Support vector machine and deep-learning object detection for localisation of hard exudates

Abstract Hard exudates are one of the main clinical findings in the retinal images of patients with diabetic retinopathy. Detecting them early significantly impacts the treatment of underlying diseases; therefore, there is a need for automated systems with high reliability. We propose a novel method...

Descripción completa

Guardado en:
Detalles Bibliográficos
Autores principales: Veronika Kurilová, Jozef Goga, Miloš Oravec, Jarmila Pavlovičová, Slavomír Kajan
Formato: article
Lenguaje:EN
Publicado: Nature Portfolio 2021
Materias:
R
Q
Acceso en línea:https://doaj.org/article/7f38062e984643ea8b7d5b74483d7c12
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
id oai:doaj.org-article:7f38062e984643ea8b7d5b74483d7c12
record_format dspace
spelling oai:doaj.org-article:7f38062e984643ea8b7d5b74483d7c122021-12-02T18:49:30ZSupport vector machine and deep-learning object detection for localisation of hard exudates10.1038/s41598-021-95519-02045-2322https://doaj.org/article/7f38062e984643ea8b7d5b74483d7c122021-08-01T00:00:00Zhttps://doi.org/10.1038/s41598-021-95519-0https://doaj.org/toc/2045-2322Abstract Hard exudates are one of the main clinical findings in the retinal images of patients with diabetic retinopathy. Detecting them early significantly impacts the treatment of underlying diseases; therefore, there is a need for automated systems with high reliability. We propose a novel method for identifying and localising hard exudates in retinal images. To achieve fast image pre-scanning, a support vector machine (SVM) classifier was combined with a faster region-based convolutional neural network (faster R-CNN) object detector for the localisation of exudates. Rapid pre-scanning filtered out exudate-free samples using a feature vector extracted from the pre-trained ResNet-50 network. Subsequently, the remaining samples were processed using a faster R-CNN detector for detailed analysis. When evaluating all the exudates as individual objects, the SVM classifier reduced the false positive rate by 29.7% and marginally increased the false negative rate by 16.2%. When evaluating all the images, we recorded a 50% reduction in the false positive rate, without any decrease in the number of false negatives. The interim results suggested that pre-scanning the samples using the SVM prior to implementing the deep-network object detector could simultaneously improve and speed up the current hard exudates detection method, especially when there is paucity of training data.Veronika KurilováJozef GogaMiloš OravecJarmila PavlovičováSlavomír KajanNature PortfolioarticleMedicineRScienceQENScientific Reports, Vol 11, Iss 1, Pp 1-9 (2021)
institution DOAJ
collection DOAJ
language EN
topic Medicine
R
Science
Q
spellingShingle Medicine
R
Science
Q
Veronika Kurilová
Jozef Goga
Miloš Oravec
Jarmila Pavlovičová
Slavomír Kajan
Support vector machine and deep-learning object detection for localisation of hard exudates
description Abstract Hard exudates are one of the main clinical findings in the retinal images of patients with diabetic retinopathy. Detecting them early significantly impacts the treatment of underlying diseases; therefore, there is a need for automated systems with high reliability. We propose a novel method for identifying and localising hard exudates in retinal images. To achieve fast image pre-scanning, a support vector machine (SVM) classifier was combined with a faster region-based convolutional neural network (faster R-CNN) object detector for the localisation of exudates. Rapid pre-scanning filtered out exudate-free samples using a feature vector extracted from the pre-trained ResNet-50 network. Subsequently, the remaining samples were processed using a faster R-CNN detector for detailed analysis. When evaluating all the exudates as individual objects, the SVM classifier reduced the false positive rate by 29.7% and marginally increased the false negative rate by 16.2%. When evaluating all the images, we recorded a 50% reduction in the false positive rate, without any decrease in the number of false negatives. The interim results suggested that pre-scanning the samples using the SVM prior to implementing the deep-network object detector could simultaneously improve and speed up the current hard exudates detection method, especially when there is paucity of training data.
format article
author Veronika Kurilová
Jozef Goga
Miloš Oravec
Jarmila Pavlovičová
Slavomír Kajan
author_facet Veronika Kurilová
Jozef Goga
Miloš Oravec
Jarmila Pavlovičová
Slavomír Kajan
author_sort Veronika Kurilová
title Support vector machine and deep-learning object detection for localisation of hard exudates
title_short Support vector machine and deep-learning object detection for localisation of hard exudates
title_full Support vector machine and deep-learning object detection for localisation of hard exudates
title_fullStr Support vector machine and deep-learning object detection for localisation of hard exudates
title_full_unstemmed Support vector machine and deep-learning object detection for localisation of hard exudates
title_sort support vector machine and deep-learning object detection for localisation of hard exudates
publisher Nature Portfolio
publishDate 2021
url https://doaj.org/article/7f38062e984643ea8b7d5b74483d7c12
work_keys_str_mv AT veronikakurilova supportvectormachineanddeeplearningobjectdetectionforlocalisationofhardexudates
AT jozefgoga supportvectormachineanddeeplearningobjectdetectionforlocalisationofhardexudates
AT milosoravec supportvectormachineanddeeplearningobjectdetectionforlocalisationofhardexudates
AT jarmilapavlovicova supportvectormachineanddeeplearningobjectdetectionforlocalisationofhardexudates
AT slavomirkajan supportvectormachineanddeeplearningobjectdetectionforlocalisationofhardexudates
_version_ 1718377569741438976