Comparing pixel N-grams and bag of visual word features for the classification of diabetic retinopathy

Pradnya Kulkarni, Andrew Stranieri, Herbert Jelinek

Research output: Book chapter/Published conference paperConference paper

Abstract

The extraction of Bag of Visual Words (BoVW) features from retinal images for automated classification has been shown to be effective but computationally expensive. Histogram and co-variance matrix features do not generally result in models that have the same predictive accuracy as BoVW and are still computationally expensive. The discovery of features that result in accurate image classification on computationally constrained devices such as smartphones would enable new and promising applications for image classification. For example, smartphone retinal cameras can conceivably make diabetic retinopathy widely available and potentially reduce undiagnosed retinopathy if it could be achieved with computationally simple classification algorithms. A novel image feature extraction technique inspired by N-grams in text mining, called 'Pixel N-grams' is described that can serve this purpose. Results on mammogram and texture classification have shown high accuracy despite the reduced computational complexity. However retinal scan classification results using Pixel N-grams lag behind BoVW approaches. An explanation for the relative poor performance of Pixel N-grams with diabetic retinopathy that draws on concepts associated with the No Free Lunch theorem are presented.
Original languageEnglish
Title of host publicationProceedings of the Australasian Computer Science Week Multiconference, ACSW 2019
PublisherAssociation for Computing Machinery
Pages1-7
Number of pages7
ISBN (Electronic)9781450366038
DOIs
Publication statusPublished - 29 Jan 2019
Event2019 Australasian Computer Science Week Multiconference, ACSW 2019 - Macquarie University, Sydney, Australia
Duration: 29 Jan 201931 Jan 2019
https://dl.acm.org/citation.cfm?id=3290688&picked=prox (proceedings)

Conference

Conference2019 Australasian Computer Science Week Multiconference, ACSW 2019
CountryAustralia
CitySydney
Period29/01/1931/01/19
Internet address

Fingerprint

Pixels
Image classification
Smartphones
Covariance matrix
Feature extraction
Computational complexity
Textures
Cameras

Cite this

Kulkarni, P., Stranieri, A., & Jelinek, H. (2019). Comparing pixel N-grams and bag of visual word features for the classification of diabetic retinopathy. In Proceedings of the Australasian Computer Science Week Multiconference, ACSW 2019 (pp. 1-7). [a22] Association for Computing Machinery. https://doi.org/10.1145/3290688.3290726
Kulkarni, Pradnya ; Stranieri, Andrew ; Jelinek, Herbert. / Comparing pixel N-grams and bag of visual word features for the classification of diabetic retinopathy. Proceedings of the Australasian Computer Science Week Multiconference, ACSW 2019. Association for Computing Machinery, 2019. pp. 1-7
@inproceedings{3ae4e389791b46c3a4bb6746186d599a,
title = "Comparing pixel N-grams and bag of visual word features for the classification of diabetic retinopathy",
abstract = "The extraction of Bag of Visual Words (BoVW) features from retinal images for automated classification has been shown to be effective but computationally expensive. Histogram and co-variance matrix features do not generally result in models that have the same predictive accuracy as BoVW and are still computationally expensive. The discovery of features that result in accurate image classification on computationally constrained devices such as smartphones would enable new and promising applications for image classification. For example, smartphone retinal cameras can conceivably make diabetic retinopathy widely available and potentially reduce undiagnosed retinopathy if it could be achieved with computationally simple classification algorithms. A novel image feature extraction technique inspired by N-grams in text mining, called 'Pixel N-grams' is described that can serve this purpose. Results on mammogram and texture classification have shown high accuracy despite the reduced computational complexity. However retinal scan classification results using Pixel N-grams lag behind BoVW approaches. An explanation for the relative poor performance of Pixel N-grams with diabetic retinopathy that draws on concepts associated with the No Free Lunch theorem are presented.",
keywords = "Diabetic retinopathy, Image classification, Image feature extraction, No Free Lunch Theorem, Pixel N-gram",
author = "Pradnya Kulkarni and Andrew Stranieri and Herbert Jelinek",
year = "2019",
month = "1",
day = "29",
doi = "10.1145/3290688.3290726",
language = "English",
pages = "1--7",
booktitle = "Proceedings of the Australasian Computer Science Week Multiconference, ACSW 2019",
publisher = "Association for Computing Machinery",

