Computer-Aided Diagnosis of Parkinson's Disease Using Complex-Valued Neural Networks and mRMR Feature Selection...

Area of Science:

• Neurology
• Biomedical Engineering
• Artificial Intelligence

Background:

• Parkinson's disease (PD) diagnosis relies on clinical evaluations, which can be insufficient for early detection.
• Current diagnostic methods for PD are often time-consuming and subjective.
• There is a need for automated systems to aid in the early diagnosis of Parkinson's disease.

Purpose of the Study:

• To propose a novel automated method for diagnosing Parkinson's disease (PD).
• To leverage biomedical sound measurements from phonation samples for PD diagnosis.
• To develop a system that offers a powerful tool for effective PD diagnosis.

Main Methods:

• Utilized biomedical sound measurements from continuous phonation samples as input attributes.
• Applied a minimum redundancy maximum relevance (mRMR) algorithm for effective attribute selection.
• Converted selected attributes to complex numbers and input them into a complex-valued artificial neural network (CVANN).

Main Results:

• Identified effective attributes from biomedical sound measurements using mRMR.
• Successfully processed complex-valued attributes using a CVANN.
• The proposed system demonstrates potential as a tool for PD diagnosis.

Conclusions:

• A novel method using voice analysis and CVANN shows promise for Parkinson's disease diagnosis.
• Automated diagnosis using biomedical sound measurements could improve early detection rates.
• The developed system offers a potentially powerful and efficient approach to PD diagnosis.