Deep Learning-Driven Phonopneumographic Analysis For Pulmonary Disease Recognition Using Dft And Melspectrograms

Phonopneumographic analysis involves the study of respiratory sounds to identify and diagnose pulmonary diseases. With advancements in deep learning, novel approaches using Digital Fourier Transform (DFT) and MelSpectrograms have emerged for automated and accurate disease recognition. This study proposes a deep learning-driven system that analyzes lung sounds, converts them into MelSpectrograms and frequency-domain representations, and classifies them using convolutional neural networks (CNNs). This approach enhances diagnostic accuracy and enables early detection of respiratory disorders such as pneumonia, asthma, and chronic obstructive pulmonary disease (COPD). The proposed method offers a non-invasive, efficient, and scalable solution for pulmonary disease screening.