Deteksi Dini Penyakit Parkinson Melalui Gambar Spiral Dan Gelombang Menggunakan Metode Convolutional Neural Network
Total View This Week0
Total View This Week0
Institusion
Politeknik Negeri Jember
Author
Mashwafah, Azizatul
Subject
458 - Teknik Informatika
Datestamp
2024-05-21 01:02:15
Abstract :
This research is focused on developing an early detection system for Parkinson's disease through spiral and wave images that aims to support initial screening before conducting further examinations to a Neurologist. This research aims to develop an early detection system for Parkinson's disease through spiral and wave images. The method used is to use Convolutional Neural Network (CNN) with DenseNet-169 architecture to classify handwriting as tremor or non tremor. In this study using spiral and wave handwriting to classify handwriting in an image with class division, namely tremor and non tremor. Before the classification process is carried out, the handwriting image goes through a series of preprocessing stages including Gaussian blur, conversion to grayscale, erosion, Otsu Thresholding and image resize. The results of several trials that have been carried out show that a learning rate of 0.0001 achieves an accuracy value of 93.13%. To assess the success of the Parkinson's disease early detection website through spiral and wave images, User Acceptance Testing (UAT) was carried out which showed a success rate of 91.8% which indicates that users accept the Parkinson's disease early detection website quite well.
This research is focused on developing an early detection system for Parkinson's disease through spiral and wave images that aims to support initial screening before conducting further examinations to a Neurologist. This research aims to develop an early detection system for Parkinson's disease through spiral and wave images. The method used is to use Convolutional Neural Network (CNN) with DenseNet-169 architecture to classify handwriting as tremor or non tremor. In this study using spiral and wave handwriting to classify handwriting in an image with class division, namely tremor and non tremor. Before the classification process is carried out, the handwriting image goes through a series of preprocessing stages including Gaussian blur, conversion to grayscale, erosion, Otsu Thresholding and image resize. The results of several trials that have been carried out show that a learning rate of 0.0001 achieves an accuracy value of 93.13%. To assess the success of the Parkinson's disease early detection website through spiral and wave images, User Acceptance Testing (UAT) was carried out which showed a success rate of 91.8% which indicates that users accept the Parkinson's disease early detection website quite well.
Download
book
BibTex
Latex, Jabref
cloud_download
Original Resource
url resource Institution
