Research

Computational Population Biology Group

Erasmus MC Medical Center

Data Science Venn Diagram

Computational Population Biology group develops and applies methods for the integrative analysis of large-scale biological, epidemiological and clinical data. Our goals are to improve the understanding of how various omics affect complex traits and to make use of such insights to improve the diagnosis, prevention and treatment of diseases whenever possible.

The group uses the latest approaches in genomics, medical imaging, computer science, statistics and machine learning to sort through increasingly rich and massive amounts of data.

Radiology Department

The Biomedical Imaging Group Rotterdam (BIGR) has years of expertise in machine learning algorithms and medical infrastructure development. The group has access to a large GPU cluster, making it perfect for AI-related projects.

Epidemiology Department

Long history of successful epidemiological studies, unique facilities of the Rotterdam Study (ERGO) and the Netherlands Institute for Health Sciences (NIHES) educational program. Well established within the CHARGE consortium.

Research Expertise

Big Data Analysis Omics Bioinformatics Epidemiology Software Development Machine Learning Deep Learning Data Science Data Visualization 3D Medical Imaging

Omics Analysis

We are developing novel algorithms for omics data analysis to understand better the relationship between them and various complex diseases.

We primarily focus on developing Explainable AI, which are essential for interpretability of the results.

GenNet Network Architecture

GenNet example of the network architecture

GenNet Explainable AI

Example of Explainable AI module of GenNet framework

Additionally, together with NVIDIA, we are working on a Federated Learning framework, which will connect research centers all over the world. We primarily focus on CHARGE consortium and EADB consortium. Such framework will allow running machine learning and deep learning models without sharing raw data in large multi-center settings.

NVIDIA Clara Framework

NVIDIA Clara framework for federated learning

GenNet Pipeline

GenNet pipeline for federated learning

Quantitative Traits Analysis

To explain the etiology of complex traits, we not only need multi-omics approaches, but also the best representation of such traits, i.e. endophenotypes. Therefore, we are interested in developing methods to derive such endophenotypes.

We are working on several domains:

Brain Imaging
Brain Networks
3D Shapes
3D Skull Analysis
3D Face Analysis
Heatmap Example

Data Visualization and Accessibility

Knowledge dissemination and data accessibility is an essential part of nowadays research. The group deals daily with enormous amounts of complex data, therefore it is important to find ways to reach out to other researchers and clinicians to provide them the opportunity to use and explore the results of our work.

We focus on developing open source software and online tools for data visualization, research and educational purposes.

ImageNe Tool

ImageNe Heritability Tool

Interactive tool for exploring heritability of brain imaging traits
Brain Aging Tool

Brain Aging Tool

Tool for predicting brain age from MRI scans

Collaborations

Within Erasmus MC Medical Center

  • Departments of Epidemiology
  • Radiology and Nuclear Medicine
  • Neurosurgery
  • Plastic Surgery
  • Internal Medicine
  • Psychiatry
  • Neuroscience
  • Craniomaxillofacial Surgery
  • Genetic Identification

National and International

EU COST Actions

The group members contribute as AI experts in the EU COST actions: