Raj A Sundar, Senthilkumar S, Radha R, Muthaiyan R
Department of Biomedical Engineering, E.G.S. Pillay Engineering College, Nagapattinam, 611002, Tamil Nadu, India.
Department of Electronics and Communication Engineering, E.G.S. Pillay Engineering College, Nagapattinam, 611002, Tamil Nadu, India.
Sci Rep. 2025 Mar 26;15(1):10400. doi: 10.1038/s41598-025-94413-3.
Disease detection plays an important role in shrimp aquaculture to ensure the health and sustainability of farming operations. Specifically, detecting viral infections at early stages can prevent significant losses. Image processing applications have been developed to detect different types of diseases in shrimp. However, theaccuracy of detection models needs improvement to detect various diseases through a single model. Therefore, this research presents a novel disease detection model using an Enhanced Recurrent Capsule Network (ERCN) with a hybrid optimization model for enhanced detection performance. The proposed ERCN utilizes dynamic routing of capsules to extract spatial hierarchies and patterns in shrimp images, while the recurrent layer extracts temporal dependencies. Performance is further improved by incorporating spatial and channel attention models to select optimal regions and features in the images for the fusion process. The dual-level feature fusion procedure combines local and global features, providing a final fused data to classify different types of diseases. Additionally, the proposed work incorporates a hybrid optimization that combines Harris Hawks Optimization (HHO) with the Marine Predator Algorithm (MPA) to fine-tune the classifier model parameters. Experiments evaluate the performance of the proposed disease detection model through various metrics such as accuracy, precision, recall, specificity, Matthews correlation coefficient, and F1-score. The resutls confirms that the performance of the proposed model is superior with precision of 94.9%, recall of 93.5%, F1-score of 94.6% and detection accuracy of 95.2% over conventional Recurrent Neural Network (RNN), Convolutional Neural Network (CNN), Gated Recurrent Unit (GRU), and Long Short Term Memory (LSTM) Networks.
疾病检测在对虾养殖中起着重要作用，以确保养殖作业的健康和可持续性。具体而言，早期检测病毒感染可防止重大损失。已开发出图像处理应用程序来检测对虾的不同类型疾病。然而，检测模型的准确性需要提高，以便通过单一模型检测各种疾病。因此，本研究提出了一种新颖的疾病检测模型，该模型使用增强循环胶囊网络（ERCN）和混合优化模型来提高检测性能。所提出的ERCN利用胶囊的动态路由来提取对虾图像中的空间层次结构和模式，而循环层则提取时间依赖性。通过纳入空间和通道注意力模型来选择图像中的最佳区域和特征进行融合过程，性能得到进一步提高。双层特征融合过程结合了局部和全局特征，提供最终融合数据以对不同类型的疾病进行分类。此外，所提出的工作纳入了一种混合优化方法，该方法将哈里斯鹰优化（HHO）与海洋捕食者算法（MPA）相结合，以微调分类器模型参数。实验通过各种指标评估所提出的疾病检测模型的性能，如准确率、精确率、召回率、特异性、马修斯相关系数和F1分数。结果证实，所提出模型的性能优于传统的循环神经网络（RNN）、卷积神经网络（CNN）、门控循环单元（GRU）和长短期记忆（LSTM）网络，其精确率为94.9%，召回率为93.5%，F1分数为94.6%，检测准确率为95.2%。
