TransMedCA-Net: A Transformer-MedNext Hybrid with Coordinate Attention for Optic Disc and Cup Segmentation

Fundus scans must precisely segment the optic disc (OD) and optic cup (OC) to detect glaucoma early. TransMedCANet is a new network design introduced in this study. To improve fundus picture structural awareness, we combine Transformerbased global context modelling, MedNext’s smart convolutional architecture, and the coordinate attention. To address challenges such as the low contrast between OD and OC regions and variations in shape and size, the model leverages the Transformer module to capture long-range dependencies and employs the MedNext backbone for lightweight yet effective feature extraction. The coordinate attention mechanism improves spatial feature localization by using precise spatial information. Extended experiments on benchmark datasets like REFUGE and DRISHTIGS reveal that TransMedCA-Net outperforms current approaches in OC boundary definition. The recommended network is reliable and fast for computer-aided diagnostic systems and clinical glaucoma screening.