Curtis P. Langlotz

Timeline

1. 1981–1989

   AI and medical degrees

   Raised in St. Paul, Minnesota, he earns a human biology degree at Stanford in 1981 and a master’s in computer science, on its artificial intelligence track, in 1983. He completes an MD and a PhD in medical information sciences at Stanford in 1989.

2. 1989–1998

   Radiology training at Penn

   In 1989, he moves to the University of Pennsylvania for radiology training and later joins its faculty. He founds Access Radiology, a picture archiving and communication systems company, in 1992.

3. 1998–2004

   eDictation and RadLex

   In 1998, he leaves Penn to assist the director of the National Cancer Institute’s Cancer Imaging Program and found eDictation, which combines speech recognition with structured data capture. He also leads development of the RadLex terminology for radiology reports.

4. 2004–2014

   Reporting systems at Penn

   After eDictation closes in 2004, he returns to Penn’s faculty. His work includes imaging platforms and report templates; he also co-founds Montage Healthcare Solutions, which provides natural language processing software for radiology.

5. 2014–2017

   Stanford and a career award

   He takes a faculty position at Stanford in 2014. In 2017, he receives the career achievement award of the Radiology Alliance for Health Services Research, an organization he founded and previously served as president.

6. 2018

   Directing the AIMI Center

   In 2018, he becomes director of Stanford’s Center for Artificial Intelligence in Medicine and Imaging (AIMI).

7. 2019

   The CheXpert chest X-rays

   In 2019, he coauthors CheXpert, a dataset of 224,316 chest radiographs from 65,240 patients. The work includes a labeler that extracts 14 observations from radiology reports and captures uncertainty, providing a benchmark for chest X-ray interpretation models.

8. 2024

   President of RSNA

   In 2024, he serves as president of the Radiological Society of North America.

9. 2026

   The Merlin CT model

   In 2026, he coauthors the Nature paper on Merlin, a three-dimensional vision-language model that learns from CT scans, electronic health record data and radiology reports. The team trains Merlin on 15,331 CT scans and evaluates tasks including disease prediction and report generation.

10. As of October 2026

    Professor and AIMI director

    As of October 2026, he is professor of radiology, medicine and biomedical data science, senior associate vice provost for research, and director of the AIMI Center at Stanford. His laboratory develops neural networks and other machine learning methods to analyze medical images and clinical notes.