| Summary: | Hearing is a very important feeling for humans. It helps to perceive the environment around us and can warn us of any imminent dangers around us. Hearing loss is an impaired ability to hear sound. Sensorineural hearing loss (SNHL) is very common in today's society. Classification of hearing loss is helpful for clinical diagnosis, finding suitable treatment methods and allocating appropriate medical interventions. This paper presents a novel method for hearing loss classification based on magnetic resonance images, which can automatically recognize tissue specific features in a given MRI image. The study involved magnetic resonance images (MRI) of 60 participants in three categories of balance: left SNHL, right SNHL, and a healthy control group. Our method uses AlexNet to extract the features of a single category, and uses Extreme Learning Machine to form a triclassifier for automatic SNHL classification. From the experimental results and the practicability of the algorithm, the classification performance is obviously better than the standard deep learning method and other traditional methods.
|
|---|