Dear Participants,

we have just released the data for L2R 2023:

New Task ThoraxCBCT:

Download here With the release of the ThoraxCBCT dataset the registration problem of image guided radiation therapy (IGRT) between pre-therapeutic fan beam CT (FBCT) and interventional low dose cone beam CT (CBCT) is addressed. For each patient one FBCT prior to therapy in maximum inspiration is provided which is denoted by the filename ending 0000*.nii.gz. Further, there are two CBCTs for each patient, one at the beginning of therapy (0001.nii.gz) and one at the end of therapy (0002.nii.gz). Both are acquired in maximum expiration. The task is to find one solution for the registration of two pairs of images per patient: 1. The planning FBCT (inspiration) prior to therapy and the low dose CBCT at the beginning of therapy (expiration), which are acquired at similar timepoints, and 2. The planning FBCT (inspiration) prior to therapy and the low dose CBCT at the end of therapy (expiration), where longer periods of time (usually several months) lay in between the scans.

The challenge in both pairs of images is the registration of images from the two different modalities FBCT and CBCT and also the shift in breathing phases between maximum inspiration and expiration. In the second subtask there is an additional challenge with the time shift between the planning CT at the beginning of therapy and the follow up CBCT at the end of therapy.

The released dataset includes training images of 11 patients resulting in 11 FBCT and 22 CBCT images, and validation images of 3 patients with 3 FBCT and 6 CBCT images. The correspondences for the validation cases will not be published yet, as they will be used for evaluation on GrandChallenge. In contrast to previous tasks, the focus of this task is not mainly on the lung, but also on other thoracic organs, especially organs at risk with regard to radiotherapy, which is why the keypoints were determined not only in the lung, but in the entire trunk region.

You can find more information about the task in the info.txt enclosed in the data.

Extended Task NLST:

Download here This year, we have some Updates regarding NLST: * We release 100 additional training cases that show greater movement compared to the previous data * We have selected a new set of test cases that have greater movement. In addition, this year will be evaluated only on manual landmarks * We will completely renew the validation for snapshot evaluation via Grand-Challange by evaluating on manual landmarks here as well

Snapshot Evaluation

As in previous years, we will again conduct a Snapshot Evaluation. The top 2 participants of the new tasks (ThoraxCBCT/NLST), plus an additional wildcard, will receive an invitation to talk at the workshop at MICCAI 2023 in Vancouver.

Please let us know if you have any other questions (or find irregularities in the dataset)!