Proactive-BiLSTM: Glaucoma detection using FUNDUS and OCT retinal images

Glaucoma detection identifies the early signs of eye conditions that can lead to vision loss by analyzing the retinal images to detect abnormalities, such as increased intraocular pressure, changes in the optic nerve head, or structural alterations in the retina. The challenges faced by the existing models include the difficulty in detecting subtle features, variability in image quality, and complex patterns that may resemble normal variations. Moreover, the traditional models struggle to adapt to the evolving patient data, capture long-term dependencies, and often suffer from lower accuracy. Hence, this research proposes the Proactive Hybridized Bidirectional Long Short-Term Memory (BiLSTM) model for Glaucoma detection. The proactive hybridized BiLSTM model is designed to enhance the detection of glaucoma by processing the retinal images. The proactive hybridized BiLSTM model enables the model to capture complex temporal dependencies and relationships within the data, which are crucial for identifying subtle patterns indicative of glaucoma, for which multifaceted feature extraction is employed. Moreover, the Proactive Hybridized BiLSTM model adapts to dynamic changes in the data to learn and predict glaucoma-related features, ultimately improving detection performance over time. The proposed Proactive hybridized BiLSTM model attains higher accuracy, sensitivity, and specificity of 96.65%, 96.51%, and 96.79% using the OCT and FUNDUS image dataset.