Evaluation of a New Neural Network Classifier for Diabetic Retinopathy.

BACKGROUND: Medical image segmentation is a well-studied subject within the field of image processing. The goal of this research is to create an AI retinal screening grading system that is both accurate and fast. We introduce a new segmentation network which achieves state-of-the-art results on semantic segmentation of color fundus photographs. By applying the net-work to identify anatomical markers of diabetic retinopathy (DR) and diabetic macular edema (DME), we collect sufficient information to classify patients by grades R0 and R1 or above, M0 and M1. METHODS: The AI grading system was trained on screening data to evaluate the presence of DR and DME. The core algorithm of the system is a deep learning network that segments relevant anatomical features in a retinal image. Patients were graded according to the standard NHS Diabetic Eye Screening Program feature-based grading protocol. RESULTS: The algorithm performance was evaluated with a series of 6,981 patient retinal images from routine diabetic eye screenings. It correctly predicted 98.9% of retinopathy events and 95.5% of maculopathy events. Non-disease events prediction rate was 68.6% for retinopathy and 81.2% for maculopathy. CONCLUSION: This novel deep learning model was trained and tested on patient data from annual diabetic retinopathy screenings can classify with high accuracy the DR and DME status of a person with diabetes. The system can be easily reconfigured according to any grading protocol, without running a long AI training procedure. The incorporation of the AI grading system can increase the graders' productivity and improve the final outcome accuracy of the screening process.

Hyperbaric oxygen treatment for non-arteritic central retinal artery occlusion retrospective comparative analysis from two tertiary medical centres.

OBJECTIVES: To compare the visual outcome of patients treated for non-arthritic central retinal artery occlusion (CRAO) in a medical centre that uses hyperbaric oxygen therapy (HBOT) as part of the standard of care (SOC) to medical centres that does not. METHODS: The study included data from two tertiary medical centres. The medical records of all patients diagnosed with non-arthritic CRAO without a patent cilioretinal artery between January 2010 and December 2018 in two tertiary medical centres were reviewed. RESULTS: One hundred and twenty-one patients were treated by HBOT and 23 patients received only SOC. In the HBOT group, best-corrected visual acuity (BCVA) improved from 2.89 ± 0.98 logMAR at presentation to 2.15 ± 1.07 logMAR upon the end of HBOT (P < 0.001), while the SOC group had no significant improvement, from 3.04 ± 0.82 logMAR at presentation to 2.80 ± 1.50 logMAR (P = 0.24). With adjustment for age, gender, and the duration of symptoms, final BCVA in the HBOT group was significantly better compared to the control group (P = 0.023). Rates of patients achieving vision of 20/200 or better were similar between groups (17.4% vs. 19.8%, P = 0.523). CONCLUSION: Utilizing HBOT as part of the SOC for CRAO improves the final visual outcome. HBOT is safe and can be implemented, if available, as part of SOC in all tertiary medical centres.

Risk factors for an atherothrombotic event in patients with diabetic macular edema treated with intravitreal injections of bevacizumab.

AIM: To identify risk factors for an atherothrombotic event (ATE) among patients who were treated for diabetic macular edema (DME) with intravitreal bevacizumab injections. METHODS: This retrospective study enrolled all consecutive patients with DME who were treated by intravitreal bevacizumab from 2009 through 2016 in a single center. They were divided into one group treated by bevacizumab and subsequently had an ATE and a second group also treated by bevacizumab and did not have an ATE. RESULTS: A total of 455 patients with DME were enrolled. Seventy-two of the patients had an ATE. A multivariate model adjusted for age, gender, smoking, body mass index, hemoglobin A1c (HbA1c), duration of diabetes, creatinine, and blood pressure revealed an increased risk for ATE in the patients with diabetic duration of more than 13y, a systolic blood pressure over 153.5 mm Hg at first treatment, or having been treated by more than 4 intravitreal bevacizumab injections. Additionally, patients that had an ATE within 3mo from the last intravitreal treatment underwent more bevacizumab injections (5.2±3.4 vs 3.07±1.86; P<0.001). CONCLUSION: The risk factors for an ATE identified in this study are systolic blood pressure >153.5 mm Hg, a history of diabetic mellitus for more than 13y, and treatment with more than 4 intravitreal bevacizumab injections. These factors need to be borne in mind when bevacizumab is being considered in the management of patients with DME.

The multiple mini-interviews as a predictor of peer evaluations during clinical training in medical school.

BACKGROUND: Multiple Mini-Interviews (MMI) is an admissions tool being used for more than a decade by medical schools worldwide to assess non-cognitive skills. In Israel, in addition to the MMI tool, two questionnaires are given to the candidates. It has been necessary to find a relevant criterion to examine the predictive validity of the MMI test. The measure developed in this study was a peer assessment tool which was filled out during clerkship. AIMS: The study aims at evaluating the predictive validity of the MMI and two questionnaires when compared with the peer assessment tool. METHOD: Ninety nine fourth-year medical students were included in this study. All of them went through the MMI test before the beginning of their studies. The students participated in two clinical rotations and were divided into 10 groups, each consisting of 9-12 students. This study is based on assessments given to 99 students. Every member of a group evaluated his fellow members on a number of attribute. The questionnaire included 10 statements, related to a particular attributes. In addition two concluding questions were included. Pearson correlations were used to test the relationships between pre-admission variables (MMI and two questionnaires) and the peer assessments. RESULTS: Statistically significant, moderate correlations between the general MOR and MIRKAM scores and the sum of the questionnaire assessments were 0.39 and 0.37, respectively (P-value <0.05). CONCLUSIONS: The study provides important evidence for the validity of the MMI. Additional studies are required to reevaluate the predicted validity of the MMI.