Fundus Photography as a Screening Method for Diabetic Retinopathy in Children With Type 1 Diabetes: Outcome of the Initial Photography

DEC-DRR: deep ensemble of classification model for diabetic retinopathy recognition

Most diabetes patients are liable to have diabetic retinopathy (DR); however, the majority of them might not be even aware of the ailment. Therefore, early detection and treatment of DR are necessary to prevent vision loss. But, avoiding DR is not a simple process. An ophthalmologist can typically identify DR through an optical evaluation of the fundus and through the evaluation of color pictures. However, due to the increased count of DR patients, this could not be possible as it consumes more time. To rectify this problem, a novel deep ensemble-based DR classification technique is developed in this work. Initially, a Wiener filter (WF) is applied for preprocessing the image. Then, the enhanced U-Net-based segmentation process is done. Subsequent to the segmentation process, features are extracted that include statistical features, inferior superior nasal temporal (ISNT), cup to disc ratio (CDR), and improved LGBP as well. Further, deep ensemble classifiers (DEC) like CNN, Bi-GRU, and DMN are used to recognize the disease. The outcomes from DMN, CNN, and Bi-GRU are then subjected to improved SLF. Additionally, the weights of DMN, CNN, and Bi-GRU are adjusted via pelican updated Tasmanian devil optimization (PU-TDO). Finally, outputs on DR (microaneurysms, hemorrhages, hard exudates, and soft exudates) are obtained. The performance of DEC + PU-TDO for diabetic retinopathy is computed over extant models with regard to different measures for four datasets. The results on accuracy using the DEC + PU-TDO scheme for the IDRID dataset is maximum around 0.975 at 90th LP while other models have less accuracy. The FPR of DEC + PU-TDO is less around 0.039 at the 90th LP for the SUSTech-SYSU dataset, while other extant models have maximum FPR.

Changes in early optical coherence tomography angiography among children and adolescents with type 1 diabetes: Relation to fibroblast growth factor 21

AIMS

Current diagnostic and treatment modalities target late stages of diabetic retinopathy (DR) when retinopathy has already been established. Novel and more sensitive strategies are needed. Optical coherence tomography angiography (OCTA) permits non-invasive visualisation of retinal microcirculation. Fibroblast growth factor-21 (FGF21) plays an important role in glucose and lipid homoeostasis. This study assesses early OCTA changes among children and adolescents with type 1 diabetes (T1DM) compared to fundus photography and correlates them to diabetes-duration, glycaemic control, and FGF21; hence, it determines their value in early detection of DR.

METHODOLOGY

Hundred children and adolescents with T1DM were assessed for diabetes-duration, insulin therapy, hypoglycemia, and diabetic-ketoacidosis frequency, Tanner staging, glycated-haemoglobin (HbA1c), fasting lipids, urinary albumin/creatinine ratio, and serum FGF21. OCTA and fundus photography were done for the studied patients and 100 age, gender, and Tanner matched healthy controls.

RESULTS

The mean age of the children and adolescents with T1DM was 10.84 years, their mean diabetes-duration was 3.27 years and their median FGF21 was 150 pg/ml. FGF21 was significantly higher among children and adolescents with T1DM than controls (p < 0.001). Children and adolescents with T1DM had a significantly larger foveal avascular zone (FAZ) and lower peripapillary and inside-disc capillary densities (p < 0.05); with no significant fundus photography difference (p = 0.155) than controls. FAZ was positively correlated and peripapillary and inside-disc capillary densities were negatively correlated with diabetes-duration, HbA1c, FGF21, and Tanner stage. FGF21 was significantly higher in T1DM children and adolescents having OCTA changes compared to those with normal OCTA (p = 0.002). Multivariate-regression revealed that FAZ is independently associated with diabetes-duration, HbA1c and FGF21.

CONCLUSIONS

OCTA changes start early in children and adolescents with T1DM long before the fundus changes. These changes are correlated with diabetes-duration, puberty, glycaemic, and FGF21.

