Proceedings Article10.1117/12.2527210
Machine learning for optical coherence tomography angiography
Julian Lo,Morgan Heisler,Arman Athwal,Francis Tran,Marinko V. Sarunic +4 more
- 23 Jun 2019
- Vol. 11078, pp 106-108
9
TL;DR: To further improve the clinical utility of these OCT-A images, segmentation is essential as it allows for the quantitative analysis of the microvasculature, which include the identification of the foveal avascular zone (FAZ) and areas of capillary non-perfusion (CNP), two biomarkers for the progression of DR.
read more
Abstract: 1. INTRODUCTION Optical coherence tomography angiography (OCT-A) is a non-invasive imaging modality allowing researchers and clinicians to view the retina in micrometer-scale detail. Acquired OCT-A volumes are three-dimensional, allowing the visualization of the superficial capillary plexus (SCP) and the deep capillary plexus (DCP). This provides valuable information towards the identification of pathologies such as diabetic retinopathy (DR). However, because an OCT-A volume is acquired over several seconds, motion artifacts caused by rapid movements of the subject’s eye (also known as micro-saccadic motion) can greatly reduce the quality, and subsequently the clinical utility, of the resulting volumes. Hardware motion tracking aims to reduce the effect of motion, but non-rigid registration is still often required for averaging sequentially acquired images. Furthermore, not all prototype OCT-A systems have tracking capabilities, particularly adaptive optics (AO) systems. Because of this, image registration is essential for the elimination of motion artifacts in OCT-A volumes, increasing their clinical diagnostic value. To further improve the clinical utility of these OCT-A images, segmentation is essential as it allows for the quantitative analysis of the microvasculature, which include the identification of the foveal avascular zone (FAZ) and areas of capillary non-perfusion (CNP), two biomarkers for the progression of DR.
read more
Chat with Paper
AI Agents for this Paper
Find similar papers on Google Scholar, PubMed and Arxiv
Write a critical review of this paper
Analyze citations of this paper to find unaddressed research gaps
Citations
Quantitative approaches in multimodal fundus imaging: State of the art and future perspectives
TL;DR: A comprehensive survey of multimodal imaging techniques can be found in this article , covering their limitations as well as their strengths, technical features, limitations, and interpretation, and the main imaging artifacts and their potential impact on imaging quality and reliability.
37
Advanced vascular examinations of the retina and optic nerve head in glaucoma
TL;DR: This chapter reviews vascular-oriented technologies in glaucoma with a special focus given to optical coherence tomography angiography, and gives an update on the improvements needed to bring vascular assessments closer to everyday clinical practice.
6
Optical coherence tomography and convolutional neural networks can differentiate colorectal liver metastases from liver parenchyma ex vivo
I. Amygdalos,E. Hachgenei,L. Burkl,David Vargas,Paul Goßmann,Laura I. Wolff,M. O. Druzenko,Maik Frye,Niels König,Robert Schmitt,Alexandros Chrysos,K Jöchle,Tom Florian Ulmer,Andreas Lambertz,Ruth Knüchel-Clarke,Ulf P. Neumann,Sven Arke Lang +16 more
TL;DR: In this article , the ability of OCT to differentiate colorectal liver metastases (CRLM) from healthy liver parenchyma, when combined with convolutional neural networks (CNN) was investigated.
5
Direct estimation of gas holdup in gas–liquid bubble column reactors using ultrasonic transmission tomography and artificial neural processing
TL;DR: In this article , reference indirect image-based estimates were obtained from reconstructed tomographic data, and direct (non-image) estimation of the gas holdup ratio was also obtained using trained neural processing networks.
4
Exploiting Pre-trained Architectures for Dual-Stream Classification of LCA-RCA in a Private AngioData
Hounaïda Moalla,Aiman Ghrab,Bassem Ben Hamed,Amine Bahloul,Leila Abid +4 more
- 20 Sep 2023
TL;DR: This study focuses on the classification of Left and Right Coronary Artery using Deep Learning architectures applied to frames extracted from angiographic videos, and evaluates Machine Learning methods and finds that the Random Forest algorithm achieves an accuracy of 89% when utilizing the video metadata.
2
References
•Posted Content
U-Net: Convolutional Networks for Biomedical Image Segmentation
TL;DR: It is shown that such a network can be trained end-to-end from very few images and outperforms the prior best method (a sliding-window convolutional network) on the ISBI challenge for segmentation of neuronal structures in electron microscopic stacks.
19.5K
Segmentation of the foveal microvasculature using deep learning networks.
Pavle Prentasic,Morgan Heisler,Zaid Mammo,Sieun Lee,Andrew Merkur,Eduardo V. Navajas,Mirza Faisal Beg,Marinko V. Sarunic,Sven Lončarić +8 more
TL;DR: The automated results were shown to be comparable to the human raters suggesting that segmentation using DNNs is comparable to a second manual rater, which is an important step in creating an automated output of optical coherence tomography angiography images.
Strip-based registration of serially acquired optical coherence tomography angiography
Morgan Heisler,Sieun Lee,Zaid Mammo,Yifan Jian,Myeong Jin Ju,Andrew Merkur,Eduardo V. Navajas,Chandrakumar Balaratnasingam,Mirza Faisal Beg,Marinko V. Sarunic +9 more
TL;DR: An automated method for registration and averaging of serially acquired OCT-A images can enable robust quantification and study of minute changes in retinal microvasculature and result in repeatable and consistent visualization.