Fine-Scale Vessel Extraction in Fundus Images by Registration with Fluorescein Angiography

Noh, Kyoung Jin; Park, Sang Jun; Lee, Soochahn

doi:10.1007/978-3-030-32239-7_86

Cited by 4 publications

(7 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We propose a modified version of the registration framework of Noh et al [15,16], combining keypoint matching based rigid registration, CNN based vessel segmentation, and B-spline based non-rigid registration. Our modification stems from the need to enhance the number of keypoints in the rigid registration for the peripheral frames with less amount of texture when compared to frames containing the optic disc.…”

Section: Frame Integrationmentioning

confidence: 99%

“…In the following, we review each subprocess along with the process for blending new frames into the montage. A visual summary of the frame integration pipeline, including a two-step rigid and non-rigid registration method adapted from [15], together with image blending.…”

Section: Frame Integrationmentioning

confidence: 99%

“…Figure 4 depicts a visualization of this process. We note that we train the SSANet on 554 images that were sampled from our private dataset described in Section 3.1, with ground truth being generated by the method of Noh et al [15,16], based on the registered vessels of corresponding fluorescein angiography (FA) images, with minor manual corrections.…”

Section: Vessel Segmentation and Frame Preprocessingmentioning

confidence: 99%

“…We apply a deep learning based object detection method [14] to detect the optic disc and retinal fovea, which is then used as reference landmarks to determine the order of image registration and alignment. We also adopt a recently proposed two step image registration framework [15,16] comprising rigid and non-rigid registration, but with key modifications to improve robustness for images in the peripheral of the retina. Here, CNN based vessel segmentation is used as the basis to maximize the accuracy of both rigid and non-rigid registration.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Fully Leveraging Deep Learning Methods for Constructing Retinal Fundus Photomontages

Kim

Noh

et al. 2021

Applied Sciences

Self Cite

View full text Add to dashboard Cite

Retinal photomontages, which are constructed by aligning and integrating multiple fundus images, are useful in diagnosing retinal diseases affecting peripheral retina. We present a novel framework for constructing retinal photomontages that fully leverage recent deep learning methods. Deep learning based object detection is used to define the order of image registration and blending. Deep learning based vessel segmentation is used to enhance image texture to improve registration performance within a two step image registration framework comprising rigid and non-rigid registration. Experimental evaluation demonstrates the robustness of our montage construction method with an increased amount of successfully integrated images as well as reduction of image artifacts.

show abstract

Section: Frame Integrationmentioning

confidence: 99%

Section: Frame Integrationmentioning

confidence: 99%

Section: Vessel Segmentation and Frame Preprocessingmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Fully Leveraging Deep Learning Methods for Constructing Retinal Fundus Photomontages

Kim

Noh

et al. 2021

Applied Sciences

Self Cite

View full text Add to dashboard Cite

show abstract

“…Previous works have also been proposed that demonstrate the benefits from combining the two image modalities of FA sequence and fundus image. In our previous work in [17], [18], we presented a method to take advantage of the detailed vascular structure highlighted in FA to obtain highly detailed vessel segmentation masks for fundus images. The benefit is even larger for artery/vein (A/V) classification since the two image modalities have complementary characteristics.…”

Section: Introductionmentioning

confidence: 99%

Combined Deep Learning of Fundus Images and Fluorescein Angiography for Retinal Artery/Vein Classification

Kim

Noh³

et al. 2022

IEEE Access

Self Cite

View full text Add to dashboard Cite

We present a new framework for achieving highly accurate retinal artery/vein classification based on a combined input of fundus images and corresponding fluorescein angiography (FA) images. Although the observable bloodflow in FA sequences should be a definitive cue to discern artery and vein, it is often insufficient due to limitations in frame rate and image quality. Since fundus images are acquired by default, we incorporate both the fundus image and FA as the input to a deep learning framework comprising convolutional and graph neural networks. First, a convolutional neural network with a parallel structure is learned to extract features from the two image modalities. Then, these features are used as input to dual hierarchical graph neural networks for pixelwise classification based on the long-range connectivity of vessels. We provide empirical evidence based on ablative and comparative quantitative evaluations supporting the effectiveness of the proposed network configuration in combining the fundus image and FA. We also provide comparative evaluation supporting the improvement in performance compared to previous methods.

show abstract

Combining Fundus Images and Fluorescein Angiography for Artery/Vein Classification Using the Hierarchical Vessel Graph Network

Noh¹,

Park

Lee

2020

Medical Image Computing and Computer Assisted Intervention – MICCAI 2020

View full text Add to dashboard Cite

Fine-Scale Vessel Extraction in Fundus Images by Registration with Fluorescein Angiography

Cited by 4 publications

References 13 publications

Fully Leveraging Deep Learning Methods for Constructing Retinal Fundus Photomontages

Fully Leveraging Deep Learning Methods for Constructing Retinal Fundus Photomontages

Combined Deep Learning of Fundus Images and Fluorescein Angiography for Retinal Artery/Vein Classification

Combining Fundus Images and Fluorescein Angiography for Artery/Vein Classification Using the Hierarchical Vessel Graph Network

Contact Info

Product

Resources

About