My name is Piotr Chudzik and I have recently started my new research job and Phd studentship within the REVAMMAD programme, part of the European Union’s 7th Framework (FP7) Marie Curie Initial Training Network programme. I work with Dr Luca Antiga and Orobix in Bergamo,Italy on the Data Warehouse and the Evaluation Framework. This project will address three big impediments in the retinal imaging field:
- Lack of shared data and ways to share data.
- Lack of standardized evaluation protocol.
- No ways to compare newly developed techniques against each other.
The project is divided into three main phases:
- Building the Data Warehouse and the Evaluation Framework.
- Combining multiple methods to improve performance.
- Matching new individuals against existing datasets.
The Data Warehouse will allow users to deposit data (retinal images, modelling data, statistical data) and meta-data (e.g. disease details) in a secure and confidential repository.
Moreover users will be able to store and execute algorithms on available datasets inside the Data Warehouse. This will allow to benchmark and evaluate computations against multiple datasets. The Evaluation Framework will include following functionality:
- Validation Wrappers – users will be able to define their own methods of validation.
- Statistical Validation Service – automated or semi-automated statistical analysis tools.
Sensitivity Analysis – the way to test model’s sensitivity to different parameters..
Once users will start to deposit their datasets and computations in the Data Warehouse, we will be able to experiment and combine multiple algorithms using ensemble methods to increase their performance. Last but not least we will project a single patient’s data on the existing datasets (using case-based reasoning techniques) and draw some conclusions.