Application Research of Artificial Intelligence Screening System for Diabetic Retinopathy

According to the latest data from the Bureau of Disease Control and Prevention of the National Health and Family Planning Commission, China currently has 199.6 million diabetic patients and has become the world's largest country with diabetes. The prevalence rate is as high as 14.3%, which is much higher than the world average of 5.8%. The primary-level ophthalmic screening service is one of the important tasks to improve primary-level medical services, and the corresponding ophthalmic imaging diagnosis technology is an important support for primary-level medical and health services. Therefore, it is very necessary for us to study the application of artificial intelligence image recognition technology for diabetic retinopathy under the medical consortium mode and to study the precise initial diagnosis, precise referral, and precise follow-up of diabetic retina under the medical conjoined mode, so as to better promote the transformation of the ophthalmology primary service model. Based on this background, in this article, we have proposed and carried out the following solution: (1) diabetes data collation. Based on medical artificial intelligence technology, this paper collected 2,265 electronic medical records from an eye hospital in Ningbo and selected 2,000 qualified medical records for data integration and preprocessing. The contents of electronic medical records mainly include age, gender, and examination records. (2) Establish diabetic retinopathy diagnosis model based on neural network algorithm. This article first uses the classic algorithm of BP neural network for modeling, chooses the Levenberg-Marquardt method as the training function, and selects 10 hidden layer units through comparison experiments. After that, ophthalmologists assessed 80 sets of test results and determined the right diagnosis rate. Finally, this article compares and analyzes the accuracy of the two routes in 80 tests.

Clinical and Demographic Factors Associated With Diabetic Retinopathy Among Young Patients With Diabetes

IMPORTANCE

Diabetic retinopathy (DR) is a leading cause of vision loss worldwide. As the incidence of both type 1 and type 2 diabetes among youths continues to increase around the world, understanding the factors associated with the development of DR in this age group is important.

OBJECTIVE

To identify factors associated with DR among children, adolescents, and young adults with type 1 or type 2 diabetes in the US.

DESIGN, SETTING, AND PARTICIPANTS

This cross-sectional study pooled data from 2 large academic pediatric centers in the US (Baylor College of Medicine/Texas Children's Hospital [BCM/TCH] Diabetes and Endocrine Care Center and Johns Hopkins University [JHU] Pediatric Diabetes Center) to form a diverse population for analysis. Data were collected prospectively at the JHU center (via point-of-care screening using fundus photography) from December 3, 2018, to November 1, 2019, and retrospectively at the BCM/TCH center (via electronic health records of patients who received point-of-care screening using retinal cameras between June 1, 2016, and May 31, 2019). A total of 1640 individuals aged 5 to 21 years with type 1 or type 2 diabetes (308 participants from the JHU center and 1332 participants from the BCM/TCH center) completed DR screening and had gradable images.

MAIN OUTCOME AND MEASURES

Prevalence of DR, as identified on fundus photography, and factors associated with DR.

RESULTS

Among 1640 participants (mean [SD] age, 15.7 [3.6] years; 867 female individuals [52.9%]), 1216 (74.1%) had type 1 diabetes, and 416 (25.4%) had type 2 diabetes. A total of 506 participants (30.9%) were Hispanic, 384 (23.4%) were non-Hispanic Black or African American, 647 (39.5%) were non-Hispanic White, and 103 (6.3%) were of other races or ethnicities (1 was American Indian or Alaska Native, 50 were Asian, 1 was Native Hawaiian or Pacific Islander, and 51 did not specify race or ethnicity, specified other race or ethnicity, or had unavailable data on race or ethnicity). Overall, 558 of 1216 patients (45.9%) with type 1 diabetes used an insulin pump, and 5 of 416 patients (1.2%) with type 2 diabetes used an insulin pump. Diabetic retinopathy was found in 57 of 1640 patients (3.5%). Patients with DR vs without DR had a greater duration of diabetes (mean [SD], 9.4 [4.4] years vs 6.6 [4.4] years; P < .001) and higher hemoglobin A1c (HbA1c) levels (mean [SD], 10.3% [2.4%] vs 9.2% [2.1%]; P < .001). Among those with type 1 diabetes, insulin pump use was associated with a lower likelihood of DR after adjusting for race and ethnicity, insurance status, diabetes duration, and HbA1c level (odds ratio [OR], 0.43; 95% CI, 0.20-0.93; P = .03). The likelihood of having DR was 2.1 times higher among Black or African American participants compared with White participants (OR, 2.12; 95% CI, 1.12-4.01; P = .02); this difference was no longer significant after adjusting for duration of diabetes, insurance status, insulin pump use (among patients with type 1 diabetes only), and mean HbA1c level (type 1 diabetes: OR, 1.79; 95% CI, 0.83-3.89; P = .14; type 2 diabetes: OR, 1.08; 95% CI, 0.30-3.85; P = .91).

