EyeQual: Accurate, explainable, retinal image quality assessment

Pedro Costa; Aurelio Campilho; Bryan Hooi; Asim Smailagic; Kris Kitani; Shenghua Liu; Christos Faloutsos; Adrian Galdran

doi:10.1109/ICMLA.2017.0-140

EyeQual: Accurate, explainable, retinal image quality assessment

Pedro Costa
, Aurelio Campilho
, Bryan Hooi
, Asim Smailagic
, Kris Kitani
, Shenghua Liu
, Christos Faloutsos
, Adrian Galdran

INESC TEC
Carnegie Mellon University
University of Porto
School of Computer Science
Chinese Academy of Sciences

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

22 Citations (Scopus)

Abstract

Given a retinal image, can we automatically determine whether it is of high quality (suitable for medical diagnosis)? Can we also explain our decision, pinpointing the region or regions that led to our decision? Images from human retinas are vital for the diagnosis of multiple health issues, like hypertension, diabetes, and Alzheimer's; low quality images may force the patient to come back again for a second scanning, wasting time and possibly delaying treatment. However, existing retinal image quality assessment methods are either black boxes without explanations of the results or depend heavily on feature engineering or on complex and error-prone anatomical structures' segmentation. Therefore, we propose EyeQual, that solves exactly this problem. EyeQual is novel, fast for inference, accurate and explainable, pinpointing low-quality regions on the image. We evaluated EyeQual on two real datasets where it achieved 100% accuracy taking just 36 milliseconds for each image.

Original language	English
Title of host publication	Proceedings - 16th IEEE International Conference on Machine Learning and Applications, ICMLA 2017
Editors	Xuewen Chen, Bo Luo, Feng Luo, Vasile Palade, M. Arif Wani
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	323-330
Number of pages	8
ISBN (Electronic)	9781538614174
DOIs	https://doi.org/10.1109/ICMLA.2017.0-140
Publication status	Published - 2017
Externally published	Yes
Event	16th IEEE International Conference on Machine Learning and Applications, ICMLA 2017 - Cancun, Mexico Duration: 18 Dec 2017 → 21 Dec 2017

Publication series

Name	Proceedings - 16th IEEE International Conference on Machine Learning and Applications, ICMLA 2017
Volume	2017-December

Conference

Conference	16th IEEE International Conference on Machine Learning and Applications, ICMLA 2017
Country/Territory	Mexico
City	Cancun
Period	18/12/17 → 21/12/17

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

SDG 3 Good Health and Well-being

Keywords

Medical Image Analysis
Multiple Instance Learning
Quality Assessment
Retinal Image
Weakly Supervised Learning

Access to Document

10.1109/ICMLA.2017.0-140

Cite this

Costa, P., Campilho, A., Hooi, B., Smailagic, A., Kitani, K., Liu, S., Faloutsos, C., & Galdran, A. (2017). EyeQual: Accurate, explainable, retinal image quality assessment. In X. Chen, B. Luo, F. Luo, V. Palade, & M. A. Wani (Eds.), Proceedings - 16th IEEE International Conference on Machine Learning and Applications, ICMLA 2017 (pp. 323-330). (Proceedings - 16th IEEE International Conference on Machine Learning and Applications, ICMLA 2017; Vol. 2017-December). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ICMLA.2017.0-140

Costa, Pedro ; Campilho, Aurelio ; Hooi, Bryan et al. / EyeQual : Accurate, explainable, retinal image quality assessment. Proceedings - 16th IEEE International Conference on Machine Learning and Applications, ICMLA 2017. editor / Xuewen Chen ; Bo Luo ; Feng Luo ; Vasile Palade ; M. Arif Wani. Institute of Electrical and Electronics Engineers Inc., 2017. pp. 323-330 (Proceedings - 16th IEEE International Conference on Machine Learning and Applications, ICMLA 2017).

@inproceedings{52bf2aa165804cc29c3615ea31bad7f2,

title = "EyeQual: Accurate, explainable, retinal image quality assessment",

abstract = "Given a retinal image, can we automatically determine whether it is of high quality (suitable for medical diagnosis)? Can we also explain our decision, pinpointing the region or regions that led to our decision? Images from human retinas are vital for the diagnosis of multiple health issues, like hypertension, diabetes, and Alzheimer's; low quality images may force the patient to come back again for a second scanning, wasting time and possibly delaying treatment. However, existing retinal image quality assessment methods are either black boxes without explanations of the results or depend heavily on feature engineering or on complex and error-prone anatomical structures' segmentation. Therefore, we propose EyeQual, that solves exactly this problem. EyeQual is novel, fast for inference, accurate and explainable, pinpointing low-quality regions on the image. We evaluated EyeQual on two real datasets where it achieved 100\% accuracy taking just 36 milliseconds for each image.",

keywords = "Medical Image Analysis, Multiple Instance Learning, Quality Assessment, Retinal Image, Weakly Supervised Learning",

author = "Pedro Costa and Aurelio Campilho and Bryan Hooi and Asim Smailagic and Kris Kitani and Shenghua Liu and Christos Faloutsos and Adrian Galdran",

note = "Publisher Copyright: {\textcopyright} 2017 IEEE.; 16th IEEE International Conference on Machine Learning and Applications, ICMLA 2017 ; Conference date: 18-12-2017 Through 21-12-2017",

