A Water Wheel Optimized Gated Attention Network (WOGAN) Model for an Automated Cervical Cancer Diagnosis
Sandeep Kumar, Y. Sukhi · 2024
Cervical cancer is one of the leading causes of early mortality for women worldwide, accounting for almost 90% of deaths from the disease in developing countries. Cervical cancer is usually associated with certain risk factors. The goal of this work is to establish and develop the Water Wheel Optimized Gated Attention Network (WOGAN), an automated method for cervical cancer diagnostics that is both precise and portable. In this instance, the Similarity-based Data Imputation (SDI) technique is used to identify missing values and replace them at the start of the classification process. After that, the Novel Water Wheel Plant Optimization (W2PO) method is used to choose the best characteristics and decrease the dimensionality of the data. Additionally, a classification technique based on Graph Convoluted Networks with Position & Channel Attention (GCN-PCA) is used to accurately classify cervical cancer with greater accuracy and shorter training times. In this study, the popular and open-source UCI repository dataset has been used to test and validate the proposed WOGAN model using enormous parameters. By using the proposed methodologies, the WOGAN system reaches the disease diagnosis accuracy up to 98.5% with little computational complexity.