Project
Medical imaging and reporting is getting more importance in healthcare services. Quality of imaging techniques and quality of reporting have big roles on the diagnostic success of radiological imaging. Interpretation of radiological images is done by doctors in an intuitive way for accurate analysis of images.
In current approach, radiology reports are stored as free-text in hospital archives. The process is usually done in the following order:
- Radiologists examine patients' films and medical images and record their findings loudly to a tape
- Secretaries listen to the tape and transfer them to text documents. These documents are stored in hospitals' archives to be later reviewed when necessary.
The problematic part in this approach is that it makes statistical analysis and retrieval of similar data unfeasible. To solve this problem, there is a need to get information from these freetexts by using text-mining approaches.
In this project we will try to get meaningful data from Turkish radiology reports and store these data in a proper way which will be fast and reliable to retrieve. Radiology reports are freetexts which have no formatting except the rules of Turkish.
Our team expects to provide a way to
- find similar patients, diseases,
- categorize diagnosis,
- identify integrity of treatments and variances between patients with the implementation of this project.