Relationships Between Retinal Vascular Characteristics and Renal Function in Patients With Type 2 Diabetes Mellitus

Evaluating the potential of retinal photography in chronic kidney disease detection: a review

Background

Chronic kidney disease (CKD) is a significant global health concern, emphasizing the necessity of early detection to facilitate prompt clinical intervention. Leveraging the unique ability of the retina to offer insights into systemic vascular health, it emerges as an interesting, non-invasive option for early CKD detection. Integrating this approach with existing invasive methods could provide a comprehensive understanding of patient health, enhancing diagnostic accuracy and treatment effectiveness.

Objectives

The purpose of this review is to critically assess the potential of retinal imaging to serve as a diagnostic tool for CKD detection based on retinal vascular changes. The review tracks the evolution from conventional manual evaluations to the latest state-of-the-art in deep learning.

Survey Methodology

A comprehensive examination of the literature was carried out, using targeted database searches and a three-step methodology for article evaluation: identification, screening, and inclusion based on Prisma guidelines. Priority was given to unique and new research concerning the detection of CKD with retinal imaging. A total of 70 publications from 457 that were initially discovered satisfied our inclusion criteria and were thus subjected to analysis. Out of the 70 studies included, 35 investigated the correlation between diabetic retinopathy and CKD, 23 centered on the detection of CKD via retinal imaging, and four attempted to automate the detection through the combination of artificial intelligence and retinal imaging.

Results

Significant retinal features such as arteriolar narrowing, venular widening, specific retinopathy markers (like microaneurysms, hemorrhages, and exudates), and changes in arteriovenous ratio (AVR) have shown strong correlations with CKD progression. We also found that the combination of deep learning with retinal imaging for CKD detection could provide a very promising pathway. Accordingly, leveraging retinal imaging through this technique is expected to enhance the precision and prognostic capacity of the CKD detection system, offering a non-invasive diagnostic alternative that could transform patient care practices.

Conclusion

In summary, retinal imaging holds high potential as a diagnostic tool for CKD because it is non-invasive, facilitates early detection through observable microvascular changes, offers predictive insights into renal health, and, when paired with deep learning algorithms, enhances the accuracy and effectiveness of CKD screening.

Correlation between retinal vascular geometric parameters and pathologically diagnosed type 2 diabetic nephropathy

Background

Diabetic nephropathy (DN) and diabetic retinopathy (DR) are common microvascular complications of diabetes. The purpose of this study was to investigate the correlation between retinal vascular geometric parameters and pathologically diagnosed type 2 DN and to determine the capacity of retinal vascular geometric parameters in differentiating DN from non-diabetic renal disease (NDRD).

Methods

The study participants were adult patients with type 2 diabetes mellitus (T2DM) and chronic kidney disease who underwent a renal biopsy. Univariate and multivariable regression analyses were performed to evaluate associations between retinal vessel geometry parameters and pathologically diagnosed DN. Multivariate binary logistic regression analyses were performed to establish a differential diagnostic model for DN.

Results

In total, 403 patients were examined in this cross-sectional study, including 152 (37.7%) with DN, 157 (39.0%) with NDRD and 94 (23.3%) with DN combined with NDRD. After univariate logistic regression, total vessel fractal dimension, arteriolar fractal dimension and venular fractal dimension were all found to be associated with DN. In multivariate analyses adjusting for age, sex, blood pressure, diabetes, DR and other factors, smaller retinal vascular fractal dimensions were significantly associated with DN (P < .05). We developed a differential diagnostic model for DN combining traditional clinical indicators and retinal vascular geometric parameters. The area under the curve of the model established by multivariate logistic regression was 0.930.

Conclusions

Retinal vessel fractal dimension is of great significance for the rapid and non-invasive differentiation of DN. Incorporating retinal vessel fractal dimension into the diagnostic model for DN and NDRD can improve the diagnostic efficiency.

Ocular biomarkers of cognitive decline based on deep-learning retinal vessel segmentation

BACKGROUND

The current literature shows a strong relationship between retinal neuronal and vascular alterations in dementia. The purpose of the study was to use NFN+ deep learning models to analyze retinal vessel characteristics for cognitive impairment (CI) recognition.

