HybridNet-X: A Hybrid Deep Learning Network with Fuzzy-Enhanced Firefly Algorithm for Lung Disorder Diagnosis Using X-Ray Images

Lung disorders such as pneumonia and tuberculosis present significant diagnostic challenges when detected through images of chest X-ray. This work proposes HybridNet-X, a hybrid deep learning model that integrates EfficientNet and Vision Transformer (ViT) with a hybrid Attention Mechanism to improve feature extraction and classification accuracy. Additionally, the model is optimized using the Fuzzy-Enhanced Firefly Algorithm (FEFA) which refines the network's parameters by enhancing convergence speed and solution accuracy. This combined approach leads to a more efficient and precise diagnosis of lung disorders. The methodology encompasses several stages including image preprocessing, feature extraction and classification followed by optimization using FEFA. The proposed HybridNet-X with FEFA optimization model outperforms the other baseline models like VGG+FEFA, ResNet+FEFA, DenseNet+FEFA using the NIH Chest X-ray repository containing a labeled dataset of X-ray images with a accuracy of 98.5% accuracy, precision of 98.7%, recall of 98.4% and a F1-score of 98.5%. In contrast, the proposed classifier without FEFA optimization resulted in lower performance with an accuracy of 96.8%. The results confirm that the integration of FEFA significantly improves the classification performance making the model more robust and suitable for real-time clinical application in detecting lung disorders from X-ray images.