation + Translation

IIIT

Hyd

erab

ad

Translation motion based GMP

Aim :- To capture deformation in superior-inferior and nasal-temporal direction (primary interest)

IIIT

Hyd

erab

ad

Translation motion based GMP

IIIT

Hyd

erab

ad

Translation motion based GMP

IIIT

Hyd

erab

ad

Translation motion based GMPProjection profile ; : projection of GMP in direction rIP=f GMPp ( P

Variation in profile ;1)( ifif=if ppd 1,....,0 Mi

Shape of profile ];[)( kckcc

pp

tfsf=kf

tfsf

Lkt=tf

Lks=sf

kkc

kkc

,....,1,0;)(

0,1,....,;)(

IIIT

Hyd

erab

ad

Rotation motion based GMP

Aim :- To accentuate the subtle deformation (local notching) and the presence of atrophy

IIIT

Hyd

erab

ad

Rotation motion based GMP

GMP is obtained using following equation

For every pivot one GMP image

For each GMP feature extracted – histogram after it is rebinned to 7 bins(figure in next slide)

;3

;&;30&15

));()((),(

oo

oo

ioGMP

evenoddi

prIkRrI

IIIT

Hyd

erab

ad

Fig :- Extracted feature fr for the normal(blue) and glaucomatous(red) case

IIIT

Hyd

erab

ad

Combination of translation and rotation based GMP

Aim :- To capture variation of neuroretinal rim thickness along optical disc in all directions

IIIT

Hyd

erab

ad

Combination of translation and rotation based GMP

;3

;180,...,0:

;0

oo

o

o

Range

IIIT

Hyd

erab

ad

Classification

IIIT

Hyd

erab

ad

Experiments and ResultsDataset details • Total test set -1845 images :1272 normal & 573

glaucomatous cases • Training images – 800 images : 400 normal & 400

glaucomatous cases• Groundtruth : opinion of 3 glaucoma experts as normal,

suspect and confirm case (gold standard – majority)• Size of original image = 1494 * 1996• ROI = 401*401 (all processing with a circular mask of

radius 200 pixels)

Evaluation Scheme • using ROC curve-- area under ROC curve

IIIT

Hyd

erab

ad

Results For analyzing contribution of each features, variants are generated combining features one by one.

Table of results for each variant :

ROC plot for stage -1

IIIT

Hyd

erab

ad

Proposals

Exploring global motion pattern based features for glaucoma

detection

Hierarchical system design using global

analysis only

Hierarchical system design using both global and local

analysis

Extending this global approach to the

detection of atrophy

IIIT

Hyd

erab

ad

Atrophy detection - Background

• Kolar et al. : Using active contours in Heidelberg retinal angiography images

• Joshi et al. (2011) : Using dissimilarity measure of regions adjoining disc

• Cheng et al. : Using biologically inspired feature• Muramatsu et al. : Using texture analysis (only on 26 images)

Previous work largely based on local feature based detection

Our strategy :- GMP based global features approach

IIIT

Hyd

erab

ad

GMP based features for atrophy

Aim :- To capture perceived intensity difference in the regions adjoining optic disc.

• Divide images into 18 patches of 20o each• For each patch, GMP is computed as follows:

• Feature computed on annular sector of width 40px

;3;15;,.....

));()(()(o

oo

oGMP

k

prIkRrI

IIIT

Hyd

erab

ad

Dataset Details

Results

Training images = 55 990 patchesTesting images = 59 1062 patches

Classifier Used Ensemble of decision trees

Sensitivity Specificity Accuracy

Joshi et al. (2011)

0.82 0.72 0.78

Proposed 0.812 0.57 0.687

Performance Figures

IIIT

Hyd

erab

ad

Proposals

Exploring global motion pattern based features for glaucoma

detection

Hierarchical system design using global

analysis only

Hierarchical system design using both global and local

analysis

Extending this global approach to the

detection of atrophy

IIIT

Hyd

erab

ad

Hierarchical system

Aim :-To analyze if cascaded system helps in improving detection performance using global features in both stages

Proposed PPA System (threshold at previous stage taken so as to have lesser false negatives)

IIIT

Hyd

erab

ad

ResultsDataset details Stage 1 – same as earlier

Stage 2 – Training images = 114 images Test images = left from 1st stage (1040)

Evaluation Scheme • Sensitivity

• Specificity• Accuracy

Sensitivity Specificity Accuracy

Proposed PPA 0.812 0.57 0.687

Cascade 0.817 0.65 0.73

Performance Figures

help us to remove 10% more images

IIIT

Hyd

erab

ad

Proposals

Exploring global motion pattern based features for glaucoma

detection

Hierarchical system design using global

analysis only

Hierarchical system design using both global and local

analysis

Extending this global approach to the

detection of atrophy

IIIT

Hyd

erab

ad

Hierarchical System

• This uses Stage 1 same as used in previous system (global feature based)– Aim – to remove as many as true negatives– Enables tuning second stage for higher specificity without

compromising sensitivity• Stage 2 is borrowed from a work within our group (local

feature based) – A clinical paper of this work is under review– Explained next

• So far no attempts to assess glaucoma using capabilities of both

IIIT

Hyd

erab

ad

Stage - 2• Local features used in this stage are :-

– Cup-to-disk vertical diameter ratio (CDR)– Cup-to-disk area ratio (CAR)– Atrophy presence decision in the inferior and superior

directions– RNFL presence decision in the inferior and superior

directions – Relative distributions of image structures using structural

clustering

IIIT

Hyd

erab

ad

Final System

IIIT

Hyd

erab

ad

Experiments & Results• Two experiments are done

– Using same training set of 800 images (as used earlier) in both the stages (Expt 1)