year = "2017",

doi = "10.1109/ICMLA.2017.0-140",

language = "English",

series = "Proceedings - 16th IEEE International Conference on Machine Learning and Applications, ICMLA 2017",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

pages = "323--330",

editor = "Xuewen Chen and Bo Luo and Feng Luo and Vasile Palade and Wani, \{M. Arif\}",

booktitle = "Proceedings - 16th IEEE International Conference on Machine Learning and Applications, ICMLA 2017",

address = "United States",

}

Costa, P, Campilho, A, Hooi, B, Smailagic, A, Kitani, K, Liu, S, Faloutsos, C & Galdran, A 2017, EyeQual: Accurate, explainable, retinal image quality assessment. in X Chen, B Luo, F Luo, V Palade & MA Wani (eds), Proceedings - 16th IEEE International Conference on Machine Learning and Applications, ICMLA 2017. Proceedings - 16th IEEE International Conference on Machine Learning and Applications, ICMLA 2017, vol. 2017-December, Institute of Electrical and Electronics Engineers Inc., pp. 323-330, 16th IEEE International Conference on Machine Learning and Applications, ICMLA 2017, Cancun, Mexico, 18/12/17. https://doi.org/10.1109/ICMLA.2017.0-140

EyeQual: Accurate, explainable, retinal image quality assessment. / Costa, Pedro; Campilho, Aurelio; Hooi, Bryan et al.
Proceedings - 16th IEEE International Conference on Machine Learning and Applications, ICMLA 2017. ed. / Xuewen Chen; Bo Luo; Feng Luo; Vasile Palade; M. Arif Wani. Institute of Electrical and Electronics Engineers Inc., 2017. p. 323-330 (Proceedings - 16th IEEE International Conference on Machine Learning and Applications, ICMLA 2017; Vol. 2017-December).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - EyeQual

T2 - 16th IEEE International Conference on Machine Learning and Applications, ICMLA 2017

AU - Costa, Pedro

AU - Campilho, Aurelio

AU - Hooi, Bryan

AU - Smailagic, Asim

AU - Kitani, Kris

AU - Liu, Shenghua

AU - Faloutsos, Christos

AU - Galdran, Adrian

PY - 2017

Y1 - 2017

N2 - Given a retinal image, can we automatically determine whether it is of high quality (suitable for medical diagnosis)? Can we also explain our decision, pinpointing the region or regions that led to our decision? Images from human retinas are vital for the diagnosis of multiple health issues, like hypertension, diabetes, and Alzheimer's; low quality images may force the patient to come back again for a second scanning, wasting time and possibly delaying treatment. However, existing retinal image quality assessment methods are either black boxes without explanations of the results or depend heavily on feature engineering or on complex and error-prone anatomical structures' segmentation. Therefore, we propose EyeQual, that solves exactly this problem. EyeQual is novel, fast for inference, accurate and explainable, pinpointing low-quality regions on the image. We evaluated EyeQual on two real datasets where it achieved 100% accuracy taking just 36 milliseconds for each image.

AB - Given a retinal image, can we automatically determine whether it is of high quality (suitable for medical diagnosis)? Can we also explain our decision, pinpointing the region or regions that led to our decision? Images from human retinas are vital for the diagnosis of multiple health issues, like hypertension, diabetes, and Alzheimer's; low quality images may force the patient to come back again for a second scanning, wasting time and possibly delaying treatment. However, existing retinal image quality assessment methods are either black boxes without explanations of the results or depend heavily on feature engineering or on complex and error-prone anatomical structures' segmentation. Therefore, we propose EyeQual, that solves exactly this problem. EyeQual is novel, fast for inference, accurate and explainable, pinpointing low-quality regions on the image. We evaluated EyeQual on two real datasets where it achieved 100% accuracy taking just 36 milliseconds for each image.

KW - Medical Image Analysis

KW - Multiple Instance Learning

KW - Quality Assessment

KW - Retinal Image

KW - Weakly Supervised Learning

UR - https://www.scopus.com/pages/publications/85048485145

U2 - 10.1109/ICMLA.2017.0-140

DO - 10.1109/ICMLA.2017.0-140

M3 - Conference contribution

AN - SCOPUS:85048485145

T3 - Proceedings - 16th IEEE International Conference on Machine Learning and Applications, ICMLA 2017

SP - 323

EP - 330

BT - Proceedings - 16th IEEE International Conference on Machine Learning and Applications, ICMLA 2017

A2 - Chen, Xuewen

A2 - Luo, Bo

A2 - Luo, Feng

A2 - Palade, Vasile

A2 - Wani, M. Arif

PB - Institute of Electrical and Electronics Engineers Inc.

Y2 - 18 December 2017 through 21 December 2017

ER -

Costa P, Campilho A, Hooi B, Smailagic A, Kitani K, Liu S et al. EyeQual: Accurate, explainable, retinal image quality assessment. In Chen X, Luo B, Luo F, Palade V, Wani MA, editors, Proceedings - 16th IEEE International Conference on Machine Learning and Applications, ICMLA 2017. Institute of Electrical and Electronics Engineers Inc. 2017. p. 323-330. (Proceedings - 16th IEEE International Conference on Machine Learning and Applications, ICMLA 2017). doi: 10.1109/ICMLA.2017.0-140

EyeQual: Accurate, explainable, retinal image quality assessment

Abstract

Publication series

Conference

UN SDGs

Keywords

Access to Document

Other files and links

Fingerprint

Cite this