Abstract
ObjectiveThe optical coherence tomography (OCT) model of macular edema in postoperative cataract patients is constructed by using convolutional neural network (CNN) U-Net, and its value is evaluated by relevant indicators. MethodsFrom January 2020 to December 2023, 200 binocular OCT images of 140 patients with macular edema after cataract surgery and 280 images of normal subjects in our hospital were included as data sets. They were randomly divided into a training set (70%) and valiadation set (30%). Res-SE U-Net was used to construct an automatic recognition model of macular edema in OCT images of patients after cataract surgery. The model was evaluated by the change of loss function and accuracy, and by drawing the receiver operating characteristic curve (ROC). ResultsThe segmentation efficacy between the U-Net model and its enhanced counterpart, Res-SE U-Net, was assessed in this study. Notably, the Dice coefficient for macular edema segmentation exhibited substantial improvement with the Res-SE U-Net model, reaching 0.96 compared to 0.90 for the conventional U-Net model. Additionally, evaluation based on the intersection over union (IoU) coefficient revealed a superiority of the Res-SE U-Net model with a coefficient of 0.79, in contrast to 0.76 for the U-Net model. Moreover, the area under the curve (AUC) analysis further supported the enhanced performance of the Res-SE U-Net model. The AUC values were determined as 0.837 for the Res-SE U-Net model and 0.811for the U-Net model. ConclusionThe Res-SE U-Net model of convolutional neural network based on OCT images can effectively identify and segment macular edema in patients after phacoemulsification.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
More From: Journal of Radiation Research and Applied Sciences
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.