CONCLUSIONS AND RELEVANCE

This study found that although the duration of diabetes and suboptimal glycemic control have long been associated with DR, insulin pump use (among those with type 1 diabetes) was independently associated with a lower likelihood of DR, which is likely owing to decreased glycemic variability and increased time in range (ie, the percentage of time blood glucose levels remain within the 70-180 mg/dL range). Black or African American race was found to be associated with DR in the univariable analysis but not in the multivariable analysis, which may represent disparities in access to diabetes technologies and care.

Diabetic retinopathy detection and classification using capsule networks

Nowadays, diabetic retinopathy is a prominent reason for blindness among the people who suffer from diabetes. Early and timely detection of this problem is critical for a good prognosis. An automated system for this purpose contains several phases like identification and classification of lesions in fundus images. Machine learning techniques based on manual extraction of features and automatic extraction of features with convolution neural network have been presented for diabetic retinopathy detection. The recent developments like capsule networks in deep learning and their significant success over traditional machine learning methods for a variety of applications inspired the researchers to apply them for diabetic retinopathy diagnosis. In this paper, a reformed capsule network is developed for the detection and classification of diabetic retinopathy. Using the convolution and primary capsule layer, the features are extracted from the fundus images and then using the class capsule layer and softmax layer the probability that the image belongs to a specific class is estimated. The efficiency of the proposed reformed network is validated concerning four performance measures by considering the Messidor dataset. The constructed capsule network attains an accuracy of 97.98%, 97.65%, 97.65%, and 98.64% on the healthy retina, stage 1, stage 2, and stage 3 fundus images.

Real-World Screening for Retinopathy in Youth With Type 1 Diabetes Using a Nonmydriatic Fundus Camera

OBJECTIVE

To assess the use of a portable retinal camera in diabetic retinopathy (DR) screening in multiple settings and the presence of associated risk factors among children, adolescents, and young adults with type 1 diabetes.

DESIGN AND METHODS

Five hundred youth with type 1 diabetes of at least 1 year's duration were recruited from clinics, diabetes camp, and a diabetes conference and underwent retinal imaging using a nonmydriatic fundus camera. Retinal characterization was performed remotely by a licensed ophthalmologist. Risk factors for DR development were evaluated by a patient-reported questionnaire and medical chart review.

RESULTS

Of the 500 recruited subjects aged 9-26 years (mean 14.9, SD 3.8), 10 cases of DR were identified (nine mild and one moderate nonproliferative DR) with 100% of images of gradable quality. The prevalence of DR was 2.04% (95% CI 0.78-3.29), at an average age of 20.2 years, with the youngest affected subject being 17.1 years of age. The rate of DR was higher, at 6.5%, with diabetes duration >10 years (95% CI 0.86-12.12, P = 0.0002). In subjects with DR, the average duration of diabetes was 12.1 years (SD 4.6, range 6.2-20.0), and in a subgroup of clinic-only subjects (n = 114), elevated blood pressure in the year before screening was associated with DR (P = 0.0068).

CONCLUSION

This study in a large cohort of subjects with type 1 diabetes demonstrates that older adolescents and young adults (>17 years) with longer disease duration (>6 years) are at risk for DR development, and screening using a portable retinal camera is feasible in clinics and other locations. Recent elevated blood pressure was a risk factor in an analyzed subgroup.

Evaluation of retinal neurovascular structures by optical coherence tomography and optical coherence tomography angiography in children and adolescents with type 1 diabetes mellitus without clinical sign of diabetic retinopathy

BACKGROUND

The aim of the study was to investigate whether retinal neurovascular structural impairment in children and adolescents with type 1 diabetes mellitus (T1D) without clinical signs of diabetic retinopathy (DR) could be detected early via optical coherence tomography (OCT) and OCT angiography (OCTA).

