Multimodal Feature Fusion with CNN-LSTM for Respiratory Disease Classification Using Lung Sound Analysis

  • Unique Paper ID: 166814
  • Volume: 11
  • Issue: 2
  • PageNo: 2286-2293
  • Abstract:
  • The diagnosis and treatment of respiratory disorders are extremely difficult and time-consuming, necessitating accurate and prompt care. Using advanced signal processing techniques and machine learning algorithms, lung sound analysis presents a viable option for non-invasive illness classification. In this thesis, the effectiveness of multimodal feature fusion for robust respiratory disease classification is examined. Specifically, Mel-frequency cepstral coefficients (MFCCs), wavelet transform, mel-spectrogram (mSpec), and Chroma short-time Fourier transform (Chroma STFT) are combined with convolution neural networks (CNNs) and long short-term memory networks (LSTMs).The third greatest cause of death worldwide is respiratory disorders. When it comes to treating respiratory illnesses, early detection is essential since it increases the efficacy of interventions such as medication and stopping the disease's spread. This article's primary goal is to suggest a revolutionary lightweight inception network that uses lung sound data to classify a variety of respiratory disorders. There are three phases to the suggested framework: 1) Preprocessing; 2) extraction and conversion of the mel spectrogram into a three-channel image; and 3) applying the respiratory disease lightweight inception network (RDLINet), a proposed lightweight inception network, to classify the mel spectrogram images into distinct pathological groups.

Cite This Article

  • ISSN: 2349-6002
  • Volume: 11
  • Issue: 2
  • PageNo: 2286-2293

Multimodal Feature Fusion with CNN-LSTM for Respiratory Disease Classification Using Lung Sound Analysis

Related Articles