Classification of Nail Abnormalities using Convolutional Neural Network

Abstract

Nails are one organ which can indicate the status of a health condition through its own appearance. To create a model which can be applied as a tool for self-classifying nail abnormalities, this article presents the study and analysis on seven abnormalities of nails: 1) Beau's lines on Nails, 2) Black line on Nails, 3) Nails Clubbing, 4) Muehrcke’s nails, 5) Onycholysis, 6) Terry’s nail, and 7) White spots on Nails. The data is composed of seven hundred images compiled from Google images. They are separated into training sets, validation sets, and test sets and arranged into a 64:16:20 ratio, respectively. The Convolutional Neural Network (CNN) model method was developed from the Artificial Neural Network (ANN) model. It points out that CNN achieves 81.43% accuracy, which is more efficient in classifying nail abnormalities than ANN, which only has 43.57% accuracy.