Hu Moment Invariant: A New Method for Hearing Loss Detection

Abstract

This paper proposed a novel hearing loss detection method. Our method first used seven Hu moment invariants to extract features. Afterwards, we used support vector machine to act as the classifier. The 10x5-fold cross validation shows our method yielded an overall accuracy of 77.47± 1.17%. The sensitivities of healthy control, left-sided hearing loss, and right-sided hearing loss are 77.60± 5.72%, 77.60± 4.30%, and 77.20± 5.98%, respectively. In all, our method is effective in hearing loss identification.

Hearing loss classification via AlexNet and Support Vector Machine

“…In the SWE+CSO [17] method, the healthy control group (HC), leftsided SNHL(LSHL) and right-sided SNHL(RSHL) was 91.00%, 89.00%, 90.67%, and 90.22%, respectively. The MAF1 of HMI + SVM [13], WE+CSO [14], FRFE + DAG-SVM [15], SWE+ GA [16], and [17] were 76.06%, 84.11%, 94.06%, 89.89% and 90.22%, respectively. In our AlexNet+SVM method, the left SNHL(LSHL), right SNHL(RSHL), healthy control group (HC) and MAF1 were 94.00%, 95.17%, 94.67% and 94.61%, respectively.…”

“…Through reading the recent papers published by scholars, it is found that scholars have proposed several feasible methods for image classification [12]. Tang, L. et al (2018) [13] proposed the method of combining Hu moment invariant (HMI) and Support Vector Machine. Gao, R. et al ( 2019) [14] used the hearing loss recognition method based on wavelet entropy and cat swarm optimization, Wang, L. et al (2020) [15] suggested hearing loss recognition based on fractional Fourier entropy and direct acyclic graph Support Vector Machine.…”

“…We compared the AlexNet+SVM method used in this hearing loss model with the advanced HMI+SVM [13], WE+CSO [14], FRFE+DAG-SVM [15], SWE+GA [16], SWE+CSO [17] methods and obtained the data shown in Table 2. The data comparison of each model is also shown in detail in Figure 9.…”

“…The data comparison of each model is also shown in detail in Figure 9. In HMI + SVM [13], the accuracy of healthy control group (HC), left-sided SNHL(LSHL) and right-sided SNHL(RSHL) was 76.83%, 75.00%, and 76.33%, respectively. The accuracy of SNHL(LSHL), SNHL(RSHL), and healthy control group (HC) by WE+CSO [14] method was 84.33%, 83.33%, and 84.67%, respectively.…”

Hearing loss classification via stationary wavelet entropy and Biogeography-based optimization

“…We compared the model accuracy of this experiment with other current advanced methods: WE-GA [9], HMI [36] and SVM-PSO [10]. The comparison results are shown in Table 8.…”