ConvGateRNN: Convolutional Gated Attention Recurrent Neural Network for Cervical Cancer Screening

Abstract

Cervical cancer continues to pose a substantial obstacle to global health, especially in areas where access to healthcare is restricted. Traditional screening methods are often manual, laborious, and prone to error. In this paper, a novel method for early-stage cervical cancer detection with a Convolutional Gated Attention Recurrent Neural Network (ConvGateRNN) is proposed. The methodology involves preprocessing and feature selection using the spotted hyena optimization technique, followed by the implementation of a ConvGateRNN classifier. The ConvGateRNN leverages deep learning techniques and attention mechanisms to effectively identify disease traits from cervical cancer datasets. Experimental findings showcase ConvGateRNN’ s exceptional efficacy relative to current approaches, achieving a remarkable accuracy of 96.18% alongside noteworthy precision, recall, and F1-score metrics. The promising results suggest that ConvGateRNN offers a favorable solution for unerring and efficient cervical cancer screening, potentially revolutionizing current practices and improving healthcare outcomes for women worldwide.