EC-Net: Efficient Compound Network with Dendritic Learning for Cataract Eye Detection

As the leading cause of visual impairment and blindness, automatic methods for diagnosing cataracts from retinal images are desired. With the advancement of computer technology, deep learning-based approaches have demonstrated superior performance compared to traditional manual methods. These algorithms are particularly advantageous in terms of conserving human and material resources and minimizing the dependency on clinical expertise. However, existing methods often struggle with extracting the non-linear characteristics inherent in cataract data. To alleviate this challenge, we propose a novel Efficient-dendritic Cataract-detection Network (EC-Net), which integrates the Dendritic Neuron Model for efficient and accurate detection of cataracts. This network leverages the nonlinear properties of dendritic neurons in conjunction with advanced deep learning feature extraction techniques to enhance the accuracy of cataract identification. Experimental results on publicly available datasets demonstrate that EC-Net significantly outperforms traditional models in the classification of cataract fundus images, providing a promising solution for medical image classification and supporting auxiliary diagnosis.