Exudate-based diabetic macular edema detection in fundus images using publicly available datasets

Luca Giancardo, Fabrice Meriaudeau, Thomas P. Karnowski, Yaqin Li, Seema Garg, Kenneth W. Tobin, Edward Chaum

Research output: Contribution to journalArticle

149 Citations (Scopus)

Abstract

Diabetic macular edema (DME) is a common vision threatening complication of diabetic retinopathy. In a large scale screening environment DME can be assessed by detecting exudates (a type of bright lesions) in fundus images. In this work, we introduce a new methodology for diagnosis of DME using a novel set of features based on colour, wavelet decomposition and automatic lesion segmentation. These features are employed to train a classifier able to automatically diagnose DME through the presence of exudation. We present a new publicly available dataset with ground-truth data containing 169 patients from various ethnic groups and levels of DME. This and other two publicly available datasets are employed to evaluate our algorithm. We are able to achieve diagnosis performance comparable to retina experts on the MESSIDOR (an independently labelled dataset with 1200 images) with cross-dataset testing (e.g., the classifier was trained on an independent dataset and tested on MESSIDOR). Our algorithm obtained an AUC between 0.88 and 0.94 depending on the dataset/features used. Additionally, it does not need ground truth at lesion level to reject false positives and is computationally efficient, as it generates a diagnosis on an average of 4.4. s (9.3. s, considering the optic nerve localisation) per image on an 2.6. GHz platform with an unoptimised Matlab implementation.

Original languageEnglish (US)
Pages (from-to)216-226
Number of pages11
JournalMedical Image Analysis
Volume16
Issue number1
DOIs
StatePublished - Jan 1 2012

Fingerprint

Macular Edema
Exudates and Transudates
Classifiers
Wavelet decomposition
Optics
Screening
Diabetic Retinopathy
Optic Nerve
Color
Ethnic Groups
Area Under Curve
Retina
Datasets
Testing

All Science Journal Classification (ASJC) codes

  • Radiological and Ultrasound Technology
  • Radiology Nuclear Medicine and imaging
  • Computer Vision and Pattern Recognition
  • Health Informatics
  • Computer Graphics and Computer-Aided Design

Cite this

Giancardo, L., Meriaudeau, F., Karnowski, T. P., Li, Y., Garg, S., Tobin, K. W., & Chaum, E. (2012). Exudate-based diabetic macular edema detection in fundus images using publicly available datasets. Medical Image Analysis, 16(1), 216-226. https://doi.org/10.1016/j.media.2011.07.004

Exudate-based diabetic macular edema detection in fundus images using publicly available datasets. / Giancardo, Luca; Meriaudeau, Fabrice; Karnowski, Thomas P.; Li, Yaqin; Garg, Seema; Tobin, Kenneth W.; Chaum, Edward.

In: Medical Image Analysis, Vol. 16, No. 1, 01.01.2012, p. 216-226.

Research output: Contribution to journalArticle

Giancardo, L, Meriaudeau, F, Karnowski, TP, Li, Y, Garg, S, Tobin, KW & Chaum, E 2012, 'Exudate-based diabetic macular edema detection in fundus images using publicly available datasets', Medical Image Analysis, vol. 16, no. 1, pp. 216-226. https://doi.org/10.1016/j.media.2011.07.004
Giancardo, Luca ; Meriaudeau, Fabrice ; Karnowski, Thomas P. ; Li, Yaqin ; Garg, Seema ; Tobin, Kenneth W. ; Chaum, Edward. / Exudate-based diabetic macular edema detection in fundus images using publicly available datasets. In: Medical Image Analysis. 2012 ; Vol. 16, No. 1. pp. 216-226.
@article{9bd1261c5f2a48189fca842be497b194,
title = "Exudate-based diabetic macular edema detection in fundus images using publicly available datasets",
abstract = "Diabetic macular edema (DME) is a common vision threatening complication of diabetic retinopathy. In a large scale screening environment DME can be assessed by detecting exudates (a type of bright lesions) in fundus images. In this work, we introduce a new methodology for diagnosis of DME using a novel set of features based on colour, wavelet decomposition and automatic lesion segmentation. These features are employed to train a classifier able to automatically diagnose DME through the presence of exudation. We present a new publicly available dataset with ground-truth data containing 169 patients from various ethnic groups and levels of DME. This and other two publicly available datasets are employed to evaluate our algorithm. We are able to achieve diagnosis performance comparable to retina experts on the MESSIDOR (an independently labelled dataset with 1200 images) with cross-dataset testing (e.g., the classifier was trained on an independent dataset and tested on MESSIDOR). Our algorithm obtained an AUC between 0.88 and 0.94 depending on the dataset/features used. Additionally, it does not need ground truth at lesion level to reject false positives and is computationally efficient, as it generates a diagnosis on an average of 4.4. s (9.3. s, considering the optic nerve localisation) per image on an 2.6. GHz platform with an unoptimised Matlab implementation.",
author = "Luca Giancardo and Fabrice Meriaudeau and Karnowski, {Thomas P.} and Yaqin Li and Seema Garg and Tobin, {Kenneth W.} and Edward Chaum",
year = "2012",
month = "1",
day = "1",
doi = "10.1016/j.media.2011.07.004",
language = "English (US)",
volume = "16",
pages = "216--226",
journal = "Medical Image Analysis",
issn = "1361-8415",
publisher = "Elsevier",
number = "1",

