A Review on the Extraction of Quantitative Retinal Microvascular Image Feature

Kipli, Kuryati; Hoque, Mohammed Enamul; Lim, Lik Thai; Mahmood, Muhammad Hamdi; Sahari, Siti Kudnie; Sapawi, Rohana; Rajaee, Nordiana; Joseph, Annie

doi:10.1155/2018/4019538

Cited by 39 publications

(25 citation statements)

References 114 publications

(167 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…New technologies are being developed for analysis of retinal microvasculature in fundus photographs, which may soon allow for automated screening for conditions such as diabetic retinopathy, hypertensive retinopathy, and macular degeneration. 7 Poplin et al used deep-learning models on retinal imaging to predict cardiovascular risk factors, such as age, gender, smoking status, systolic blood pressure, and major adverse cardiac events. 8 Spectral domain optical coherence tomography (SD-OCT) may demonstrate thinning of the inner retina and focal attenuation of the ellipsoid zone in areas of the Elschnig spots.…”

Section: Diagnosis Of Hypertensive Chorioretinopathymentioning

confidence: 99%

<p>A Review of Hypertensive Retinopathy and Chorioretinopathy</p>

Tsukikawa¹,

Stacey²

2020

OPTO

View full text Add to dashboard Cite

Hypertensive retinopathy and choroidopathy have important short-and long-term implications on patients' overall health and mortality. Eye care professionals should be familiar with the severity staging of these entities and be able to readily recognize and refer patients who are in need of systemic blood pressure control. This paper will review the diagnosis, staging, treatment, and long-term implications for vision and mortality of patients with hypertensive retinopathy and choroidopathy.

show abstract

Section: Diagnosis Of Hypertensive Chorioretinopathymentioning

confidence: 99%

<p>A Review of Hypertensive Retinopathy and Chorioretinopathy</p>

Tsukikawa¹,

Stacey²

2020

OPTO

View full text Add to dashboard Cite

show abstract

“…From the microcirculation perspective, retinal microvasculopathy may reflect small vessel disease. The lower fractal dimension which reflects a sparser retinal microvascular network is associated with stroke, Alzheimer's disease, and more diffuse and severe coronary artery disease [20][21][22][23]. As demonstrated, albuminuria is significantly and independently associated with the presence and severity of atherosclerosis [12].…”

Section: Discussionmentioning

confidence: 89%

Chinese ASCVD risk equations rather than pooled cohort equations are better to identify macro- and microcirculation abnormalities

Lin

Gao

et al. 2020

BMC Cardiovasc Disord

View full text Add to dashboard Cite

Background: We hypothesized that discriminating the early subclinical organ damage would serve as a great opportunity for prevention against atherosclerotic cardiovascular disease (ASCVD). Brachial-ankle pulse wave velocity (baPWV), low retinal vascular fractal dimension, and albuminuria are surrogates of subclinical vascular changes. Methods: The aim of this study was to use Pooled Cohort Equations (PCE) and ASCVD risk equations derived from "Prediction for ASCVD Risk in China project (CHINA-PAR)" to observe the prevalence of macro-and microcirculation abnormalities. A total of 2166 subjects were involved. Characteristics were investigated using questionnaire and physical examinations. We calculated the urine albumin to creatinine ratio (UACR). The baPWV was measured using a fully automatic arteriosclerosis detector. The retinal vascular fractal dimension was measured by a semiautomated computer-based program. The 10-year ASCVD risk was estimated using the PCE and CHINA-PAR model. Results: The cutoff values for the elevated baPWV were 2.82 and 2.92% in the PCE model and CHINA-PAR model, respectively, with nearly 85% sensitivity and an average specificity of 74%. For low retinal fractal dimension, at the cutoff point of 3.8%, we acquired an acceptable sensitivity of 66.27-68.24% and specificity of 62.57-67.45%. All the C-statistics presented a significant improvement from the PCE model to the CHINA-PAR model (P < 0.05). For all categories-net reclassification improvement (NRI) values were significant and clearly varied (0.329, 0.183, and 0.104, respectively) depending on the cutoff set at 3%. Conclusion: Our study demonstrated that the CHINA-PAR equations rather than PCE could provide better identification of macro-and microcirculation abnormalities. A lower cutoff point for the subclinical vascular changes may be selected in a population from southeast China.

show abstract

“…Meanwhile, it demonstrates that our strategy of adopting the newly constructed structured dropout convolutional block to build Backbone is effective. (2) With only 98 parameters added, the AUC, F1, and MCC of SA-UNet are 0.05% / 0.08%, 0.29% / 0.31%, and 0.34% / 0.36% higher than Backbone, respectively, which proves the strategy of the introduction of spatial attention is effective. (3) Compared with the original U-Net with 23 convolutional layers, our SA-UNet has a much smaller amount of parameters, so for the task of retinal blood vessel segmentation, SA-UNet is a lightweight and effective network .…”

Section: A Ablation Experimentsmentioning

confidence: 85%

SD-Unet: A Structured Dropout U-Net for Retinal Vessel Segmentation

Guo

Szemenyei

Yang

et al. 2019

2019 IEEE 19th International Conference on Bioinformatics and Bioengineering (BIBE)

View full text Add to dashboard Cite

The precise segmentation of retinal blood vessel is of great significance for early diagnosis of eye-related diseases such as diabetes and hypertension. In this work, we propose a lightweight network named Spatial Attention U-Net (SA-UNet) that does not require thousands of annotated training samples and can be utilized in a data augmentation manner to use the available annotated samples more efficiently. SA-UNet introduces a spatial attention module which infers the attention map along the spatial dimension, and then multiply the attention map by the input feature map for adaptive feature refinement. In addition, the proposed network employees a kind of structured dropout convolutional block instead of the original convolutional block of U-Net to prevent the network from overfitting. We evaluate SA-UNet based on two benchmark retinal datasets: the Vascular Extraction (DRIVE) dataset and the Child Heart and Health Study (CHASE_DB1) dataset. The results show that our proposed SA-UNet achieves the state-of-the-art retinal vessel segmentation accuracy on both datasets. 

show abstract

A Review on the Extraction of Quantitative Retinal Microvascular Image Feature

Cited by 39 publications

References 114 publications

<p>A Review of Hypertensive Retinopathy and Chorioretinopathy</p>

<p>A Review of Hypertensive Retinopathy and Chorioretinopathy</p>

Chinese ASCVD risk equations rather than pooled cohort equations are better to identify macro- and microcirculation abnormalities

SD-Unet: A Structured Dropout U-Net for Retinal Vessel Segmentation

Contact Info

Product

Resources

About