Related Experiment Video
Updated: Oct 10, 2025
04:04
Asthma Detection Research Based on Voice Signal Processing and Machine Learning
Published on: July 22, 2025
521
Inception-Based Network and Multi-Spectrogram Ensemble Applied To Predict Respiratory Anomalies and Lung Diseases
Summary
This study introduces a novel deep neural network for lung disease detection using respiratory sounds. The model effectively identifies respiratory anomalies and diseases from spectrograms, showing promising results on a benchmark dataset.
Area of Science:
- Medical Informatics
- Artificial Intelligence in Medicine
- Respiratory Medicine
Background:
- Lung diseases pose a significant global health burden.
- Accurate and early detection of lung diseases is crucial for effective treatment.
- Traditional diagnostic methods can be invasive or time-consuming.
Purpose of the Study:
- To develop and evaluate an inception-based deep neural network for detecting lung diseases.
- To utilize respiratory sound analysis for non-invasive disease identification.
- To assess the network's performance on a recognized respiratory sound dataset.
Main Methods:
- Respiratory sound recordings were collected from patients.
- Front-end feature extraction transformed audio into spectrograms, capturing spectral and temporal data.
- Back-end classification employed an inception-based deep neural network for disease detection.
Main Results:
- The proposed network achieved competitive scores on the ICBHI benchmark dataset.
- Scores for respiratory anomaly detection were 0.53/0.45.
- Scores for respiratory disease detection were 0.87/0.85.
Conclusions:
- Deep neural networks, particularly inception-based models, show significant potential for lung disease detection using respiratory sounds.
- Spectrogram analysis is an effective method for representing respiratory sound features for AI analysis.
- The study demonstrates a viable non-invasive approach for identifying respiratory anomalies and diseases.
