Hearing Loss Detection in Medical Multimedia Data by Discrete Wavelet Packet Entropy and Single-Hidden Layer Neural Network Trained by Adaptive Learning-Rate Back Propagation

摘要

In order to develop an efficient computer-aided diagnosis system for detecting left-sided and right-sided sensorineural hearing loss, we used artificial intelligence in this study. First, 49 subjects were enrolled by magnetic resonance imaging scans. Second, the discrete wavelet packet entropy (DWPE) was utilized to extract global texture features from brain images. Third, single-hidden layer neural network (SLNN) was used as the classifier with training algorithm of adaptive learning-rate back propagation (ALBP). The 10 times of 5-fold cross validation demonstrated our proposed method yielded an overall accuracy of 95.31%, higher than standard back propagation method with accuracy of 87.14%. Besides, our method also outperforms the "FRFT + PCA (Yang, 2016)", "WE + DT (Kale, 2013)", and "WE + MRF (Vasta 2016)". In closing, our method is efficient.