Media Inquiries

Mary Ellen Hackett
Manager, Communications Office

Image of original whole-slide image of liver tumor with annotation by pathologists, courtesy PAIP 2019.
Image courtesy PAIP 2019, CC by-NC 4.0, original whole-slide image of liver tumor with annotation by pathologists.

FREDERICK, Md. – Frederick National Laboratory for Cancer Research’s medical imaging expertise received the highest rating at a recent grand challenge for pathology that featured nearly 1,000 competitors from around the world.

Hyun Jung, Ph.D., a bioinformatics analyst, won first place in both parts of the Pathology AI Platform Challenge (PAIP 2019). The event, held in China in October, was part of the International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI 2019), which drew leading scientists, engineers, and clinicians from disciplines associated with medical imaging and computer assisted intervention.

Jung participated in a challenge that evaluated new and existing algorithms for automated detection of liver cancer using whole-slide images. The two tasks given were liver cancer segmentation and viable tumor burden estimation.

Jung is part of the FNL’s Imaging and Visualization Group in the Advanced Biomedical Computational Sciences (ABCS), which supports and accelerates basic research by developing and implementing technologies in image analysis, scientific visualization, IT infrastructure, and data access. 

“The challenge is a good way for us to gauge our methods, because this field changes quite fast,” said Jung.

The Imaging and Visualization Group works with collaborators at NIH and within the FNL. Jung works with researchers who require segmentation of certain types of cancer. The process divides an image into parts for investigators to analyze, particularly for tumor staging, which can help researchers quantify and assess tumors when evaluating potential treatments.

By participating in a challenge such as PAIP 2019, Jung said the motivation is twofold. It provides an opportunity to assess FNL’s medical imaging methods compared to others in the field, and it allows FNL to keep up to date with the latest technologies. 

Challenge participants also receive access to datasets that are labeled and can be available for download for pathologists or other scientists to aid with their research. 

To view complete challenge results, visit the PAIP 2019 website