Attention-Based Recurrent Neural Network for Accurate Detection of Gastrointestinal Diseases

Gastrointestinal (GI) diseases represent a significant global health challenge, necessitating timely and accurate diagnosis to improve treatment outcomes. Therefore, in this study, an Attention Based Recurrent Neural Network (RNN) framework is proposed for the accurate detection of gastrointestinal diseases. The proposed methodology begins with pre-processing of input images using the scharr filter to enhance image edges and improve feature clarity. Segmentation is performed through global thresholding, isolating the regions of interest for further analysis. For feature selection, Ant Bee Colony optimization (ABC) is employed, enabling the capture of multiscale and directional image features critical for distinguishing pathological patterns. Finally, an Attention-Based RNN is applied for classification, leveraging its ability to focus on relevant spatial and temporal features, ensuring precise and robust disease detection. Experimental results demonstrate the integration of Scharr filtering, global thresholding, contourlet based feature extraction, and attention mechanisms significantly improves diagnostic accuracy and the system implemented using Python, provides a promising automated solution for GI disease detection, with an accuracy of 95% and precision of 94%, aiding clinicians in early diagnosis and enhancing patient care outcomes, highlighting its potential for real-world clinical application.