The goal of Dr. Langlotz’s laboratory is to reduce diagnostic errors and to improve the accuracy and consistency of radiology communication through real-time decision support systems and other information technologies. The lab develops novel machine learning and natural language processing algorithms that provide intelligent assistance to radiologists, clinicians, patients, and other consumers of the radiology report. The laboratory supports translational work and, after assessment of clinical effectiveness, encourages commercialization and dissemination of the resulting software.

Dr. Langlotz is responsible for the strategic technical direction of a filmless, paperless radiology practice that interprets hundreds of thousands of exams annually. This unified technology platform contains over 1 billion clinical images occupying 0.5 petabytes, and serves as a test bed to develop new information technologies and evaluate their effect on clinical practice.

Active projects include:

1. Deep learning algorithms for differential diagnosis of bone tumors,
2. Machine learning for information extraction from multi-institutional radiology reports, and
3. Creation of a cloud-based, petabyte-scale, multi-institutional repository of annotated diagnostic images for the development of intelligent clinical image analysis systems.