Chuying Shi

Objectives: Train an automatic retinal image analysis (ARIA) method to screen glaucomatous optic neuropathy (GON) on non-mydriatic retinal images labelled with the additional results of optical coherence tomography (OCT) and assess different models for the GON classification.Methods: All the images were obtained from the hospital for training and 10-fold cross-validation. Two methods were used to improve the classification performance: (1) using images labelled with the additional results of OCT as the reference standard and (2) generating models using retinal features from the entire images, the region of interest (ROI) of the optic disc, and the ROI of the macula, and the combination of all the features.Results: Overall, we collected 1338 images with paired OCT scans. In 10-fold validation, ARIA achieved sensitivities of 92.2 %, 92.7% and 85.7%, specificities of 88.8%, 86.7% and 80.2% and accuracies of 90.6%, 89.9% and 83.1% using the retinal features from the entire images, the ROI of the optic disc and the ROI of the macula, respectively. We found the model combining all the features has the best classification performance and obtained a sensitivity of 92.5%, a specificity of 92.1% and an accuracy of 92.4%, which is significantly different from other models (p<0.001).Conclusion: We used two methods to improve the classification performance and found the best model to detect glaucoma on colour fundus retinal images. It can become a cost-effective and relatively more accurate glaucoma screening tool than conventional methods.

We appreciate the interest of Sun et al. [1] in our article [2]. They suggested in their Letter to the Editor that we could provide detailed information on sedatives and analgesics administered to patients and perform subgroup analysis based on these confounding factors. There were eight studies included in our meta-analysis. Sharma et al. administered small doses of intravenous agents (midazolam, propofol, and fentanyl) to all individual patients who needed to achieve adequate relaxation on the day after the surgery [3]. Hari-Kovacs et al. routinely used 10 mg/os temazepam on the day ward [4]. Zhang et al. administered pranoprofen eye drops to patients in the treatment group only before their second-eye surgery, but did not mention any sedatives/analgesics in the control group [5]. One study did not describe the use of sedatives/analgesics in detail, and sedatives/analgesics were not routinely administered to all eyes in first-eye and second-eye surgery [6]. The remaining four of the eight studies used neither sedatives nor analgesics. In total, Sharma et al. and Hari-Kovacs et al. administered sedatives or analgesics in both first-eye and second-eye surgery. However, subgroup analysis cannot be performed because these two studies assessed pain on different days—one was on the day of surgery and another was on the day after surgery. With high levels of heterogeneity, meta-analyses would have low predictive values no matter what the I² values are [7]. Therefore, we only did a subgroup analysis on the four studies that did not use any sedatives/analgesics. The result (WMD: 0.62; 95% CI: 0.53, 0.71; ; I² = 0%) showed that the pain scores of the first eye on the day of the surgery were significantly lower compared to those of the second eye when no sedatives/analgesics were used, which was consistent with our original result. A possible pharmacological explanation of greater eye pain in second-eye surgery is that the previous exposure to analgesic and sedative medications during the first-eye surgery may cause drug tolerance so that the response to the same medications was decreased during the second-eye surgery. However, this seems unlikely according to the result of this subgroup analysis where no sedatives/analgesics were used. The meta-analysis of using sedatives/analgesics therefore requires more studies to further investigate.