Brain-inspired algorithms for retinal image analysis

Romeny, Bart M. Haar; Bekkers, Erik J.; Zhang, Jiong; Abbasi-Sureshjani, Samaneh; Huang, Fan; Duits, Remco; Dashtbozorg, Behdad; Berendschot, Tos T. J. M.; Smit-Ockeloen, Iris; Eppenhof, Koen A. J.; Feng, Jie; Hannink, Julius; Schouten, Jan S.A.G.; Tong, Mengmeng; Wu, Hanhui; Triest, Han W.; Zhu, Shenghuo; Chen, Dali; He, Wei; Xu, Ling; Han, Pei; Kang, Yan

doi:10.1007/s00138-016-0771-9

Cited by 29 publications

(27 citation statements)

References 59 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A selection of four biomarkers was chosen from an extensive list of features that were automatically calculated by the pipeline of methods described by ter Haar Romeny et al 7 The biomarkers were chosen such that they represent key vessel geometries: vessel caliber, tortuosity and bifurcations. Descriptions of the biomarkers are given in Tab.…”

Section: Methodsmentioning

confidence: 99%

“…6 A pipeline that combines the validated algorithms has been assembled for the RetinaCheck project, a large-scale diabetes screening program in China. 7 The current pipeline is computationally expensive (taking 24 minutes to process an image) and could benefit substantially from advances in machine learning techniques to speed up the biomarker extraction. This paper is a proof of principle for the approximation of part of the pipeline of retinal image analysis methods into a single convolutional neural network (CNN).…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Approximation of a pipeline of unsupervised retina image analysis methods with a CNN

Heslinga

Pluim

Bozorg

et al. 2019

Medical Imaging 2019: Image Processing

Self Cite

View full text Add to dashboard Cite

A pipeline of unsupervised image analysis methods for extraction of geometrical features from retinal fundus images has previously been developed. Features related to vessel caliber, tortuosity and bifurcations, have been identified as potential biomarkers for a variety of diseases, including diabetes and Alzheimer's. The current computationally expensive pipeline takes 24 minutes to process a single image, which impedes implementation in a screening setting. In this work, we approximate the pipeline with a convolutional neural network (CNN) that enables processing of a single image in a few seconds. As an additional benefit, the trained CNN is sensitive to key structures in the retina and can be used as a pretrained network for related disease classification tasks. Our model is based on the ResNet-50 architecture and outputs four biomarkers that describe global properties of the vascular tree in retinal fundus images. Intraclass correlation coefficients between the predictions of the CNN and the results of the pipeline showed strong agreement (0.86 -0.91) for three of four biomarkers and moderate agreement (0.42) for one biomarker. Class activation maps were created to illustrate the attention of the network. The maps show qualitatively that the activations of the network overlap with the biomarkers of interest, and that the network is able to distinguish venules from arterioles. Moreover, local high and low tortuous regions are clearly identified, confirming that a CNN is sensitive to key structures in the retina.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Approximation of a pipeline of unsupervised retina image analysis methods with a CNN

Heslinga

Pluim

Bozorg

et al. 2019

Medical Imaging 2019: Image Processing

Self Cite

View full text Add to dashboard Cite

show abstract

“…Based on this infrastructure, an application has been developed which automatically computes several biomarkers from retinal images in a repeatable and objective manner. The algorithms used in this application outperform most of the state-of-the-art techniques [10].…”

Section: Introductionmentioning

confidence: 96%

Infrastructure for Retinal Image Analysis

Dashtbozorg

Abbasi-Sureshjani

Zhang

et al. 2016

Proceedings of the Ophthalmic Medical Image Analysis Third International Workshop

Self Cite

View full text Add to dashboard Cite

Abstract. This paper introduces a retinal image analysis infrastructure for the automatic assessment of biomarkers related to early signs of diabetes, hypertension and other systemic diseases. The developed application provides several tools, namely normalization, vessel enhancement and segmentation, optic disc and fovea detection, junction detection, bifurcation/crossing discrimination, artery/vein classification and red lesion detection. The pipeline of these methods allows the assessment of important biomarkers characterizing dynamic properties of retinal vessels, such as tortuosity, width, fractal dimension and bifurcation geometry features.

show abstract

“…Many studies [1], [2] have been performed on various retinal imaging modalities to analyze vascular changes. A comprehensive overview of technical and clinical studies is provided by Kashani et al [1] to show the great promise of using OCT-A to analyze multiple diseases like glaucomatous optic neuropathy, DR, AMD, uveitis and macular telangiectasia type 2.…”

Section: Introductionmentioning

confidence: 99%

3D Shape Modeling and Analysis of Retinal Microvasculature in OCT-Angiography Images

Zhang

Qiao

Sarabi

et al. 2020

IEEE Trans. Med. Imaging

Self Cite

View full text Add to dashboard Cite

3D optical coherence tomography angiography (OCT-A) is a novel and non-invasive imaging modality for analyzing retinal diseases. The studies of microvasculature in 2D en face projection images have been widely implemented, but comprehensive 3D analysis of OCT-A images with rich depth-resolved microvascular information is rarely considered. In this paper, we propose a robust, effective, and automatic 3D shape modeling framework to provide a high-quality 3D vessel representation and to preserve valuable 3D geometric and topological information for vessel analysis. Effective vessel enhancement and extraction steps by means of curvelet denoising and optimally oriented flux (OOF) filtering are first designed to produce 3D microvascular networks.

show abstract

Brain-inspired algorithms for retinal image analysis

Cited by 29 publications

References 59 publications

Approximation of a pipeline of unsupervised retina image analysis methods with a CNN

Approximation of a pipeline of unsupervised retina image analysis methods with a CNN

Infrastructure for Retinal Image Analysis

3D Shape Modeling and Analysis of Retinal Microvasculature in OCT-Angiography Images

Contact Info

Product

Resources

About