– Using the left over images from stage 1 to train the stage 2 (Expt 2)• 1040 (from stage 1)

• Dataset details – In Expt 1 – testing set similar as used before– In Expt 2 – 800 new images (added to 1040) 652 normal and 148

glaucomatous

• Evaluation scheme– AUC– Sensitivity– Specificity

IIIT

Hyd

erab

ad

Result of Expt 1• Comparison of Stage 2 and combined system (below)

Sensitivity Specificity AUC

Tao et al. 0.65 0.44 0.558

Bock et al. (2010) 0.70 0.46 0.615

Stage-2 CDR 0.70 0.48 0.636

Stage-2 multifactorial 0.79 0.62 0.8012

Hierarchical System 0.8128 0.7658 0.8353

• Comparison with state of the art methods (table)

IIIT

Hyd

erab

ad

Result of Expt 2

• The dataset details (given earlier)• Comparison of whole system with and without retraining the

second system using leftover images (table below)

Sensitivity Specificity AUC

Hierarchical System

0.801 0.734 0.8123

Hierarchical System (retrained)

0.8539 0.7935 0.844

IIIT

Hyd

erab

ad

Conclusion A global feature based approach for glaucoma detection from

retinal images was proposed

The Generalized Moment Pattern representation is extended for detecting structural distortions in Optic Disk

Hierarchical design (using both global and local analysis) posited to be best to avoid a trade off between sensitivity and specificity

Evaluation of glaucoma detection is performed on a large retinal image dataset

IIIT

Hyd

erab

ad

Note

• Computing cost may be an issue but alternative ways can be explored for that (especially for Stage-2)

• Given the variabilities that occur in screening scenarios, the parameters may need to be retrained for desired results

IIIT

Hyd

erab

ad

Possible Future Directions

• Exploring more efficient global feature for RNFL detection.– Performance presented in literature not great– Due to subtleness– Adding this as new stage or in the stage-1 of the proposed system

will boost performance• Automatic detection of parameters of GMP

– will be a major step forward in generalization of detection system using GMP

– Can be trained automatically as and when new data arrives• Optimizing the computational cost.

IIIT

Hyd

erab

ad

References[1] G. D. Joshi, J. Sivaswamy, and S. R. Krishnadas. Optic disk and cup segmentation from monocular colour

retinal images for glaucoma assessment. IEEE Trans on Medical Imaging, 30(6):1192-1205,2011.[2] J. Liu, D. Wong, J. Lim, H. Li, N. Tan, and T. Wong. Argali- an automatic cup-to-disc ratio measurement

system for glaucoma detection and analysis framework. In Proc. SPIE, Medical Imaging, pages 72 603k-8, 2009.

[3] R. Bock, J. Meier, G. Michelson, L. Nyul, and J. Hornegger. Classifying glaucoma with image-based features from fundus photographs. Proc. DAGM, pages 355-364, 2007.

[4] R. Bock, J. Meier, L. Nyul, and G. Michelson. Glaucoma risk index: automated glaucoma detection from color fundus images. Medical Image Analysis, 14(3):471-481, 2010.

[5] J. Meier, R. Bock, G. Michelson, L. Nyul, and J. Hornegger. Effects of preprocessing eye fundus images on appearance based glaucoma classification. Proc. CAIP, pages 165-172, 2007.

[6] K. S. Deepak, N. K. Medathati, and J. Sivaswamy. Detection and discrimination of disease-related abnormalities based on learning normal cases. Pattern Recogn., 45(10):3707-3716, Oct. 2012.

[7] K. S. Deepak and J. Sivaswamy. Automatic assessment of macular edema from color retinal images. Medical Imaging, IEEE Trans on, 31(3):766 -776, march 2012.

[8] D. Tao, F. Yin, D. Kee, Y. Xu, T. Yin, J. Cheng, and J. Liu. Superpixel classification based optic cup segmentation. Medical Image Computing and Computer-assisted Intervention (MICCAI), 8151:421–428, 2013.

[9] N. Annu and J. Justin. Automated classification of glaucoma images by wavelet energy features. International Journal of Engineering and Technology, 2013.

[10] R. Kolar, J. Jan, R. Laemmer, and R. Jirik. Semiautomatic detection and evaluation of autofluorescent areas in retinal images. Proc. EMBS, pages 3327–3330, 2007

IIIT

Hyd

erab

ad

References[11] G. D. Joshi, J. Sivaswamy, R. Prashanth, and S. R. Krishnadas. Detection of peri-papillary atrophy and rnfl defect from retinal images. ICIAR, 2011.[12] C. Muramatsu, Y. Hatanaka, A. Sawada, T. Yamamoto, and H. Fujita. Computerized detection of peripapillary chorioretinal atrophy by texture analysis. 33rd Annual International Conference of the IEEE EMBS Boston, Massachusetts USA, 2011.[13] J. Cheng*, D. Tao, J. Liu, D. W. K. Wong, N.-M. Tan, T. Y. Wong, and S. M. Saw. Peripapillary atrophy detection by sparse biologically inspired feature manifold. IEEE TRANSACTIONS ON MEDICAL IMAGING, 2012.

IIIT

Hyd

erab

ad

We gratefully acknowledge

Aravind Eye Hospital, Madurai (for data and expert diagnosis)

IIIT

Hyd

erab

ad

THANK YOU

IIIT

Hyd

erab

ad

IIIT

Hyd

erab

ad

IIIT

Hyd

erab

ad

IIIT

Hyd

erab

ad

IIIT

Hyd

erab

ad