Motivation

Radiology reports contain crucial diagnostic information that form the backbone of patient care and clinical decision-making. However, the unstructured nature of these text-based reports poses challenges for efficient data retrieval and analysis. At the Klinikum rechts der Isar, Technical University of Munich (TUM) Clinic, we have amassed a substantial dataset of approximately 2,000,000 radiology reports spanning a decade. This wealth of information presents a unique opportunity to develop advanced natural language processing (NLP) techniques for automated information extraction.

The primary objective of this project is to develop and implement a robust system utilizing large language models (LLMs) to extract structured information from unstructured radiology reports. While the system will be designed for broad applicability in clinical data extraction, a particular focus will be placed on automating the extraction of procedural codes, which are currently manually assigned—a process that is both time-consuming and prone to errors. The work should be concluded with a scientific paper for publication in a top-tier journal.

Apply to this project here