Fundus image segmentation based on improved M-Net network

With the continuous advancement of computer vision technology, deep learning networks show great potential in medical imaging diagnosis. Glaucoma is the second largest blinding eye disease in the world, and the traditional optic nerve head (ONH) evaluation method is costly and time-consuming, which does not apply to mass screening. Based on this, this paper proposes a deep learning network-based fundus image segmentation method, which exhibits high accuracy in glaucoma detection and is sensitive and robust to early lesion images. Compared with traditional manual screening methods, this deep learning network shows significant improvement in speed and accuracy. Meanwhile, we utilize image super-resolution-based fundus image segmentation to improve image quality and microstructure clarity, providing ophthalmologists with more detailed and accurate diagnostic aids.