}

Kulkarni, P, Stranieri, A & Jelinek, H 2019, Comparing pixel N-grams and bag of visual word features for the classification of diabetic retinopathy. in Proceedings of the Australasian Computer Science Week Multiconference, ACSW 2019., a22, Association for Computing Machinery, pp. 1-7, 2019 Australasian Computer Science Week Multiconference, ACSW 2019, Sydney, Australia, 29/01/19. https://doi.org/10.1145/3290688.3290726

Comparing pixel N-grams and bag of visual word features for the classification of diabetic retinopathy. / Kulkarni, Pradnya; Stranieri, Andrew; Jelinek, Herbert.

Proceedings of the Australasian Computer Science Week Multiconference, ACSW 2019. Association for Computing Machinery, 2019. p. 1-7 a22.

Research output: Book chapter/Published conference paperConference paper

TY - GEN

T1 - Comparing pixel N-grams and bag of visual word features for the classification of diabetic retinopathy

AU - Kulkarni, Pradnya

AU - Stranieri, Andrew

AU - Jelinek, Herbert

PY - 2019/1/29

Y1 - 2019/1/29

N2 - The extraction of Bag of Visual Words (BoVW) features from retinal images for automated classification has been shown to be effective but computationally expensive. Histogram and co-variance matrix features do not generally result in models that have the same predictive accuracy as BoVW and are still computationally expensive. The discovery of features that result in accurate image classification on computationally constrained devices such as smartphones would enable new and promising applications for image classification. For example, smartphone retinal cameras can conceivably make diabetic retinopathy widely available and potentially reduce undiagnosed retinopathy if it could be achieved with computationally simple classification algorithms. A novel image feature extraction technique inspired by N-grams in text mining, called 'Pixel N-grams' is described that can serve this purpose. Results on mammogram and texture classification have shown high accuracy despite the reduced computational complexity. However retinal scan classification results using Pixel N-grams lag behind BoVW approaches. An explanation for the relative poor performance of Pixel N-grams with diabetic retinopathy that draws on concepts associated with the No Free Lunch theorem are presented.

AB - The extraction of Bag of Visual Words (BoVW) features from retinal images for automated classification has been shown to be effective but computationally expensive. Histogram and co-variance matrix features do not generally result in models that have the same predictive accuracy as BoVW and are still computationally expensive. The discovery of features that result in accurate image classification on computationally constrained devices such as smartphones would enable new and promising applications for image classification. For example, smartphone retinal cameras can conceivably make diabetic retinopathy widely available and potentially reduce undiagnosed retinopathy if it could be achieved with computationally simple classification algorithms. A novel image feature extraction technique inspired by N-grams in text mining, called 'Pixel N-grams' is described that can serve this purpose. Results on mammogram and texture classification have shown high accuracy despite the reduced computational complexity. However retinal scan classification results using Pixel N-grams lag behind BoVW approaches. An explanation for the relative poor performance of Pixel N-grams with diabetic retinopathy that draws on concepts associated with the No Free Lunch theorem are presented.

KW - Diabetic retinopathy

KW - Image classification

KW - Image feature extraction

KW - No Free Lunch Theorem

KW - Pixel N-gram

UR - http://www.scopus.com/inward/record.url?scp=85061237652&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85061237652&partnerID=8YFLogxK

U2 - 10.1145/3290688.3290726

DO - 10.1145/3290688.3290726

M3 - Conference paper

AN - SCOPUS:85061237652

SP - 1

EP - 7

BT - Proceedings of the Australasian Computer Science Week Multiconference, ACSW 2019

PB - Association for Computing Machinery

ER -

Kulkarni P, Stranieri A, Jelinek H. Comparing pixel N-grams and bag of visual word features for the classification of diabetic retinopathy. In Proceedings of the Australasian Computer Science Week Multiconference, ACSW 2019. Association for Computing Machinery. 2019. p. 1-7. a22 https://doi.org/10.1145/3290688.3290726