Rainer Grobholz

Title: Rainer Grobholz: Advancing Cross-Site Calibration in Deep Learning

Introduction

Rainer Grobholz is an inventor based in Rupperswil, Switzerland. His work includes a patent for a deep-learning method designed to mitigate domain shift across imaging data from different sites.

Latest Patents

Grobholz is associated with 1 patent: “Multi-site cross-organ calibrated deep learning (MUSCID) method and apparatus.” The disclosed method accesses imaging data from two sites, identifies on-target regions in images from the first site and off-target regions in images from the second, and uses the off-target regions to calibrate the on-target regions. The aim is to reduce the domain shift between the calibrated regions and the second-site data compared with the shift affecting the original regions.

Career Highlights

Grobholz’s listed institutional affiliations include Case Western Reserve University and Emory University. His patent reflects work at the intersection of medical imaging and deep learning, with a focus on addressing differences between data collected at separate sites.

Collaborations

Grobholz’s listed collaborators include Yufei Zhou and Can Koyuncu.

Conclusion

Rainer Grobholz’s patent describes a cross-site calibration approach intended to make deep-learning methods more robust to domain shifts in medical imaging.