MICCAI 2025 – The Human Radiome Project represented in Seoul!

The divisions of Medical Image Computing (MIC) and Intelligent Medical Systems (IMSY) from the German Cancer Research Center (DKFZ) sent several of their members to the renowned MICCAI conference, held this year in Seoul, South Korea. Among this group were key members of The Human Radiome Project (THRP) as well as Helmholtz Imaging, who contributed significantly with three accepted papers, one challenge workshop, and five challenge wins.

Challenge Highlights:

• PANTHER Challenge: Our intern Omer Faruk Durugol, supervised by Maximilian Rokuss and Yannick Kirchhoff, achieved 1st place in both tasks, tackling pancreatic tumor segmentation in diagnostic MRI and MR-Linac images.
• autoPET/CT IV: Yannick Kirchhoff and Maximilian Rokuss combined LesionLocator and LongiSeg to win 1st and 2nd place in Task 1 and 2 respectively!
• ODELIA Breast Cancer Classification: Benjamin Hamm, Yannick Kirchhoff and Maximilian Rokuss ranked 1st in this challenge advancing breast cancer screening in MRI and presented our newly released BreastDivider dataset – 17k+ MRI scans and 3k+ annotated targets.
• TriALS Challenge: Congratulations to Andrés Martínez Mora, Katharina Eckstein, Jonathan Deissler, and the team for an excellent performance in the TriALS Challenge, winning both automatic liver lesion segmentation tasks (Task 1 & 3) and placing 3rd in the interactive task (Task 2)!
• FOMO Challenge: Andrés Martínez Mora, Partha Ghosh and Constantin Ulrich participated in the open track and secured 3rd place – great job!

Quotes from the Challenge Winners:

Maximilian Rokuss: “The MICCAI challenges this year showed how transformative large-scale pretrained models have become for medical imaging. Nearly every winning approach relied on pretrained or distilled models, indicating that the era of training from scratch is coming to an end.”

Benjamin Hamm: “Connecting with other challenge participants at MICCAI was incredibly valuable. We had the chance to exchange ideas about our different approaches, future improvements, open research questions, and clinical translation. These conversations truly foster collaboration and open the door for exciting future projects.”

The group photo shows members of Medical Image Computing, led by Klaus Maier-Hein, together with members of the Intelligent Medical Systems Lab at DKFZ, led by Prof. Lena Maier-Hein.

Author: Dr. Daniel Walther, The Human Radiome Project Co-Coordinator