Identification and Assessment of Pre-processing Techniques for Parkinson's Diagnosis

Parkinson's is an irreversible neurological disorder that causes motor and non-motor disabilities. To date, Parkinson's disease (PD) is detected via manual inspection of its symptoms by the neurologist. Manual detection is purely based on a doctor's expertise and analysis which may lead to misdiagnosis. Researchers have proposed various PD detection techniques based on MRI sequences. However, they yet found unsuitable for PD detection clinically due to their relatively lower accuracy. Artefacts and noise-induced during MRI acquisition degrade the quality of scans thereby resulting in poor classification performances. Hence, pre-processing techniques must be applied to the MRI dataset before the disease classification. This paper presents various pre-processing approaches available for different MRI sequences and also suggests their applicability. Parkinson's detection through pre-processed MRI sequences may yield better diagnostic performances.