• Version
• Download 23
• File Size 1,014.75 KB
• File Count 1
• Create Date 28 May 2025
• Last Updated 28 May 2025

Neurodegenerative Disorder Disease Prediction Using RCNN Classification and Spiral Image

Authors:

T.Annalakshmi

Student, Dept of computer Application

Dr.M.G.R Educational and Research Institute

Saisuve03@gmail.com

Ms.K.Annalakshmi

Assistant Professor, Dept of computer Application

Dr.M.G.R Educational and Research Institute

annalakshmi.mca@drmgrdu.ac.in

ABSTRACT

In recent years, the integration of machine learning within the field of biological and medical sciences has significantly enhanced our ability to interpret complex clinical data. This advancement has been especially impactful in the domain of neurodegenerative disorders such as Parkinson’s Disease (PD), where early and precise diagnosis remains a persistent challenge. In this study, we propose a novel approach utilizing a Region-based Convolutional Neural Network (R-CNN) to distinguish between healthy individuals and those affected by PD.

The method focuses on identifying subtle but meaningful patterns in visual diagnostic tools such as spiral and wave drawings, which are known to reflect motor impairments typical of the disease. Parkinson’s Disease is characterized by the progressive degeneration of motor control, often manifesting in tremors, rigidity, and bradykinesia. While traditional diagnostic techniques depend heavily on the observational skills of neurologists—assessing speech, writing, gait, and hand-drawn spirals—the process can be both time-consuming and subjective.

To overcome these limitations, we adopted a deep learning strategy, using Faster R-CNN to directly analyze input patterns without manual feature extraction. This model automates the learning of distinguishing features from a dataset comprising MRI images and spiral drawing samples, enabling accurate classification with minimal human intervention.

The system was trained and tested on real biomedical drawing datasets, including spiral and wave patterns from both PD patients and healthy controls. We achieved an accuracy of 96% during training and 86–97% during testing using batch normalization, which stabilized learning and minimized overfitting. This high performance clearly demonstrates the effectiveness of our approach in capturing the fine-grained motor distortions associated with Parkinson’s Disease.

The model’s ability to learn from nuanced visual cues reflects its strength in identifying early-stage abnormalities, which are often difficult to detect with conventional tools. Moreover, this research highlights the superiority of the proposed model when compared to existing state-of-the-art methods. The results demonstrated not only robustness in pattern recognition but also the potential for real-world clinical applications. By automating a significant portion of the diagnostic workflow, this approach could reduce diagnostic delays and support neurologists in making more informed decisions. The system has proven effective in its current implementation and will be further validated using a broader, more diverse set of clinical data in future work.

Keywords - Deep Learning - R-CNN - Spiral Drawing Analysis - Medical Image Classification - Early Diagnosis.