A Novel Deep-Learning Based Approach for Mild Cognitive Impairment Screening in Patients with Parkinson’s Disease

Parkinson's disease (PD) is a progressive neurodegenerative disorder with motor and non-motor complications including mild cognitive impairment (MCI) and dementia. Early detection of MCI is crucial to initiate appropriate therapeutic treatments that may possibly delay the progression of the disease to dementia and severe cognitive dysfunction. In this paper, we propose a three-dimensional convolutional neural network applied on the time-scale representation of sleep electroencephalography to classify PD patients into subjects with normal cognition (NC) or MCI. The proposed framework was capable of screening MCI at a high accuracy, sensitivity, specificity, and quadratic weighted kappa up to 95.9%, 94.9%, 95% and 91% respectively. This represents an important tool that may allow clinicians to monitor the progression of PD and support an objective diagnosis of MCI.