METHODS

We included 908 participants from a community-based cohort followed for over 15 years (the prospective KaiLuan Study) who underwent brain magnetic resonance imaging (MRI) and fundus photography between 2021 and 2022. The cohort consisted of both cognitively healthy individuals (N = 417) and those with cognitive impairment (N = 491). We employed the NFN+ deep learning framework for retinal vessel segmentation and measurement. Associations between Retinal microvascular parameters (RMPs: central retinal arteriolar / venular equivalents, arteriole to venular ratio, fractal dimension) and CI were assessed by Pearson correlation. P < 0.05 was considered statistically significant. The correlation between the CI and RMPs were explored, then the correlation coefficients between CI and RMPs were analyzed. Random Forest nonlinear classification model was used to predict whether one having cognitive decline or not. The assessment criterion was the AUC value derived from the working characteristic curve.

RESULTS

The fractal dimension (FD) and global vein width were significantly correlated with the CI (P < 0.05). Age (0.193), BMI (0.154), global vein width (0.106), retinal vessel FD (0.099), and CRAE (0.098) were the variables in this model that were ranked in order of feature importance. The AUC values of the model were 0.799.

CONCLUSIONS

Establishment of a predictive model based on the extraction of vascular features from fundus images has a high recognizability and predictive power for cognitive function and can be used as a screening method for CI.

The association of global vessel width with cognitive decline and cerebral small vessel disease burden in the KaiLuan study

Background

As the retinal microvasculature shares similarities with the cerebral microvasculature, numerous studies have shown that retinal vascular is associated with cognitive decline. In addition, several population-based studies have confirmed the association between retinal vascular and cerebral small vessel disease (CSVD) burden. However, the association of retinal vascular with CSVD burden as well as cognitive function has not been explored simultaneously. This study investigated the relations of retinal microvascular parameters (RMPs) with CSVD burden and cognitive function.

Methods

We conducted a cross-sectional study of participants in the KaiLuan study. Data were collected from subjects aged ≥18 years old who could complete retinal photography and brain magnetic resonance imaging (MRI) between December 2020 to October 2021 in the Kailuan community of Tangshan. RMPs were evaluated using a deep learning system. The cognitive function was measured using the Montreal Cognitive Assessment (MoCA). We conducted logistic regression models, and mediation analysis to evaluate the associations of RMPs with CSVD burden and cognitive decline.

Results

Of the 905 subjects (mean age: 55.42±12.02 years, 54.5% female), 488 (53.9%) were classified with cognitive decline. The fractal dimension (FD) [odds ratio (OR), 0.098, 95% confidence interval (CI): 0.015-0.639, P=0.015] and global vein width (OR: 1.010, 95% CI: 1.005-1.015, P<0.001) were independent risk factors for cognitive decline after adjustment for potential confounding factors. The global artery width was significantly associated with severe CSVD burden (OR: 0.985, 95% CI: 0.974-0.997, P=0.013). The global vein width was sightly associated with severe CSVD burden (OR: 1.005, 95% CI: 1.000-1.010, P=0.050) after adjusting for potential confounders. The multivariable-adjusted odds ratios (95% CI) in highest tertile versus lowest tertile of global vein width were 1.290 (0.901-1.847) for cognitive decline and 1.546 (1.004-2.290) for severe CSVD burden, respectively. Moreover, CSVD burden played a partial mediating role in the association between global vein width and cognitive function (mediating effect 6.59%).

Conclusions

RMPs are associated with cognitive decline and the development of CSVD. A proportion of the association between global vein width and cognitive decline may be attributed to the presence of CSVD burden.

Deep learning-based fundus image analysis for cardiovascular disease: a review

It is well established that the retina provides insights beyond the eye. Through observation of retinal microvascular changes, studies have shown that the retina contains information related to cardiovascular disease. Despite the tremendous efforts toward reducing the effects of cardiovascular diseases, they remain a global challenge and a significant public health concern. Conventionally, predicting the risk of cardiovascular disease involves the assessment of preclinical features, risk factors, or biomarkers. However, they are associated with cost implications, and tests to acquire predictive parameters are invasive. Artificial intelligence systems, particularly deep learning (DL) methods applied to fundus images have been generating significant interest as an adjunct assessment tool with the potential of enhancing efforts to prevent cardiovascular disease mortality. Risk factors such as age, gender, smoking status, hypertension, and diabetes can be predicted from fundus images using DL applications with comparable performance to human beings. A clinical change to incorporate DL systems for the analysis of fundus images as an equally good test over more expensive and invasive procedures may require conducting prospective clinical trials to mitigate all the possible ethical challenges and medicolegal implications. This review presents current evidence regarding the use of DL applications on fundus images to predict cardiovascular disease.

