A Deep Learning Architecture with Spatio-Temporal Focusing for Detecting Respiratory Anomalies

TL;DR

This paper introduces a deep learning system utilizing spatio-temporal focusing and attention mechanisms to detect respiratory anomalies from spectrograms, achieving top performance on a benchmark dataset.

Contribution

The work proposes a novel deep learning architecture with spatial-temporal focusing and multi-head attention for respiratory anomaly detection, evaluated on a challenging benchmark.

Findings

Achieved Top-1 performance on the SPRSound dataset

Score of 0.810 in Task 1-1, demonstrating high accuracy

Robust system with consistent results across tasks

Abstract

This paper presents a deep learning system applied for detecting anomalies from respiratory sound recordings. Our system initially performs audio feature extraction using Continuous Wavelet transformation. This transformation converts the respiratory sound input into a two-dimensional spectrogram where both spectral and temporal features are presented. Then, our proposed deep learning architecture inspired by the Inception-residual-based backbone performs the spatial-temporal focusing and multi-head attention mechanism to classify respiratory anomalies. In this work, we evaluate our proposed models on the benchmark SPRSound (The Open-Source SJTU Paediatric Respiratory Sound) database proposed by the IEEE BioCAS 2023 challenge. As regards the Score computed by an average between the average score and harmonic score, our robust system has achieved Top-1 performance with Scores of 0.810, 0.667, 0.744, and 0.608 in Tasks 1-1, 1-2, 2-1, and 2-2, respectively.