Project ID: P202205130002
Investigating the relationship between whole-exome sequence analysis results of clinical features and features extracted from DaT SPECT images Description
Deep Learning, Machine Learning, Image Processing
People with the below expertise are able to apply for this project: 1-The individual with enough experience in medical image processing techniques 2-The individual with enough experience in 3-D Deep learning methods 3- The individual with enough experience in traditional and deep learning fusion techniques. Both programming languages Matlab and Python are acceptable, but Python works better for some people who aim at working on google Colab.
This project includes:
Parkinson’s disease (PD) is the second prevalent neurodegenerative disorder after Alzheimer’s disease (AD), is affected 1–5% of the general population and 1-2% of the world population above 65 years, increasing to 4% over the age of 85. Over 20 genes have been associated with familial PD. 5–10% of cases are only inherited in familial PD with mutations identified in several genes such as leucine-rich repeat kinase 2 (LRRK2), glucocerebrosidase (GBA), alpha-synuclein (SNCA), Parkin (PRKN), PTEN-induced kinase 1 (PINK1), DJ-1, and VPS35. In this study, the PD genome is analyzed using whole-exome sequence (WES) and we will investigate the relationship between mutations in PD patients with clinical, motor, non-motor features as well as Radiomics features extracted from DAT SPECT images. We will also classify PD subgroups using a combination of semi-supervised methods and clustering. In the first stage, genetic data of patients will be analyzed by WES method. We employed 298 features, including 55 clinical features (CFs, motor and non-motor symptoms), 28 conventional imaging features (CIFs) and 215 RFs extracted from each ROI of DAT-SPECT image using our standardized SERA software. In this study, three groups of algorithms were employed, including 3 semi-supervised algorithms (SSA) linked with 10 Feature Selection algorithms (FSAs) and 11 Feature Extraction algorithms (FEAs). The hyperparameters of these models were tuned using a 5-fold cross-validation technique and grid search.