Defining cutting-edge imaging informatics
Dr. Erickson's lab creates new informatics tools that can clearly and concisely extract and convey the wealth of information in medical images to improve patient outcomes. In this example, CT images of the abdomen are segmented by the Radiology Informatics Lab's Body Composition model. Pixels in red are muscle, green is subcutaneous fat, brown is visceral fat, purple are visceral organs, and white is bone. The ratios of these tissues are useful clinical predictors of robustness and risk factors for cardiometabolic diseases.
Overview
Mayo Clinic's Radiology Informatics Lab directed by Bradley J. Erickson, M.D., Ph.D., adds value to medical imaging by designing tools that:
- Extract quantitative and diagnostic information, including subtle textures that humans often can't perceive or quantify.
- Improve the quality of medical images, including deep learning-based reconstruction.
- Improve the workflow for planning, executing and reporting medical imaging.
The lab's original work focused on improving diagnoses from radiological images. This work has expanded to include multimodal image and non-image-derived biomarkers of diseases. The lab works on a wide range of diseases and medical images from Mayo Clinic's Department of Radiology and other image-producing departments.
This is the main view of the open-source annotation software Sigma, which includes classic annotation tools and a number of deep learning models.
Sigma image annotation tool
The Radiology Informatics Lab has created an image annotation tool called Sigma. It allows users to divide medical images quickly and easily into parts. Imaging experts call this "segmentation." It is written in JavaScript, and it can be executed on most modern computers. It also can be accessed remotely via a web browser.
Sigma has advanced image processing tools for segmentation, including external deep learning tools that can produce segmentations. It also allows users to refine these segmentations.
Learn more about the lab
Dr. Erickson provides an introduction to the lab in this video:
Affiliations
Dr. Erickson's Radiology Informatics Lab is affiliated with several Mayo Clinic research areas, including: