Diabetic retinopathy classification for supervised machine learning algorithms

Inter-rater reliability in labeling quality and pathological features of retinal OCT scans: A customized annotation software approach

OBJECTIVES

Various imaging features on optical coherence tomography (OCT) are crucial for identifying and defining disease progression. Establishing a consensus on these imaging features is essential, particularly for training deep learning models for disease classification. This study aims to analyze the inter-rater reliability in labeling the quality and common imaging signatures of retinal OCT scans.

METHODS

500 OCT scans obtained from CIRRUS HD-OCT 5000 devices were displayed at 512x1024x128 resolution on a customizable, in-house annotation software. Each patient's eye was represented by 16 random scans. Two masked reviewers independently labeled the quality and specific pathological features of each scan. Evaluated features included overall image quality, presence of fovea, and disease signatures including subretinal fluid (SRF), intraretinal fluid (IRF), drusen, pigment epithelial detachment (PED), and hyperreflective material. The raw percentage agreement and Cohen's kappa (κ) coefficient were used to evaluate concurrence between the two sets of labels.

RESULTS

Our analysis revealed κ = 0.60 for the inter-rater reliability of overall scan quality, indicating substantial agreement. In contrast, there was slight agreement in determining the cause of poor image quality (κ = 0.18). The binary determination of presence and absence of retinal disease signatures showed almost complete agreement between reviewers (κ = 0.85). Specific retinal pathologies, such as the foveal location of the scan (0.78), IRF (0.63), drusen (0.73), and PED (0.87), exhibited substantial concordance. However, less agreement was found in identifying SRF (0.52), hyperreflective dots (0.41), and hyperreflective foci (0.33).

CONCLUSIONS

Our study demonstrates significant inter-rater reliability in labeling the quality and retinal pathologies on OCT scans. While some features show stronger agreement than others, these standardized labels can be utilized to create automated machine learning tools for diagnosing retinal diseases and capturing valuable pathological features in each scan. This standardization will aid in the consistency of medical diagnoses and enhance the accessibility of OCT diagnostic tools.

A Beginner's Guide to Artificial Intelligence for Ophthalmologists

The integration of artificial intelligence (AI) in ophthalmology has promoted the development of the discipline, offering opportunities for enhancing diagnostic accuracy, patient care, and treatment outcomes. This paper aims to provide a foundational understanding of AI applications in ophthalmology, with a focus on interpreting studies related to AI-driven diagnostics. The core of our discussion is to explore various AI methods, including deep learning (DL) frameworks for detecting and quantifying ophthalmic features in imaging data, as well as using transfer learning for effective model training in limited datasets. The paper highlights the importance of high-quality, diverse datasets for training AI models and the need for transparent reporting of methodologies to ensure reproducibility and reliability in AI studies. Furthermore, we address the clinical implications of AI diagnostics, emphasizing the balance between minimizing false negatives to avoid missed diagnoses and reducing false positives to prevent unnecessary interventions. The paper also discusses the ethical considerations and potential biases in AI models, underscoring the importance of continuous monitoring and improvement of AI systems in clinical settings. In conclusion, this paper serves as a primer for ophthalmologists seeking to understand the basics of AI in their field, guiding them through the critical aspects of interpreting AI studies and the practical considerations for integrating AI into clinical practice.

Association between obstructive coronary disease and diabetic retinopathy: Cross-sectional study of coronary angiotomography and multimodal retinal imaging

AIMS

To investigate the association between diabetic retinopathy (DR) and coronary artery disease (CAD) using coronary angiotomography (CCTA) and multimodal retinal imaging (MMRI) with ultra-widefield retinography and optical coherence tomography angiography and structural domain.

METHODS

Single-center, cross-sectional, single-blind. Patients with diabetes who had undergone CCTA underwent MMRI. Uni and multivariate analysis were used to assess the association between CAD and DR and to identify variables independently associated with DR.

RESULTS

We included 171 patients, 87 CAD and 84 non-CAD. Most CAD patients were males (74 % vs 38 %, P < 0.01), insulin users (52 % vs 38 %, p < 0.01) and revascularized (64 %). They had a higher prevalence of DR (48 % vs 22 %, p = 0.01), microaneurysms (25 % vs 13 %, p = 0.04), intraretinal cysts (22 % vs 8 %, p = 0.01) and areas of reduced capillary density (46 % vs 20 %, p < 0.01). CAD patients also had lower mean vascular density (MVD) (15.7 % vs 16.5,%, p = 0.049) and foveal avascular zone (FAZ) circularity (0.64 ± 0.1 vs 0.69 ± 0.1, p = 0.04). There were significant and negative correlations between Duke coronary score and MVD (r = -0.189; p = 0.03) and FAZ circularity (r = -0,206; p = 0.02). CAD, DM duration and insulin use independently associated with DR.