Panretinal photocoagulation plus intravitreal conbercept for diabetic retinopathy in real world: a retrospective study

BACKGROUND

To evaluate outcomes of panretinal photocoagulation (PRP) plus intravitreal conbercept (IVC) for diabetic retinopathy (DR) in real world and explore risk factors for patients with poor reactivity and presence of vision-threatening complications after combination treatment.

METHODS

Retrospective review of DR patients received PRP plus IVC over 6 months. The main outcome was improvement ≥ 2 steps in ETDRS diabetic retinopathy severity scale (DRSS) levels. Different strategies for eyes receiving IVC within or over 1 month after PRP were analyzed. For patients with DRSS improvement < 2 steps and presence of vision-threatening adverse events, a binary logistic regression method was used to select risk factors.

RESULTS

Sixty one eyes were involved in this study. After treated with combination therapy with a median number of 3 injections, 44% of eyes improved ≥ 2 steps in DRSS levels. A total of 14 eyes (23%) occurred vision-threatening adverse events. No significant difference was found in eyes receiving conbercept within or over 1 month after PRP. Duration of diabetes (OR 0.849, 95%CI 0.734-0.982, P = 0.027), GFR (OR 0.961, 95%CI 0.933-0.990, P = 0.010) and baseline DRSS levels (OR 3.290, 95%CI 1.483-7.295, P = 0.003) were independent risk factors for DRSS improvement < 2 steps after treatment. Occurrence of vision-threatening complications was only related to high DRSS levels (OR 3.668, 95%CI 1.710-7.868, P = 0.001).

CONCLUSIONS

The combination therapy was effective for most patients with DR in real world. Eyes received PRP combined with earlier or later conbercept was demonstrated no significant difference for outcomes. For patients with poor renal function, high DRSS levels or occurred DR at the early stage of diabetes, follow-up should be strengthened.

Correlation of Glucose and Lipid Metabolism Levels and Serum Uric Acid Levels with Diabetic Retinopathy in Type 2 Diabetic Mellitus Patients

Objective. The aim of this study was to observe the association between the development of diabetic retinopathy (DR) in type 2 diabetes mellitus (T2DM) and the levels of glucose and lipid metabolism and serum uric acid (SUA) levels. Methods. A retrospective analysis was performed on 97 patients with T2DM who were admitted to our endocrinology department from June 2019 to April 2021 with complete data; the patients were divided into DR and no DR groups (NDR) according to the presence or absence of DR. Their clinical history and biochemical test indexes were collected, and the fundus was examined by fundus photography and the fundoscopic examination method, and the vascular diameter was measured by using a computer software. All clinical data, medical history, and biochemical test indexes were compared between the two groups, and logistic regression was used to analyze the risk factors of DR. Results. The duration of DM disease, fasting blood glucose (FBG), glycosylated hemoglobin, type A1C (HbA1c) levels, cholesterol (TC), triacylglycerol (TG), low-density lipoprotein cholesterol (LDL-C), and SUA levels were higher in the DR group than those in the NDR group, and the differences were significant ( $P<0.05$ ). The difference between the NDR group and the DR group in terms of gender, age, BMI, DBP, SBP, family history of DM, FINS, and HDL-C levels was not significant ( $P>0.05$ ). The results of multifactorial analysis showed that the four variables of DM duration, HbA1c, TG, and SUA were still risk factors for the development of DR ( $P<0.05$ ). Further receiver operating characteristic (ROC) analysis showed that the areas under the curves (AUCs) for the duration of DM disease, HbA1c, TG, and SUA to predict the occurrence of DR were 0.740 (95% CI 0.639–0.841), 0.767 (95% 0.672–0.862), 0.721 (95% CI 0.617–0.826), and 0.693 (95% CI 0.588∼0.797), respectively. Conclusion. The lesions of DR in T2DM patients have a close relationship with the course of DM, HbA1c, TG, and SUA, and the course of DM, HbA1c, TG, and SUA has a good predictive value for the occurrence of DR.

Uric Acid and Diabetic Retinopathy: A Systematic Review and Meta-Analysis

Background

The relationship between uric acid (UA) and diabetic retinopathy (DR) remains ambiguous, and the results of current studies on the UA levels in patients with DR are conflicting. A meta-analysis was performed to provide a better understanding of the relationship between UA levels and DR.

