Hybrid QDCNN-DNFN for Laryngeal Cancer Detection Using Image and Voice Analysis in Federated Learning

Laryngeal cancer is a form of cancer that affects the larynx. It is a malignant tumor that occurs in the neck and head region. However, it remains challenging to identify laryngeal cancer in the primary stage to avert the mortality rate. This research aims to devise a hybrid Quantum Dilated Convolutional Neural Network- Deep Neuro-Fuzzy Network (QDCNN-DNFN) for laryngeal cancer detection by utilizing image and voice analysis in federated learning. Federated learning has two types of units, such as nodes and a server. Primarily, the local nodes are trained, based on the local data. At the server, the global model is generated by considering the weights of the local node, and this method is copied by the local nodes based on which subsequent training is done. At the local node, laryngeal cancer detection is performed simultaneously by utilizing the features from both the image and the voice by using the QDCNN-DNFN. The efficacy of the proposed QDCNN-DNFN is evaluated using certain metrics, like Mean average precision, loss function, Mean Square Error (MSE), root mean square error, Mean Square Error (MSE), accuracy, and F1-score, and the value attained is 0.878, 0.083, 0.046, 0.210, 0.891, and 0.871. The proposed method is used in the early detection of Laryngeal cancer with high accuracy, which could contribute to reducing the mortality rate associated with this disease.