CONCLUSIONS

CAD patients had higher prevalence of DR and lower MVD. CAD, DM duration and insulin use were independently associated with DR.

Developing and evaluating the usability of a web-based Diabetic Retinopathy Health Education Program for patients with type 2 diabetes mellitus

OBJECTIVES

To describe the development of a webpage based on the Intervention Mapping (IM) protocol and usability testing of the Diabetic Retinopathy Health Education Profram (DRHEP).

METHODS

The mixed methods pilot feasibility study was carried out between April and September 2021, involving 16 patients with type 2 diabetes mellitus and 5 experts. The usability score was rated according to the System Usability Scale (SUS).

RESULTS

The average SUS score by the experts was 88. The patients gave a higher score of 85 for SUS, with 58 as the lowest. The average SUS score was 72. The findings indicate that the webpage is acceptable, good, and highly usable for users.

CONCLUSION

The outcomes of this study signify the relationship between effective health applications and how their design might hamper their effectiveness in changing patients' behavior.

In Vitro Evaluation of Apoptosis, Inflammation, Angiogenesis, and Neuroprotection Gene Expression in Retinal Pigmented Epithelial Cell Treated with Interferon α-2b

Angiogenesis, retinal neuropathy, and inflammation are the main molecular features of diabetic retinopathy (DR) and should be taken into consideration for potential treatment approaches. Retinal pigmented epithelial (RPE) cells play a major role in DR progression. This study evaluated the in vitro effect of interferon (IFN) α-2b on the expression of genes involved in apoptosis, inflammation, neuroprotection, and angiogenesis in RPE cells. RPE cells were cocultured with IFN α-2b at 2 doses (500 and 1,000 IU) and treatment periods (24 and 48 h). The quantitative relative expression of genes (BCL-2, BAX, BDNF, VEGF, and IL-1b) was evaluated in the treated versus control cells through real-time polymerase chain reaction (PCR). The result of this study demonstrated that IFN treatment at 1,000 IU (48 h) led to significant upregulation of BCL-2, BAX, BDNF, and IL-1b; however, the BCL-2/BAX ratio was not statistically altered from 1:1, in any of the treatment patterns. We also showed that VEGF expression was downregulated in RPE cells treated with 500 IU for 24 h. It can be concluded that IFN α-2b was safe (BCL-2/BAX ∼1:1) and enhanced neuroprotection at 1,000 IU (48 h); however-at the same time-IFN α-2b induced inflammation in RPE cells. Moreover, the antiangiogenic effect of IFN α-2b was solely observed in RPE cells treated with 500 IU (24 h). It seems that IFN α-2b in lower doses and short duration exerts antiangiogenic effects and in higher doses and longer duration has neuroprotective and inflammatory effects. Hence, appropriate concentration and duration of treatment, according to the type and stage of the disease, should be considered to achieve success in IFN therapy.

Pneumatic vitreolysis versus vitrectomy for the treatment of vitreomacular traction syndrome and macular holes: complication analysis and systematic review with meta-analysis of functional outcomes

BACKGROUND

We conducted a systematic review to compare  the effects of pneumatic vitreolysis (PV), enzymatic vitreolysis (EVL) with ocriplasmin, and pars plana vitrectomy (PPV) on vitreomacular traction (VMT) syndrome and macular holes (MHs) to assess their efficacy as treatment options.

METHODS

Databases, including PubMed, ClinicalTrials.gov ( www.

CLINICALTRIALS

gov ), the Cochrane Central Register of Controlled Trials (CENTRAL)-including the Cochrane Eyes and Vision Group Trials Register (The Cochrane Library 2013, Issue 2)-, Ovid MEDLINE, and EMBASE (January 2000-October 2022), were searched to identify studies comparing the outcomes of PV versus PPV, PPV versus ocriplasmin and ocriplasmin versus PV. RevMan 5.1 was used for the meta-analysis of the studies.