}

TY - JOUR

T1 - Exudate-based diabetic macular edema detection in fundus images using publicly available datasets

AU - Giancardo, Luca

AU - Meriaudeau, Fabrice

AU - Karnowski, Thomas P.

AU - Li, Yaqin

AU - Garg, Seema

AU - Tobin, Kenneth W.

AU - Chaum, Edward

PY - 2012/1/1

Y1 - 2012/1/1

N2 - Diabetic macular edema (DME) is a common vision threatening complication of diabetic retinopathy. In a large scale screening environment DME can be assessed by detecting exudates (a type of bright lesions) in fundus images. In this work, we introduce a new methodology for diagnosis of DME using a novel set of features based on colour, wavelet decomposition and automatic lesion segmentation. These features are employed to train a classifier able to automatically diagnose DME through the presence of exudation. We present a new publicly available dataset with ground-truth data containing 169 patients from various ethnic groups and levels of DME. This and other two publicly available datasets are employed to evaluate our algorithm. We are able to achieve diagnosis performance comparable to retina experts on the MESSIDOR (an independently labelled dataset with 1200 images) with cross-dataset testing (e.g., the classifier was trained on an independent dataset and tested on MESSIDOR). Our algorithm obtained an AUC between 0.88 and 0.94 depending on the dataset/features used. Additionally, it does not need ground truth at lesion level to reject false positives and is computationally efficient, as it generates a diagnosis on an average of 4.4. s (9.3. s, considering the optic nerve localisation) per image on an 2.6. GHz platform with an unoptimised Matlab implementation.

AB - Diabetic macular edema (DME) is a common vision threatening complication of diabetic retinopathy. In a large scale screening environment DME can be assessed by detecting exudates (a type of bright lesions) in fundus images. In this work, we introduce a new methodology for diagnosis of DME using a novel set of features based on colour, wavelet decomposition and automatic lesion segmentation. These features are employed to train a classifier able to automatically diagnose DME through the presence of exudation. We present a new publicly available dataset with ground-truth data containing 169 patients from various ethnic groups and levels of DME. This and other two publicly available datasets are employed to evaluate our algorithm. We are able to achieve diagnosis performance comparable to retina experts on the MESSIDOR (an independently labelled dataset with 1200 images) with cross-dataset testing (e.g., the classifier was trained on an independent dataset and tested on MESSIDOR). Our algorithm obtained an AUC between 0.88 and 0.94 depending on the dataset/features used. Additionally, it does not need ground truth at lesion level to reject false positives and is computationally efficient, as it generates a diagnosis on an average of 4.4. s (9.3. s, considering the optic nerve localisation) per image on an 2.6. GHz platform with an unoptimised Matlab implementation.

UR - http://www.scopus.com/inward/record.url?scp=82355185836&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=82355185836&partnerID=8YFLogxK

U2 - 10.1016/j.media.2011.07.004

DO - 10.1016/j.media.2011.07.004

M3 - Article

VL - 16

SP - 216

EP - 226

JO - Medical Image Analysis

JF - Medical Image Analysis

SN - 1361-8415

IS - 1

ER -