Methods

PubMed, Web of Science, Embase, and the Cochrane Library databases were searched until December 11, 2021 to identify eligible studies, that compared the UA levels of the case group (patients with DR) and control group (controls with diabetes and healthy participants). The weighted mean difference (WMD) with a 95% confidence interval (CI) was used to evaluate the difference in UA levels between the case and control groups.

Results

Twenty-one studies involving 4,340 patients with DR and 8,595 controls (8,029 controls with diabetes and 566 healthy participants) were included in this meta-analysis. We found that patients with DR had significantly higher UA levels than those in the controls with diabetes (WMD = 36.28; 95% CI: 15.68, 56.89; P < 0.001) and healthy participants (WMD = 70.80; 95% CI: 19.85, 121.75; P = 0.006). There was an obvious heterogeneity among the 21 studies (I² = 97%, P < 0.001). Subgroup analyses of different phases of DR showed that UA levels were significantly increased in participants with proliferative diabetic retinopathy (PDR) (WMD = 46.57; 95% CI: 28.51, 64.63; P < 0.001) than in controls with diabetes; however, the difference is not statistically significant when comparing UA levels in patients with non-proliferative diabetic retinopathy (NPDR) and controls with diabetes (WMD = 22.50; 95% CI: −6.07, 51.08; P = 0.120). In addition, UA levels were higher in participants with a body mass index (BMI) ≥25.0 kg/m² and over 15 years of diabetes. Univariate meta-regression analysis revealed that BMI (P = 0.007, Adj R² = 40.12%) and fasting blood glucose (FBG) (P = 0.040, Adj R² = 29.72%) contributed to between-study heterogeneity.

Conclusions

In conclusion, our study provides evidence that UA levels are higher in patients with DR than those in the controls, but this difference is not statistically significant in the early phases. UA might be a potential biomarker for identifying disease severity in patients with DR, rather than predicting the onset of DR among patients with diabetes. However, more prospective and high-quality clinical evidence is required to confirm these present findings.

Systematic Review Registration

https://www.crd.york.ac.uk/PROSPERO/display_record.php?RecordID=297708.

Prediction of Diabetes through Retinal Images Using Deep Neural Network

Microvascular problems of diabetes, such as diabetic retinopathy and macular edema, can be seen in the eye's retina, and the retinal images are being used to screen for and diagnose the illness manually. Using deep learning to automate this time-consuming process might be quite beneficial. In this paper, a deep neural network, i.e., convolutional neural network, has been proposed for predicting diabetes through retinal images. Before applying the deep neural network, the dataset is preprocessed and normalised for classification. Deep neural network is constructed by using 7 layers, 5 kernels, and ReLU activation function, and MaxPooling is implemented to combine important features. Finally, the model is implemented to classify whether the retinal image belongs to a diabetic or nondiabetic class. The parameters used for evaluating the model are accuracy, precision, recall, and F1 score. The implemented model has achieved a training accuracy of more than 95%, which is much better than the other states of the art algorithms.

Deep-Learning-Based Hemoglobin Concentration Prediction and Anemia Screening Using Ultra-Wide Field Fundus Images

Background: Anemia is the most common hematological disorder. The purpose of this study was to establish and validate a deep-learning model to predict Hgb concentrations and screen anemia using ultra-wide-field (UWF) fundus images.

Methods: The study was conducted at Peking Union Medical College Hospital. Optos color images taken between January 2017 and June 2021 were screened for building the dataset. ASModel_UWF using UWF images was developed. Mean absolute error (MAE) and area under the receiver operating characteristics curve (AUC) were used to evaluate its performance. Saliency maps were generated to make the visual explanation of the model.

Results: ASModel_UWF acquired the MAE of the prediction task of 0.83 g/dl (95%CI: 0.81–0.85 g/dl) and the AUC of the screening task of 0.93 (95%CI: 0.92–0.95). Compared with other screening approaches, it achieved the best performance of AUC and sensitivity when the test dataset size was larger than 1000. The model tended to focus on the area around the optic disc, retinal vessels, and some regions located at the peripheral area of the retina, which were undetected by non-UWF imaging.

Conclusion: The deep-learning model ASModel_UWF could both predict Hgb concentration and screen anemia in a non-invasive and accurate way with high efficiency.