The present study aimed to conduct a real-time automatic analysis of two important surgical phases, which are continuous curvilinear capsulorrhexis (CCC), nuclear extraction, and three other surgical phases of cataract surgery using artificial intelligence technology. A total of 303 cases of cataract surgery registered in the clinical database of the Ophthalmology Department of Tsukazaki Hospital were used as a dataset. Surgical videos were downsampled to a resolution of 299 × 168 at 1 FPS to image each frame. Next, based on the start and end times of each surgical phase recorded by an ophthalmologist, the obtained images were labeled correctly. Using the data, a neural network model, known as InceptionV3, was developed to identify the given surgical phase for each image. Then, the obtained images were processed in chronological order using the neural network model, where the moving average of the output result of five consecutive images was derived. The class with the maximum output value was defined as the surgical phase. For each surgical phase, the time at which a phase was first identified was defined as the start time, and the time at which a phase was last identified was defined as the end time. The performance was evaluated by finding the mean absolute error between the start and end times of each important phase recorded by the ophthalmologist as well as the start and end times determined by the model. The correct response rate of the cataract surgical phase classification was 90.7% for CCC, 94.5% for nuclear extraction, and 97.9% for other phases, with a mean correct response rate of 96.5%. The errors between each phase’s start and end times recorded by the ophthalmologist and those determined by the neural network model were as follows: CCC’s start and end times, 3.34 seconds and 4.43 seconds, respectively and nuclear extraction’s start and end times, 7.21 seconds and 6.04 seconds, respectively, with a mean of 5.25 seconds. The neural network model used in this study was able to perform the classification of the surgical phase by only referring to the last 5 seconds of video images. Therefore, our method has performed like a real-time classification.

DL model is better than SVM model in detecting ERM by using 3D-OCT images.

No abstract

Purpose: To evaluate the efficacy of deep learning in judging the need for rebubbling after Descemet's endothelial membrane keratoplasty (DMEK). Methods: This retrospective study included eyes that underwent rebubbling after DMEK (rebubbling group: RB group) and the same number of eyes that did not require rebubbling (non-RB group), based on medical records. To classify the RB group, randomly selected images from anterior segment optical coherence tomography at postoperative day 5 were evaluated by corneal specialists. The criterion for rebubbling was the condition where graft detachment reached the central 4.0-mm pupil area. We trained nine types of deep neural network structures (VGG16, VGG19, ResNet50, InceptionV3, InceptionResNetV2, Xception, DenseNet121, DenseNet169, and DenseNet201) and built nine models. Using each model, we tested the validation data and evaluated the model. Results: This study included 496 images (31 eyes from 24 patients) in the RB group and 496 images (31 eyes from 29 patients) in the non-RB group. Because 16 picture images were obtained from the same point of each eye, a total of 992 images were obtained. The VGG19 model was found to have the highest area under the receiver operating characteristic curve (AUC) of all models. The AUC, sensitivity, and specificity of the VGG19 model were 0.964, 0.967, and 0.915, respectively, whereas those of the best ensemble model were 0.956, 0.913, and 0.921, respectively. Conclusions: This automated system that enables the physician to be aware of the requirement of RB might be clinically useful.

Purpose. The present study aimed to compare the accuracy of diabetic retinopathy (DR) staging with a deep convolutional neural network (DCNN) using two different types of fundus cameras and composite images. Method. The study included 491 ultra-wide-field fundus ophthalmoscopy and optical coherence tomography angiography (OCTA) images that passed an image-quality review and were graded as no apparent DR (NDR; 169 images), mild nonproliferative DR (NPDR; 76 images), moderate NPDR (54 images), severe NPDR (90 images), and proliferative DR (PDR; 102 images) by three retinal experts by the International Clinical Diabetic Retinopathy Severity Scale. The findings of tests 1 and 2 to identify no apparent diabetic retinopathy (NDR) and PDR, respectively, were then assessed. For each verification, Optos, OCTA, and Optos OCTA imaging scans with DCNN were performed. Result. The Optos, OCTA, and Optos OCTA imaging test results for comparison between NDR and DR showed mean areas under the curve (AUC) of 0.79, 0.883, and 0.847; sensitivity rates of 80.9%, 83.9%, and 78.6%; and specificity rates of 55%, 71.6%, and 69.8%, respectively. Meanwhile, the Optos, OCTA, and Optos OCTA imaging test results for comparison between NDR and PDR showed mean AUC of 0.981, 0.928, and 0.964; sensitivity rates of 90.2%, 74.5%, and 80.4%; and specificity rates of 97%, 97%, and 96.4%, respectively. Conclusion. The combination of Optos and OCTA imaging with DCNN could detect DR at desirable levels of accuracy and may be useful in clinical practice and retinal screening. Although the combination of multiple imaging techniques might overcome their individual weaknesses and provide comprehensive imaging, artificial intelligence in classifying multimodal images has not always produced accurate results.

Surgical skill levels of young ophthalmologists tend to be instinctively judged by ophthalmologists in practice, and hence a stable evaluation is not always made for a single ophthalmologist. Although it has been said that standardizing skill levels presents difficulty as surgical methods vary greatly, approaches based on machine learning seem to be promising for this objective. In this study, we propose a method for displaying the information necessary to quantify the surgical techniques of cataract surgery in real-time. The proposed method consists of two steps. First, we use InceptionV3, an image classification network, to extract important surgical phases and to detect surgical problems. Next, one of the segmentation networks, scSE-FC-DenseNet, is used to detect the cornea and the tip of the surgical instrument and the incisional site in the continuous curvilinear capsulorrhexis, a particularly important phase in cataract surgery. The first and second steps are evaluated in terms of the area under curve (i.e., AUC) of the figure of the true positive rate versus (1—false positive rate) and the intersection over union (i.e., IoU) obtained by the ground truth and prediction associated with the region of interest. As a result, in the first step, the network was able to detect surgical problems with an AUC of 0.97. In the second step, the detection rate of the cornea was 99.7% when the IoU was 0.8 or more, and the detection rates of the tips of the forceps and the incisional site were 86.9% and 94.9% when the IoU was 0.1 or more, respectively. It was thus expected that the proposed method is one of the basic techniques to achieve the standardization of surgical skill levels.

PURPOSE: We demonstrated real-time evaluation technology for cataract surgery using artificial intelligence (AI) to residents and supervising doctors (doctors), and performed a comparison between the two groups in terms of risk indicators and duration for two of the important processes of surgery, continuous curvilinear capsulorhexis (CCC) and phacoemulsification (Phaco). MATERIALS AND METHODS: Each of three residents with operative experience of fewer than 100 cases, and three supervising doctors with operative experience of 1000 or more cases, performed cataract surgeries on three cases, respectably, a total of 18 cases. The mean values of the risk indicators in the CCC and Phaco processes measured in real-time during the surgery were statistically compared between the residents’ group and the doctors’ group. RESULTS: The mean values (standard deviation) of the risk indicator (the safest, 0 to most risky, 1) for CCC were 0.556 (0.384) in the residents and 0.433 (0.421) in the doctors, those for Phaco were 0.511 (0.423) in the residents and 0.377 (0.406) in the doctors. The doctors’ risk indicators were significantly better in both processes ( P = 0.0003, P < 0.0001 by Wilcoxon test). CONCLUSION: We successfully implemented a real-time surgical technique evaluation system for cataract surgery and collected data. The risk indicators were significantly better in the doctors than in the resident's group, suggesting that AI can objectively serve as a new indicator to intraoperatively identify surgical risks.

No abstract