A Hybrid CNN-RNN Model for Enhanced Pneumonia Detection using X-Ray Imaging

This paper introduces a robust deep learning framework for the detection of pneumonia using chest X-ray images, emphasizing the integration of both spatial and temporal data for enhanced diagnostic accuracy. Initially, the research employed standard CNNs to assess their baseline efficacy in identifying pneumonia-related abnormalities within X-ray images. Building upon this, the investigation progressed to more sophisticated hybrid architectures, including CNN-RNNs, CNN-LSTMs, and CNN-GRUs. These hybrid models synergize the powerful spatial feature extraction capabilities of CNNs with the temporal sequence processing strengths of RNNs, LSTMs, and GRUs. Experimental results demonstrated that the CNN-LSTM model achieved the highest accuracy of $\mathbf{9 8 . 5 \%}$, significantly outperforming traditional CNNs, which achieved $93 \%$ accuracy. The findings underscore the potential of these hybrid deep learning models in providing reliable, real-time clinical decision support for pneumonia detection, offering notable improvements in diagnostic accuracy and recall over traditional models.