AIPD-DC10-EMC

• Host Institution: Erasmus Medical Center (EMC)
  
• PhD Enrolment: Erasmus Medical Center (EMC)
  
• Start Date: October 2025
  
• Duration: 36 + 12 months (first 36 months funded by AIPD)
• Official PhD Supervisor: Juan Hernandez Tamames

Research Objectives

To investigate the difference among groups with PD and atypical parkinsonism (AP) based on advanced quantitative MRI biomarkers and AI/ML algorithms and, secondly, to be able to provide an early diagnosis to patients classified as CUP (Clinically Unclassifiable Parkinson). In this project, we will extract quantitative data from available MRI databases with confirmed diagnosis from these groups: PD, MSA and PSP. We will include an Erasmus MC database of CUPS (currently being recruited). We will use AI techniques: first, to classify the different groups of patients based on quantitative MRI biomarkers and, second, to predict the final diagnosis of CUPS. We will develop an AI “one-stop-shop” tool for the diagnosis of CUPS that could be used in Multi-Center Clinical Trials.

Working Plan:

1. To review open-access databases of MRI.
2. To review the potential MRI structural and functional biomarkers.
3. To develop standardized postprocessing pipelines for extracting quantitative biomarkers.
4. To develop supervised classifiers based on AI and MRI quantitative biomarkers. XAI techniques will be used to provide model explanations.
5. To develop an AI tool for predicting diagnosis of CUPs.
6. To clinically validate the performance of the AI tools over pre-existing MRI databases of Parkinson, AP and CUPS.

DCs10 and 11 will work closely together due to their common focus on MRI imaging.

Expected Results

• Novel AI/ML models for better diagnosis of CUP patients
   

Planned Secondment(s)

• Host: GE Healthcare
  • Duration: 24 months
    
  • Purpose: Learning about AI/ML applications to MRI

This project is part of the "Digital Health" work package.