Medical Data Donors e.V. was founded on the 24th of January 2018 with the intention of making larger databases of medical images available for research. To realize this vision, we desire patients to voluntarily donate their data to us so that we can use it for exciting research projects.
Building a large database of annotated medical images is extremely expensive and therefore normally reserved for large-scale industry. As a rule, this leads to a monopoly position for global players.
Our not-for-profit database approach can be the foundation for major discoveries in medicine as countless research groups and start-ups are given access to the collected data under strict ethical supervision. This results in enormous benefits for all parties involved.
For researchers, a major benefit of using donated data is that the problem of limited data availability in the medical domain can be counteracted.
However, there are also great advantages for the data donors: In every phase of our ongoing research projects, he receives detailed insights into the use of his data and current research results.
The privacy of the data donors is very important to us. Therefore, we always adhere to the General Data Protection Regulation (GDPR) of the European Union (EU). This ensures the protection of personal data within the EU and gives the patient full control over their data.
As an example, data donors have the right to be informed whether or not their personal data is being processed, where, and for what purpose. In addition, patients can revoke the permission to use their data at any time and have their data deleted.
In order to make the donated image data usable for machine learning tasks, it has to be annotated. This means that the actual image data is supplemented by metadata (e.g. disease labels or organ outlines). This additional information is required to train machine learning algorithms to make future decisions.
Since medical image annotation is an expensive and time-consuming task, we follow a crowdsourcing approach. In this way, the process of generating metadata is outsourced to a group of voluntary users.
For attracting a crowd of users to accomplish medical tasks, gamification has been proven to be a very effective strategy. Gamified crowdsourcing concepts aim at generating annotations of adequate quality while motivating and entertaining the user through fun.
We had a wonderful first seminar at FAU Erlangen-Nuremberg in Summer Term 2019. Within the seminar, we were able to create multiple “annotation games” in the context of Medical Data Donors e.V. You will find short descriptions and download links for Android on our games page.