Emphysema Recognition by Multi-input Neural Network
GUO Tao
GU Yicong
LIU Qiming
LI Cheng
SHI Shuai
Abstract：Multiple input neural network is used to classify the two typical characteristics of emphysema(stridor and blister),so as to determine whether high altitude emphysema is present or not.For lung sound data,four spectral feature extraction methods including Mel spectrum(Mel),constant Q transform(CQT),Wavelet transform(WT)and short-time Fourier transform(STFT)are used after filtering and denoising.LBP and Mixup are used for data enhancement,and lung sound classification is performed in Mul-CNN.The accuracy,specificity,sensitivity and ICBHI scores of lung sounds are 93.6%,92.3%,94%and 93.1%when WT and Mel are used as inputs.
Keywords：emphysemaMul-CNNLBPMixup
Publication Date：2025-03-20
Online Publishing Date：2025-08-15(First online date of this platform, not the publication date of the document)
Pages：6( 678-683 )