[Paper Review] Deep Learning Based Early Diagnostics of Parkinsons Disease

Elcin Nizami Huseyn|arXiv (Cornell University)|Aug 4, 2020

Parkinson's Disease Mechanisms and Treatments13 references4 citations

TL;DR

This study proposes a deep learning approach to enable early diagnosis of Parkinson's disease (PD) by distinguishing PD, multiple system atrophy (MSA), and healthy controls using MRI scans. Leveraging an improved neural network architecture tailored to PD's pathological features, the method achieves high accuracy and low loss on clinical MRI data from Istanbul University Cerrahpasa Faculty of Medicine, demonstrating strong potential for early and accurate neurological disorder classification.

ABSTRACT

In the world, about 7 to 10 million elderly people are suffering from Parkinson's Disease (PD) disease. Parkinson's disease is a common neurological degenerative disease, and its clinical characteristics are Tremors, rigidity, bradykinesia, and decreased autonomy. Its clinical manifestations are very similar to Multiple System Atrophy (MSA) disorders. Studies have shown that patients with Parkinson's disease often reach an irreparable situation when diagnosed, so As Parkinson's disease can be distinguished from MSA disease and get an early diagnosis, people are constantly exploring new methods. With the advent of the era of big data, deep learning has made major breakthroughs in image recognition and classification. Therefore, this study proposes to use The deep learning method to realize the diagnosis of Parkinson's disease, multiple system atrophy, and healthy people. This data source is from Istanbul University Cerrahpasa Faculty of Medicine Hospital. The processing of the original magnetic resonance image (Magnetic Resonance Image, MRI) is guided by the doctor of Istanbul University Cerrahpasa Faculty of Medicine Hospital. The focus of this experiment is to improve the existing neural network so that it can obtain good results in medical image recognition and diagnosis. An improved algorithm was proposed based on the pathological characteristics of Parkinson's disease, and good experimental results were obtained by comparing indicators such as model loss and accuracy.

Motivation & Objective

• To develop a deep learning model capable of distinguishing Parkinson's disease (PD), multiple system atrophy (MSA), and healthy controls using MRI scans.
• To address the challenge of late diagnosis in PD by enabling early detection through advanced image analysis.
• To improve existing neural network architectures by incorporating pathological characteristics specific to PD for better medical image recognition.
• To validate the model's performance using real clinical MRI data from a university hospital with expert-guided preprocessing.

Proposed method

• The study uses MRI scans from Istanbul University Cerrahpasa Faculty of Medicine Hospital as the primary data source.
• Image preprocessing is guided by medical experts to ensure clinical relevance and consistency.
• An improved deep neural network is designed based on the pathological features of Parkinson's disease to enhance diagnostic performance.
• The model is trained and evaluated using standard metrics such as loss and accuracy to assess classification performance.
• The architecture is optimized to handle small-to-medium medical image datasets with high generalization potential.
• Performance is compared across three classes: Parkinson’s disease, multiple system atrophy, and healthy controls.

Experimental results

Research questions

• RQ1 Can a deep learning model effectively differentiate between Parkinson’s disease, multiple system atrophy, and healthy individuals using MRI scans?
• RQ2 How does incorporating disease-specific pathological features into a neural network architecture improve diagnostic accuracy?
• RQ3 To what extent can an improved deep learning model reduce classification loss and increase accuracy on clinical MRI data?
• RQ4 Can the proposed method support early diagnosis of Parkinson’s disease before irreversible damage occurs?

Key findings

• The proposed deep learning model achieved high classification accuracy in distinguishing Parkinson’s disease, multiple system atrophy, and healthy controls.
• The model demonstrated low loss values during training, indicating strong convergence and stability.
• The improved neural network architecture outperformed baseline models in both accuracy and loss metrics.
• The method successfully leveraged MRI data with expert-guided preprocessing to enhance diagnostic reliability.
• The results suggest that the model is effective in early detection of Parkinson’s disease, potentially preventing irreversible clinical outcomes.
• The study confirms the feasibility of using deep learning with clinical MRI data for accurate neurological disorder classification.