METHODS

In the current prospective, cross-sectional, observational clinical study children and adolescents with T1D without DR were evaluated between December 2018 and May 2019. Retinal neurovascular structures in the macular and optic disc regions were examined in detail and quantitatively assessed using OCT and OCTA. Data from subjects with T1D were compared with data from healthy controls. Whether retinal neurovascular structural changes were significantly associated with puberty stage, diabetes duration, and HbA1c level was also investigated.

RESULTS

The T1D group included 110 eyes and the control group included 84 eyes. In the T1D group the mean inside disc vessel density (VD) was significantly lower than that of the control group (p < 0.001), as was the mean superior temporal disc VD (p < 0.043). Puberty stage was significantly associated with retinal thickness, parafoveal superficial capillary plexus VD, and peripapillary retinal nerve fiber layer thickness (p < 0.05). Diabetes duration and HbA1c level was significantly correlated with retinal layer thickness, foveal avascular zone diameter, and superficial and deep capillary plexus VDs.

CONCLUSION

In children and adolescents with T1D without clinical signs of DR, the VD of the disc region is affected earlier than the macular region. In these patients, early neurovascular impairment can be detected non-invasively via OCT and OCTA.

Changes in Retinal Microcirculation Precede the Clinical Onset of Diabetic Retinopathy in Children With Type 1 Diabetes Mellitus

PURPOSE

To investigate whether abnormal glucose metabolism in diabetes mellitus (DM) affects the retinal microcirculation of children with well-controlled type 1 DM and to compare these results with those obtained from healthy children.

DESIGN

Cross-sectional prospective study.

METHODS

This study enrolled 60 patients with DM without clinically detectable diabetic retinopathy (DR) and 57 age-matched control subjects. Optical coherence tomography angiography (OCT-A) was performed using AngioVue (Avanti, Optivue). Foveal avascular zone (FAZ) area, nonflow area, superficial and deep vessel densities, FAZ perimeter, acircularity index of FAZ (AI; the ratio of the perimeter of FAZ and the perimeter of a circle with equal area), and foveal density (FD-300; vessel density in 300 μm around FAZ) were analyzed. Correlations between the investigated OCT-A parameters with DM duration and glycated hemoglobin (HbA1c) levels were evaluated among patients with type 1 DM.

RESULTS

Differences in the mean values for FAZ perimeter, AI, and FD-300 were statistically significant between DM group and control group (P < .001, P = .001, and P = .009, respectively). There were also statistically significant differences between the groups for vessel densities of deep superior hemi-parafovea, deep temporal parafovea, and deep superior parafoveal zones (P = .008, P = .015, and P = .005, respectively). There were no significant correlations between DM duration and HbA1c levels with the investigated OCT-A parameters.

CONCLUSION

Diabetic eyes without clinically detectable DR exhibited alterations in FD-300, AI, perimeter, and vessel density of parafoveal capillaries in deep capillary plexus preceding the enlargement of FAZ; therefore, these new parameters might be sensitive imaging biomarkers to define early DR.

Diabetes technology: improving care, improving patient-reported outcomes and preventing complications in young people with Type 1 diabetes

With the evolution of diabetes technology, those living with Type 1 diabetes are given a wider arsenal of tools with which to achieve glycaemic control and improve patient-reported outcomes. Furthermore, the use of these technologies may help reduce the risk of acute complications, such as severe hypoglycaemia and diabetic ketoacidosis, as well as long-term macro- and microvascular complications. In addition, diabetes technology can have a beneficial impact on psychosocial health by reducing the burden of diabetes. Unfortunately, diabetes goals are often unmet and people with Type 1 diabetes too frequently experience acute and long-term complications of this condition, in addition to often having less than ideal psychosocial outcomes. Increasing realization of the importance of patient-reported outcomes is leading to diabetes care delivery becoming more patient-centred. Diabetes technology in the form of medical devices, digital health and big data analytics have the potential to improve clinical care and psychosocial support, resulting in lower rates of acute and chronic complications, decreased burden of diabetes care, and improved quality of life.