Challenge Banner

News:

Learn2Reg 2026 - Task 1 - PSMAReg

PSMAReg: registration of pre-therapy and follow-up whole-body PSMA PET/CT scans for therapy planning, dosimetry, and response assessment.

For this task, Learn2Reg 2026 is partnering with the Society of Nuclear Medicine and Molecular Imaging (SNMMI) to promote a clinically grounded benchmark for longitudinal registration in nuclear medicine imaging.

Participants are asked to estimate deformable displacement fields that align each follow-up moving scan to its corresponding baseline fixed scan. The task evaluates longitudinal whole-body PSMA PET/CT registration before and after therapy, with an emphasis on robustness to anatomical change, multimodal image appearance differences, preservation of PET quantification, and variability across patient cohorts.

Data Summary:
  • Training and validation: 597 PSMA PET/CT scans from 378 patients, including 164 patients with two or more follow-up scans.
  • Testing: 262 internally acquired PSMA PET/CT scans from 131 patients, acquired using similar whole-body protocols.

Learn2Reg 2026 - Task 2 - Learn2Breath

Learn2Breath: focuses on registration of inspiratory and expiratory chest CT scans to enable robust quantification of lung biomechanics and disease progression.

For this task, Learn2Reg 2026 is partnering with the large, multicenter, Genetic Epidemiology of COPD (COPDGene) study for clinically relevant evaluation of image registration quality.

Participants are asked to estimate deformable displacement fields that align each expiratory chest CT scan to its corresponding fixed scan (inspiratory chest CT).

Data Summary:
  • Training and validation: 200 paired inspiratory and expiratory images for training, and 10 pairs for validation.
  • Testing: 100 internally acquired inspiratory and expiratory chest CT scans from the COPDGene study with a large volume change of at least 2 liters (L).

Learn2Reg Newsletter:

If you like to be added to the Learn2Reg newsletter, please fill out this form.

L2R online:
  • learn2reg.grand-challenge.org (this page) provides all information about the current Learn2Reg-Challenge, as well as our support forum and challenge archives
  • github.com/MDL-UzL/L2R provides code for evaluation, zero-deformation fields for sanity checks and other useful utilities for developing your image registration algorithm.
Motivation:

Standardised benchmark for the best conventional and learning-based medical registration methods:

  • Analyse accuracy, robustness and speed on complementary tasks for clinical impact. 
  • Remove entry barriers for new teams with expertise in deep learning but not necessarily registration.

Learn2Reg removes pitfalls for learning and applying transformations by providing:

  • python evaluation code for voxel displacement fields and open-source code all evaluation metrics
  • anatomical segmentation labels, manual landmarks, masks and keypoint correspondences for deep learning

Learn2Reg addresses four of the imminent challenges of medical image registration:

  • learning from relatively small datasets
  • estimating large deformations
  • dealing with multi-modal scans
  • learning from noisy annotations

Evaluation: Comprehensive and fair evaluation criteria that include:

  • Dice/surface distance and TRE to measure accuracy and robustness of transferring anatomical annotations 
  • standard deviation and extreme values of Jacobian determinant to promote plausible deformations,
  • low computation time for easier clinical translation evaluated using docker containers on GPUs provided by organisers.

Organisers / Contact:

  • Junyu Chen (Johns Hopkins University, MD, USA)
  • M. Faizyab A. Chaudhary (Johns Hopkins University, MD, USA)
  • Shuwen Wei (Johns Hopkins University, MD, USA)
  • Yihao Liu (Vanderbilt University, TN, USA)
  • Lianrui Zuo (Vanderbilt University, TN, USA)
  • Harrison Bai (Johns Hopkins University, MD, USA)
  • Yong Du (Johns Hopkins University, MD, USA)
  • Jing Tang (University of Cincinnati, OH, USA)
  • Sarah E. Gerard (University of Iowa, IA, USA)
  • Adam Alessio (Michigan State University, MI, USA)
  • Alessa Hering (Radboud University Medical Center, Nijmegen, Netherlands)
  • Joseph M. Reinhardt (University of Iowa, IA, USA)
  • Gary E. Christensen (University of Iowa, IA, USA)
  • Aaron Carass (Johns Hopkins University, MD, USA)