|
|
 |
| |
|
|
|
Study and Evaluation of Parameters Influencing Parkison’s Disease Using 3D CNN |
|
|
|
PP: 579-588 |
|
|
|
doi:10.18576/amis/170406
|
|
|
|
Author(s) |
|
|
|
Caroline El Fiorenza J,
V. Sellam,
|
|
|
|
Abstract |
|
|
| A diagnostic method of determining protein structure in persons for Parkinson’s disease (PD) would be a 3D imaging scan. Convolutional Neural Networks (CNNs), which are efficient when used with spatial data, are suitable candidates for automating this diagnostic procedure to assist medical workers. The protein structure of PD patients was evaluated or diagnosed using a Improved Faster Recurrent Convolutional Neural Network (IFRCNN) ordinal model in this article. A data preprocessing technique was modified to operate PD because IFRCNNs require big datasets to operate acceptably. We take into account the Improved Shortest Paths Ordinal Graph-based Oversampling (ISP-OGO) technique that employs a gamma probabilistic model for the creation of Inter-Class Data (ICD). It is proposed that ISP-OGO be modified to become the ISP-OGO-β method that uses the beta distribution, which is more appropriate than gamma for creating synthetic samples in the ICD. A novel 3D image dataset is the foundation for the evaluation of the various approaches demonstrates how ISP-OGO-β produces greater performance than ISP-OGO and the ordinal method enhances the effectiveness compared to the nominal method. |
|
|
|
|
|
|
|
