Ear disease determination on computer-assisted outer and middle ear images

Abstract

This paper proposes a method to determine the outer and middle ear disease from image files in computer data efficiently and developed the environment friendly unit. The proposed method starts its method by analyzing the accessed file. If a file is labeled as a file that needs to be classified, the proposed method collects the list of image files in the directory, then it starts to run CNN-based classification to determine the type of ear disease suffered by the patient based on the information contained in the input image. If an image file is detected as a certain outer and middle ear disease, that file will be given a label and its probability. The CNN engine work for classifying the input image captured by Endoscope Ear Cleaning Tool Kit. The classification machine was developed using 1800 images for 18 disease classes with each class having a sample of 100 to 105 sample images. The sample image was augmented from 900 patient data. The experimental result shows that the Ear Disease Determination on Computer-Assisted Outer and Middle Ear Images which is based on CNN has run properly which is indicated by 80% accuracy and potentially implemented for the rural health center facility.