RESULTS

Among the 89 studies, 79 were considered eligible for qualitative analysis, and 10 quantitative studies were subjected to meta-analysis. PPV resulted in better postoperative visual acuity improvement than ocriplasmin (standardized mean deviation (SMD) = 0.38, 95% CI 0.03-0.73, p = 0.0003). PV resulted in no significant difference in visual improvement compared  with  PPV (SMD = - 0.15, 95% CI - 0.47 to 0.16, p = 0.35). PPV was significantly more effective in terms of the VMT release rate (risk ratio = 0.48, 95% CI 0.38-0.62, p = 0.00001) and MH closure rate (risk ratio = 0.49, 95% CI 0.30-0.81, p = 0.006) than ocriplasmin. PV was more effective than ocriplasmin in terms of the VMT release rate (risk ratio = 0.49, 95% CI 0.35-0.70, p = 0.0001). Qualitative analysis showed MH closure rates of 46%, 47.8%, and 95% and VMT releases rates of 46%, 68% and 100% after ocriplasmin, PV, and PPV treatments, respectively.  Adverse events and postoperative complications occurring after treatment have also been documented in these studies.

CONCLUSION

PPV appears to be the most promising option for MH closure and VMT release, with fewer serious complications than EVL  or PV. However, given the limited number of studies comparing these treatments, further research is needed to establish the superiority of PPV over the other options.

Maximum home blood pressure and cardiovascular outcomes in patients with type 2 diabetes: KAMOGAWA-HBP study

OBJECTIVE

The maximum blood pressure was reported as a possible marker of organ damage. We previously showed that maximum home blood pressure was significantly associated with development of diabetic nephropathy. In the same cohort of patients with diabetes as in the previous study, this study aimed to evaluate the prognostic blood pressure values for the onset of first cardiovascular events.

METHODS

This retrospective cohort study included 1082 patients with type 2 diabetes (47.0% female, median age 65.0) without a history of macrovascular complications. Blood pressure measurements were performed in triplicates every morning and evening for 14 consecutive days from the start of the study. Cox hazards model was used to evaluate the risk of primary endpoint, which was defined as the onset of first major cardiovascular event.

RESULTS

The primary endpoint occurred in 119 patients (incidence rate, 15.7/1000 person-years) during an average of 7.0-year follow-up. The adjusted hazard ratios (95% confidence interval [CI]) of maximum morning systolic blood pressure (SBP) and maximum evening SBP for cardiovascular events were 1.12 (1.01-1.24) and 1.19 (1.07-1.31), respectively, adjusted by sex, duration of diabetes, body mass index, hemoglobin A1c, low density lipoprotein cholesterol, smoking status, and use of antihypertensive medications. The cutoff values of maximum blood pressure for the events were 150 mmHg in the morning (hazard ratio, 1.73; 95% CI, 1.07-2.81) and 157 mmHg in the evening (hazard ratio, 2.30; 95% CI, 1.46-3.61), using the Youden's index.

CONCLUSION

Maximum home blood pressure is a predictor of subsequent cardiovascular events in patients with type 2 diabetes.

Impact of sustained hypertension on new cardiovascular events in patients with type 2 diabetes: KAMOGAWA-HBP study

We have previously shown that masked hypertension (MH) and sustained hypertension (SH) contribute to the progression of diabetic nephropathy. Although the risk of target organ damage and cardiovascular events in MH and SH is significantly higher than that in normotension and white coat hypertension, the role of MH or SH in cardiovascular events has never been reported in studies specific to diabetic patients. Therefore, in this study, we aimed to determine whether blood pressure control status contributes to the development of new cardiovascular events. A longitudinal study of 1082 patients with type 2 diabetes mellitus and no history of cardiovascular events was conducted. Patients were instructed to have their blood pressure measured three times, every morning and evening, for 14 consecutive days. Hypertension status was classified into four groups based on the systolic blood pressure measurements in the clinic and at home. The primary endpoint was the first cardiovascular event. After a median follow-up of 7.0 (interquartile range, 4.0-9.0) years, 119 patients developed cardiovascular events. The hazard ratio (95% confidence interval) for the risk of developing cardiovascular events was significantly higher in the SH group than in the controlled blood pressure group (1.63 [1.02-2.59]). SH is a useful predictor of cardiovascular events. Both at home and in the clinic, blood pressure monitoring should be assessed in routine clinical practice to predict future cardiovascular events in patients with type 2 diabetes.